Communication data transmission method and device of train approach warning intercom

Abstract

This invention relates to the field of communication data transmission technology, and discloses a communication data transmission method and device for a train approach warning walkie-talkie. The method includes: collecting real-time signals and historical status data of the train's surrounding environment; performing signal attenuation and noise analysis to obtain preliminary position data; when the signal strength is too low, using a random forest algorithm to analyze and compensate for noise to obtain an adjusted dataset; obtaining the position trajectory through three-dimensional coordinate calculation and attenuation correction; if it does not match the historical status, optimizing the position information through a noise gain matrix; fusing multi-source communication correction signals to form a stable transmission dataset; monitoring latency and signal-to-noise ratio in real time, re-optimizing and calculating when limits are exceeded, and finally fusing to generate accurate communication data. This invention improves the stability and accuracy of train positioning and communication in complex environments.

Description

Technical Field

[0001] This invention relates to the field of communication data transmission technology, and in particular to a communication data transmission method and device for a train approach warning walkie-talkie. Background Technology

[0002] Train approach warning systems are crucial in the field of railway safety. By monitoring the position and movement of trains in real time, they ensure safe train operation and efficient dispatching, and are an indispensable technological pillar of modern railway transportation.

[0003] In existing technologies, the most traditional and widely used method is voice notification via walkie-talkie. Dispatch centers, station duty officers, or locomotive drivers use high-power base stations or locomotive radios to broadcast train approach warnings via voice announcements on specific operating frequencies, including train number, direction, and estimated arrival time. On-site personnel receive these announcements using portable walkie-talkies, thus becoming aware of the train's approach. However, external environmental factors such as terrain or electromagnetic interference can cause signal fluctuations. For example, when a high-speed train passes through a tunnel, the signal may be temporarily lost due to obstruction, leading to data transmission distortion or interruption.

[0004] In summary, existing technologies suffer from the problem that signals are easily interfered with in complex environments, such as tunnels, mountainous areas, or areas with strong electromagnetic interference, leading to data transmission distortion or signal transmission interruption. Summary of the Invention

[0005] This invention provides a communication data transmission method and device for a train approach warning intercom, so as to achieve communication stability of the train in environments such as tunnels, mountainous areas or areas with strong electromagnetic interference.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a communication data transmission method for a train approach warning intercom, comprising:

[0007] Real-time signals, including speed, direction, and distance parameters, are collected from the train's surrounding environment, and the train's historical motion status and multi-dimensional communication data streams are obtained.

[0008] The real-time signal is subjected to signal attenuation analysis and noise source identification to obtain preliminary location dataset and noise source identification results;

[0009] When the signal strength in the preliminary location dataset is lower than a preset strength threshold, the random forest algorithm is used to analyze the noise source identification results to obtain a compensation coefficient. The dataset is then adjusted based on the compensation coefficient and the preliminary location dataset to obtain the adjusted dataset.

[0010] Based on the adjusted dataset, perform iterative calculation of the three-dimensional coordinates to obtain a preliminary position estimate.

[0011] The preliminary position estimate is corrected for signal attenuation to obtain the corrected position trajectory data;

[0012] When the corrected position trajectory data does not match the historical motion state of the train, a noise gain matrix is ​​calculated based on the noise source, and the corrected position trajectory data is optimized based on the noise gain matrix to obtain optimized position information.

[0013] An initial signal is obtained from the multi-dimensional communication data stream and corrected to obtain a calibration signal. The optimized location information is then fused with the calibration signal to obtain a stable transmission dataset.

[0014] The system acquires the latency and signal-to-noise ratio data of the stable transmission dataset in real time. When the latency or signal-to-noise ratio data exceeds a preset threshold, the system optimizes the calculation of the train's three-dimensional coordinates to obtain optimized position data. The optimized position data is then fused with the train's historical motion state to obtain the final accurate communication data.

[0015] In one optional implementation, the step of performing signal attenuation analysis and noise source identification on the real-time signal to obtain a preliminary location dataset and noise source identification results includes:

[0016] The velocity direction and distance parameters in the real-time signal are converted from analog to digital to obtain the first signal dataset.

[0017] The time-domain signal in the first signal dataset is converted into a frequency-domain signal to obtain a frequency-converted signal. The signal attenuation features are extracted from the frequency-converted signal to obtain an attenuation feature dataset.

[0018] Based on the attenuation feature dataset, noise sources are separated to obtain a noise separation dataset, and the noise source type is determined to obtain the noise source identification result;

[0019] A preliminary location dataset is obtained by performing a Kalman filter fusion operation on the noise separation dataset.

[0020] In one optional implementation, when the signal strength in the preliminary location dataset is lower than a preset strength threshold, a random forest algorithm is used to analyze the noise source identification results to obtain a compensation coefficient. The dataset is then adjusted based on the compensation coefficient and the preliminary location dataset to obtain an adjusted dataset, including:

[0021] When the signal strength is lower than a preset signal strength threshold, it is determined that there is interference, and the type of interference is determined based on the noise source identification result.

[0022] The environmental noise characteristics are obtained by extracting the noise source features corresponding to the interference type from a pre-established noise source database.

[0023] The environmental noise characteristics were analyzed and calculated using the random forest algorithm to obtain compensation coefficients;

[0024] The preliminary location dataset is weighted and adjusted according to the compensation coefficient to obtain the adjusted dataset.

[0025] In one optional implementation, the step of performing iterative calculation of three-dimensional coordinates based on the adjusted dataset to obtain a preliminary position estimate includes:

[0026] The base station delay, signal strength, and angle of arrival data in the adjusted dataset are obtained, and the base station delay, signal strength, and angle of arrival data are smoothed to obtain optimized multi-source data.

[0027] The train's three-dimensional coordinates are iteratively calculated based on the optimized multi-source data to obtain preliminary three-dimensional coordinates.

[0028] The characteristic parameters of temperature, humidity and electromagnetic interference are extracted from the pre-established environmental database, and the preliminary three-dimensional coordinates are corrected by multipath fading model to obtain the corrected three-dimensional coordinates.

[0029] When the corrected three-dimensional coordinates meet the preset coordinate convergence threshold, the temperature compensation, humidity compensation, and multipath fading correction parameters are integrated to obtain a preliminary position estimate.

[0030] In one optional implementation, the step of performing signal attenuation correction on the preliminary position estimate to obtain corrected position trajectory data includes:

[0031] Obtain the time delay data and environmental parameters from the preliminary position estimate, and smooth the time delay data and environmental parameters to obtain the smoothed state vector;

[0032] The preliminary position trajectory is obtained by iterative calculation based on the smoothed state vector.

[0033] If the initial position trajectory does not reach the preset convergence threshold, the time delay data and the environmental parameters are integrated to obtain the corrected state vector;

[0034] The corrected position trajectory data is obtained by optimizing the state vector in conjunction with the signal attenuation model.

[0035] In one optional implementation, when the corrected position trajectory data does not match the historical motion state of the train, a noise gain matrix is ​​calculated based on the noise source, and the corrected position trajectory data is optimized based on the noise gain matrix to obtain optimized position information, including:

[0036] An initial state vector is obtained from sensor data and environmental parameters in the corrected position trajectory data, and a smoothed state vector is obtained by smoothing the initial state vector.

[0037] When the smoothed state vector does not match the train's historical motion state, the noise gain matrix is ​​calculated based on the noise source.

[0038] The smoothed state vector is corrected based on the noise gain matrix and combined with environmental parameters to obtain the corrected state vector;

[0039] The corrected state vector is subjected to error minimization processing to obtain optimized position information.

[0040] In one optional implementation, the step of obtaining an initial signal from the multi-dimensional communication data stream and correcting it to obtain a calibration signal, and then fusing the optimized location information with the calibration signal to obtain a stable transmission dataset, includes:

[0041] An initial signal is obtained from a multi-dimensional communication data stream, and the initial signal is adaptively processed to obtain a preprocessed signal.

[0042] When the preprocessed signal is inconsistent with the signal output by the preset prediction equation, calibration is performed based on the preprocessed signal to obtain a calibration signal.

[0043] By combining the current channel parameters, the calibration signal and the optimized location information are integrated to obtain fused data;

[0044] Channel optimization processing is performed on the fused data to obtain a stable transmission dataset.

[0045] In one optional implementation, the real-time acquisition of latency data and signal-to-noise ratio data of the stable transmission dataset, when the latency data or signal-to-noise ratio data exceeds a preset threshold, optimizes the train's three-dimensional coordinate calculation to obtain optimized position data, including:

[0046] The system acquires the latency data and signal-to-noise ratio data of the stable transmission dataset in real time. When the latency data or the signal-to-noise ratio data exceeds a preset data threshold, the system performs deviation detection on the stable transmission dataset to obtain abnormal data points.

[0047] Triangulation is performed based on the signal strength and time difference of arrival of the abnormal data points to obtain optimized location data;

[0048] In one optional implementation, the step of fusing the optimized location data and the train's historical motion state to obtain the final accurate communication data includes:

[0049] Based on the optimized position data, the characteristics of multipath effect and external noise are identified, and combined with the historical motion state of the train, the classified interference data is obtained.

[0050] The time delay data and signal-to-noise ratio data of the optimized location data are optimized, and the classified interference data and the optimized location data are fused to obtain the final accurate communication data.

[0051] Secondly, the present invention provides a communication data transmission device for a train approach warning intercom, comprising:

[0052] The data acquisition module is used to collect real-time signals, including speed, direction, and distance parameters, from the train's surrounding environment and to acquire the train's historical motion status.

[0053] The signal analysis module is used to perform signal attenuation analysis and noise source identification on the real-time signal to obtain preliminary location dataset and noise source identification results;

[0054] The signal compensation module is used to analyze the noise source identification results using a random forest algorithm to obtain compensation coefficients when the signal strength in the preliminary location dataset is lower than a preset strength threshold. The compensation coefficients and the preliminary location dataset are then adjusted to obtain an adjusted dataset.

[0055] The coordinate calculation module is used to perform iterative calculation of three-dimensional coordinates based on the adjusted dataset to obtain preliminary position estimates.

[0056] The coordinate correction module is used to correct the signal attenuation of the preliminary position estimate to obtain the corrected position trajectory data.

[0057] The coordinate optimization module is used to calculate a noise gain matrix based on the noise source when the corrected position trajectory data does not match the historical motion state of the train, and to optimize the corrected position trajectory data based on the noise gain matrix to obtain optimized position information.

[0058] The data fusion module is used to obtain initial signal data from multi-dimensional communication data streams and correct it to obtain calibration signal data, and then fuse the optimized location information with the calibration signal data to obtain a stable transmission dataset.

[0059] The data output module is used to acquire the latency data and signal-to-noise ratio data of the stable transmission dataset in real time. When the latency data or signal-to-noise ratio data exceeds a preset threshold, the three-dimensional coordinate calculation of the train is optimized to obtain optimized position data. The optimized position data and the historical motion state of the train are fused to obtain the final accurate communication data.

[0060] Thirdly, the present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the communication data transmission method of the train approach warning intercom as described in any one of the above.

[0061] Fourthly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the communication data transmission method of the train approach warning intercom as described in any one of the above.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] (1) This invention obtains preliminary location data through signal attenuation analysis and noise identification. When the signal strength is insufficient, a random forest algorithm is used to generate a noise compensation coefficient optimized dataset. The location trajectory is obtained through three-dimensional coordinate calculation and attenuation correction. The trajectory is further optimized through a noise gain matrix, and a stable transmission dataset is generated by combining it with calibrated multi-dimensional communication signals. The system monitors the transmission quality (delay, signal-to-noise ratio) in real time. When the limit is exceeded, the coordinate calculation is re-optimized, and finally, high-precision communication results are output by fusing historical motion data. This method effectively solves the problem that signals are easily interfered with in complex environments, such as tunnels, mountainous areas, or areas with strong electromagnetic interference, leading to data transmission distortion or signal transmission interruption.

[0064] (2) Based on multi-source signal fusion and intelligent compensation, this invention uses random forest to process noise interference and combines three-dimensional coordinate iterative calculation and dynamic noise gain adjustment to effectively suppress errors caused by environmental factors. Multiple verifications of historical motion states and real-time data ensure trajectory consistency, and adaptive channel parameter calibration further guarantees data transmission quality. Thus, at the technical level, this invention improves the feasibility and reliability of high-speed rail positioning and communication under complex conditions.

[0065] (3) This invention significantly improves the communication stability of trains in environments with strong interference and multiple attenuation, and effectively overcomes the error accumulation problem caused by weak signals and high noise in traditional methods. Through intelligent noise compensation, multi-source data fusion and real-time optimization mechanisms, the system can achieve continuous high-precision positioning and reliable data transmission, providing key technical support for train operation safety, scheduling efficiency and early warning response, and has important engineering application value. Attached Figure Description

[0066] Figure 1 This is a schematic diagram of the communication data transmission method of the train approach warning walkie-talkie provided in the first embodiment of the present invention;

[0067] Figure 2 This is a schematic diagram of the communication data transmission device of the train approach warning walkie-talkie provided in the second embodiment of the present invention. Detailed Implementation

[0068] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0069] Reference Figure 1 The first embodiment of the present invention provides a communication data transmission method for a train approach warning walkie-talkie, comprising the following steps:

[0070] S11 collects real-time signals from the train's surrounding environment, including speed, direction, and distance parameters, and acquires the train's historical motion status and multi-dimensional communication data streams.

[0071] S12, perform signal attenuation analysis and noise source identification on the real-time signal to obtain preliminary location dataset and noise source identification results;

[0072] S13, when the signal strength in the preliminary location dataset is lower than a preset strength threshold, the random forest algorithm is used to analyze the noise source identification results to obtain a compensation coefficient, and the dataset is adjusted according to the compensation coefficient and the preliminary location dataset to obtain the adjusted dataset;

[0073] S14, perform iterative calculation of three-dimensional coordinates based on the adjusted dataset to obtain a preliminary position estimate;

[0074] S15, perform signal attenuation correction on the preliminary position estimate to obtain the corrected position trajectory data;

[0075] S16, when the corrected position trajectory data does not match the historical motion state of the train, the noise gain matrix is ​​calculated based on the noise source, and the corrected position trajectory data is optimized based on the noise gain matrix to obtain optimized position information;

[0076] S17, an initial signal is obtained from the multi-dimensional communication data stream and corrected to obtain a calibration signal, and the optimized position information is fused with the calibration signal to obtain a stable transmission dataset;

[0077] S18, real-time acquisition of delay data and signal-to-noise ratio data of the stable transmission dataset; when the delay data or signal-to-noise ratio data exceeds a preset range threshold, optimization of train three-dimensional coordinate calculation to obtain optimized position data; fusion of the optimized position data and the train's historical motion state to obtain final accurate communication data.

[0078] In step S11, real-time signals including speed, direction and distance parameters need to be collected from the train's surrounding environment, and the train's historical motion status and multi-dimensional communication data stream need to be obtained.

[0079] It should be noted that in the scenario of train operation, the sensor network collects real-time data on the train's surrounding environment, covering real-time signals of parameters such as speed, direction, and distance. The sensor network consists of various sensors installed on the train body and beside the track, such as lidar, accelerometers, and gyroscopes, which are used to measure changes in distance, speed, and direction, respectively.

[0080] The train's historical motion state includes its historical trajectory and the vector velocity of that trajectory at that point, for example, 21 m / s, in the direction of due east. The multi-dimensional communication data stream consists of multi-dimensional information about the wireless signal captured by the train's communication module, including signal delay, bandwidth occupancy, and signal-to-noise ratio.

[0081] In step S12, the process of performing signal attenuation analysis and noise source identification on the real-time signal to obtain preliminary location datasets and noise source identification results includes:

[0082] The velocity direction and distance parameters in the real-time signal are converted from analog to digital to obtain the first signal dataset.

[0083] The time-domain signal in the first signal dataset is converted into a frequency-domain signal to obtain a frequency-converted signal. The signal attenuation features are extracted from the frequency-converted signal to obtain an attenuation feature dataset.

[0084] Based on the attenuation feature dataset, noise sources are separated to obtain a noise separation dataset, and the noise source type is determined to obtain the noise source identification result;

[0085] A preliminary location dataset is obtained by performing a Kalman filter fusion operation on the noise separation dataset.

[0086] First, the acquired real-time signals are digitized using an analog-to-digital converter to obtain the first signal dataset. Assuming the train is traveling at 80 km / h, the sensors collect data 1000 times per second, generating a digital signal sequence containing speed, heading angle, and obstacle distance. This digitization ensures high accuracy for subsequent signal analysis, providing a reliable data foundation for train positioning and environmental perception.

[0087] Subsequently, a Fast Fourier Transform (FFT) was used to convert the time-domain signal into a frequency-domain signal to extract signal attenuation characteristics. The time-domain signal may contain composite signals such as train vibration and wind noise. The FFT can identify attenuation characteristics within a specific frequency range, yielding an attenuation characteristic dataset. For example, when a train passes through a tunnel, the signal may experience attenuation in a specific frequency band due to reflection, manifesting as a decrease in signal strength within the 0.5-2kHz frequency range. This attenuation characteristic dataset reflects the impact of the environment on the signal, such as external factors like tunnel walls and rain or snow. This frequency-domain analysis helps identify patterns of environmental interference, providing a basis for subsequent noise reduction.

[0088] Next, the attenuation feature dataset is subjected to dimensionality reduction to separate noise signals and determine the noise source category, thus obtaining the noise source identification results. Principal component analysis (PCA) is used to extract the main variation components of the data, reducing the high-dimensional signal to a low-dimensional space. The original data may contain 100 feature dimensions; PCA retains the first five principal components, explaining 90% of the signal variation. Noise sources may be identified as wind noise, mechanical vibration, or electromagnetic interference. For example, wind noise is characterized by high-frequency random fluctuations, while mechanical vibration is mostly a low-frequency periodic signal. This dimensionality reduction and classification improves data processing efficiency and reduces the interference of noise on positioning accuracy.

[0089] Finally, based on the noise-separated dataset and combining speed, direction, and distance parameters, a Kalman filter algorithm is used to fuse multi-source data, obtaining a preliminary position dataset. Kalman filtering is then used for prediction and update steps, fusing sensor data to optimize train position estimation. Assuming the train's speed at time t is 80 km / h, its heading angle is 30°, and its distance from an obstacle ahead is 100 m, Kalman filtering can integrate historical data and current measurements, correcting for noise-induced biases and outputting more accurate train position coordinates. This fusion method significantly improves positioning accuracy, especially in complex environments such as tunnels or strong winds.

[0090] In step S13, when the signal strength in the preliminary location dataset is lower than a preset strength threshold, a random forest algorithm is used to analyze the noise source identification results to obtain a compensation coefficient. The dataset is then adjusted based on the compensation coefficient and the preliminary location dataset to obtain an adjusted dataset, including:

[0091] When the signal strength is lower than a preset signal strength threshold, it is determined that there is interference, and the type of interference is determined based on the noise source identification result.

[0092] The environmental noise characteristics are obtained by extracting the noise source features corresponding to the interference type from a pre-established noise source database.

[0093] The environmental noise characteristics were analyzed and calculated using the random forest algorithm to obtain compensation coefficients;

[0094] The preliminary location dataset is weighted and adjusted according to the compensation coefficient to obtain the adjusted dataset.

[0095] First, signal strength is extracted from the preliminary location dataset acquired by the sensor network. Signal strength reflects the communication quality between the sensor and the target object, and is typically measured in decibels (dB / mW). Assuming the train is running in complex terrain, the signal strength captured by the sensor network might be -75 dBm, while a similar preset strength threshold derived from historical data analysis is -70 dBm. When the signal strength is found to be below the threshold, it indicates the presence of interference. This interference could originate from electromagnetic interference or physical obstacles. The process of identifying interference factors involves real-time monitoring of signal strength fluctuations combined with environmental parameters, resulting in the interference type.

[0096] It is worth noting that the preset strength threshold is an empirical value derived from statistical analysis of a large amount of historical operating data, and can be dynamically adjusted according to the actual environment and operating conditions. By collecting communication success rate and positioning accuracy data between the train and the base station under different signal-to-noise ratio conditions in typical operating environments (such as plains, tunnels, and mountainous areas), the lowest average signal strength corresponding to a communication success rate higher than 95% and a positioning error lower than 1 meter is set as the threshold.

[0097] Subsequently, based on the interference type, a database is queried, and corresponding noise source features are extracted from a pre-established noise source database to obtain environmental noise characteristics. The pre-established noise source database is a relational database, whose structure includes, but is not limited to, interference type (e.g., electromagnetic interference, wind noise, mechanical vibration), center frequency (Hz), bandwidth (Hz), typical amplitude (mV), typical duration (s), and environmental characteristic identifiers (e.g., 'in a tunnel', 'strong wind area'). Test equipment is deployed in various typical environments to collect raw noise signals. Feature parameters are extracted using methods such as Fourier transform and wavelet analysis, and then labeled and stored in the database. Assuming the judgment indicates electromagnetic interference, the extracted feature parameters may include the frequency, amplitude, and duration of the interference signal, such as a frequency of 55Hz, an amplitude of 0.2mV, and a duration of 0.1 seconds. These feature parameters serve as input features for the random forest algorithm.

[0098] Next, the random forest algorithm is used to analyze and calculate the characteristic parameters of environmental noise to obtain compensation coefficients. The random forest algorithm classifies and regresses noise features by constructing decision trees, outputting compensation coefficients to compensate for the impact of noise on the signal. For example, noise samples from historical train operation data are used as training data, and hyperparameters of the model, such as the number and depth of trees, are determined through methods such as cross-validation (exemplarily, the number of decision trees is set to 100, the maximum depth to 10, and the mean squared error (MSE) is used as the splitting criterion). Historical motion state data of similar trains and the training set of the pre-built model are collected, preprocessed, and labeled with features. Samples are then taken from these to form the training subset of each tree model. At the same time, some noise feature parameters, such as frequency, amplitude, and signal-to-noise ratio, are randomly selected as candidate splitting attributes when splitting at each node. The optimal splitting point is determined by calculating the Gini coefficient or information gain. The complete decision tree is recursively generated until a preset termination condition is reached, such as the number of node samples being less than the tree depth limit. The noise characteristic parameters (frequency 55Hz, amplitude 0.2mV) acquired in real time are input into all constructed decision trees. Each tree outputs a predicted compensation coefficient. For example, if a decision tree, based on different training processes, yields a similar compensation coefficient of 1.1, the final comprehensive compensation coefficient of 1.2 is calculated by aggregating the outputs of all trees and using regression averaging. This coefficient reflects the overall attenuation effect of noise on the signal; a higher value indicates a higher required compensation intensity. The algorithm's advantage lies in its ability to handle high-dimensional data and model the nonlinear relationships of noise characteristics.

[0099] Finally, the input data is weighted and adjusted based on the compensation coefficient calculation results to correct the initial position dataset, resulting in an adjusted dataset. Assuming the original position data is the train coordinates (x=100m, y=200m), and the signal strength is attenuated by noise, causing positional deviations, a compensation coefficient of 1.2 is used to weight and adjust the coordinates, resulting in adjusted coordinates (x=120m, y=240m). This data adjustment improves the accuracy of the position data by correcting for noise, ensuring the reliability of subsequent navigation or positioning systems.

[0100] In step S14, the step of performing iterative calculation of three-dimensional coordinates based on the adjusted dataset to obtain a preliminary position estimate includes:

[0101] The base station delay, signal strength, and angle of arrival data in the adjusted dataset are obtained, and the base station delay, signal strength, and angle of arrival data are smoothed to obtain optimized multi-source data.

[0102] The train's three-dimensional coordinates are iteratively calculated based on the optimized multi-source data to obtain preliminary three-dimensional coordinates.

[0103] The characteristic parameters of temperature, humidity and electromagnetic interference are extracted from the pre-established environmental database, and the preliminary three-dimensional coordinates are corrected by multipath fading model to obtain the corrected three-dimensional coordinates.

[0104] When the corrected three-dimensional coordinates meet the preset coordinate convergence threshold, the temperature compensation, humidity compensation, and multipath fading correction parameters are integrated to obtain a preliminary position estimate.

[0105] First, base station delay, signal strength, and angle of arrival (AHA) data from the adjusted dataset are acquired and used for high-precision positioning. Base station delay, signal strength, and AHA are core multi-source data, reflecting signal propagation time, communication quality, and signal direction, respectively. For example, base station delay is obtained by measuring the signal propagation time from the train to the base station, measured in nanoseconds (ns), with a typical value of 200 ns; signal strength is measured in decibels and milliwatts (dBm), assumed to be -65 dBm; and the AHA is obtained through analysis of the base station antenna array, with a typical value of 30°. These data are collected in real time via a sensor network, providing the foundation for subsequent positioning.

[0106] Subsequently, Kalman filtering is used to smooth the base station delay, signal strength, and angle of arrival data, reducing noise impact and obtaining optimized multi-source data. The principle is to fuse historical data and current measurements through prediction and update steps to generate a more stable data output. For example, assuming a significant fluctuation in signal strength during a measurement, Kalman filtering, combining previous measurements, outputs a smoothed signal strength of -66 dBm and a delay of 198 ns. This smoothing process improves data stability, providing reliable input for subsequent positioning calculations.

[0107] Next, high-precision triangulation is performed to calculate the train's three-dimensional coordinates based on the optimized multi-source data, yielding preliminary three-dimensional coordinates. Initial distance values ​​are calculated using light speed conversion based on base station delay data (e.g., 200ns, 210ns, 205ns). This is combined with signal strength data and angle-of-arrival (AOA) measurements. Signal strength data corrects for non-line-of-sight errors using a path loss model, while AOA measurements provide directional constraints. This establishes a geometric positioning equation set, and the least squares method is used for initial coordinate estimation, resulting in x=150m, y=250m, z=10m. The theoretical delay / angle of the current coordinates relative to each base station is calculated and compared with actual observations to generate residuals. The Gauss-Newton algorithm is used to adjust the coordinate values ​​and reduce the residuals. Each iteration introduces signal strength weights and AOA confidence levels, reducing the weight of weak signal base stations and assigning higher weights to narrow beams. After multiple iterations until the coordinate change is lower than the residual convergence, the optimized three-dimensional coordinates are finally output. This method improves the initial positioning accuracy through multi-source data fusion.

[0108] The path loss model is built based on multi-source heterogeneous data fusion and machine learning optimization. Data sources include historical environmental parameter databases (temperature, humidity, air pressure, electromagnetic interference spectrum), multi-base station channel measurement reports (signal strength, delay spread, multipath amplitude and phase characteristics), and GIS topographic data. Supervised learning is used for training, with environmental parameters and distance information as inputs and signal attenuation as a label. A nonlinear mapping relationship is fitted using Gaussian process regression or a neural network. When the input parameters include real-time environmental data (e.g., temperature 25℃, humidity 70%, interference intensity 0.2mV / m), distance estimates, and signal frequency, the output is the path loss value or calibration coefficient (e.g., 1.02). The calculation model integrates physical models (e.g., free-space propagation formulas) with data-driven correction. Adaptive attenuation is generated through dynamically weighted environmental factors (e.g., humidity weight 0.3, interference weight 0.5), ultimately achieving high-precision calibration (e.g., delay corrected from 2.05ms to 2.09ms).

[0109] We conduct offline pre-training using large-scale historical data (covering measured signal propagation values ​​under different seasons, weather, and terrain conditions) to establish a basic attenuation model. Then, we introduce real-time environmental parameters for online fine-tuning and use a sliding window mechanism to update model parameters (such as weighted training with data from the most recent 30 minutes) to ensure that the model adapts to dynamic environmental changes.

[0110] Then, to avoid environmental factors such as temperature, humidity, and electromagnetic interference from causing deviations in the initial coordinates, relevant feature parameters need to be extracted from the environmental database. The initial 3D coordinates are then corrected using a multipath fading model to obtain the corrected 3D coordinates. After obtaining the initial coordinates, the system collects environmental data in real time (e.g., temperature 25℃, humidity 80%, detected 55Hz electromagnetic interference), matches the closest environmental record from the database, extracts the associated parameters, calibrates the signal propagation speed based on the 0.1% velocity offset corresponding to temperature, corrects the path loss in base station distance calculation using a humidity attenuation coefficient of 1.05, and activates the interference suppression algorithm in the multipath fading model within the electromagnetic interference frequency range of 50-60Hz. Finally, the above corrections are substituted into the coordinate solution equations, and the initial coordinates are adjusted using weighted least squares. For example, the original (x=150m, y=250m, z=10m) is corrected to (x=152m, y=252m, z=10.5m), thereby eliminating systematic deviations caused by environmental factors.

[0111] The environmental database collects environmental factors for at least one full year and fits the impact of each environmental factor on signal propagation. It stores a multi-dimensional environmental parameter mapping table in structured tabular form. Each record includes temperature, humidity, electromagnetic interference characteristics (such as center frequency, bandwidth, and intensity threshold), and their corresponding signal propagation correction parameters (such as velocity offset percentage, attenuation coefficient, multipath error weight, and compensation coefficient). For example, at 71% humidity, the speed is 0.99 times the standard speed of light. It should be noted that this mapping relationship is based on industry standard models regarding atmospheric gas attenuation and refractive index. Under standard atmospheric conditions of 20°C, 71% humidity, and 1013 hPa atmospheric pressure, the propagation speed of 2.4 GHz radio waves in air is approximately 0.990 to 0.995 times the speed of light. This invention selects a conservative value of 0.99 times for calculation and fine-tunes it using field test data to ensure accuracy.

[0112] Furthermore, the multipath fading model is a mathematical model describing the random changes in signal amplitude, phase, and delay caused by wireless signals reaching the receiver via multiple paths (such as reflection, diffraction, and scattering) during propagation. Its core characteristic is that the signal superposition caused by multipath propagation generates constructive or destructive interference, leading to rapid fluctuations in the received signal strength (small-scale fading). This model quantifies the impact of the environment on the signal through parameterization, including delay spread (describing the degree of multipath delay dispersion), coherence bandwidth (the critical value for judging frequency-selective fading), Doppler spread (reflecting the time-varying characteristics of the channel), and Rice factor or Rayleigh distribution parameters (distinguishing the presence or absence of the dominant path). A predictive model is established by learning the nonlinear mapping relationship between environmental characteristics and multipath parameters (such as delay spread and Rice factor) through maximum likelihood estimation or deep learning networks (such as CNN-LSTM hybrid networks). In train positioning correction, the model uses real-time parameters (such as humidity attenuation coefficient and interference spectrum) provided by the environmental database to calculate the power distribution and phase shift of multipath components, thereby estimating the degree of signal distortion. The model then uses a compensation algorithm to reverse-correct the measured values ​​of time of arrival (TOA) and angle of arrival (AOA), ultimately reducing the impact of multipath effects on positioning coordinate errors.

[0113] Finally, when the corrected coordinates meet the convergence threshold, such as a deviation from the true location of less than 0.5m, the environmental parameters are integrated using a Bayesian weighted fusion algorithm. A preliminary location estimate is calculated based on the weights of temperature, humidity, and multipath fading. For example, firstly, confidence weights for each correction source are calculated based on the uncertainty of the environmental parameters, such as a temperature weight of 0.4, a humidity weight of 0.3, and a multipath fading weight of 0.3. The initial weight values ​​are allocated based on the inverse variance of each environmental factor in historical data and are dynamically updated through the real-time error covariance matrix. Subsequently, the temperature-compensated coordinates, humidity-corrected coordinates, and multipath fading-corrected coordinates are multiplied by their respective weights and summed to obtain the weighted average coordinates. That is, (x=152, y=252, z=10.5)×0.4+(x=151.5, y=251.8, z=10.2)×0.3+(x=151.7, y=251.2, z=10.1)×0.3=(x=151.8, y=251.5, z=10.3). Simultaneously, the correlation of errors from each correction source is evaluated using the covariance matrix, and the weight allocation is dynamically adjusted. For example, given the more pronounced multipath effect in enclosed environments such as tunnels, the multipath weight is increased to 0.5, while other weights are correspondingly reduced. This process iterates until the coordinate change is less than the threshold of 0.5m, ensuring that the track information remains essentially consistent. This process continuously integrates scene feature parameters from the environmental database, such as the tunnel reflection delay of 0.02 seconds, and updates the weights using Bayesian probability, ultimately achieving high-precision adaptive positioning. This extended scheme improves positioning reliability in complex environments and adapts to different scenario requirements.

[0114] In step S15, the process of correcting the preliminary position estimate by signal attenuation to obtain corrected position trajectory data includes:

[0115] Obtain the time delay data and environmental parameters from the preliminary position estimate, and smooth the time delay data and environmental parameters to obtain the smoothed state vector;

[0116] The preliminary position trajectory is obtained by iterative calculation based on the smoothed state vector.

[0117] If the initial position trajectory does not reach the preset convergence threshold, the time delay data and the environmental parameters are integrated to obtain the corrected state vector;

[0118] The corrected position trajectory data is obtained by optimizing the state vector in conjunction with the signal attenuation model.

[0119] First, multi-source sensors are used to collect time delay data and environmental parameters for the preliminary location estimate. Time delay data typically refers to the propagation time of a signal from the train to the base station, measured in nanoseconds (ns), with a typical value of 180 ns. Environmental parameters include temperature and humidity, such as a temperature of 20°C and humidity of 70%. This data is acquired in real time through a sensor network, providing input for subsequent processing. Kalman filtering is used to smooth the time delay data and environmental parameters to reduce noise interference. The principle is to obtain a smoothed state vector by predicting and updating, combining historical data and current measurements. Specifically, assuming the time delay data fluctuates between 175 and 185 ns in a certain measurement, the Kalman filter fuses previous measurements, outputting a smoothed time delay of 182 ns, and environmental parameters such as humidity are adjusted to 71%. This smoothing process improves the reliability of the data.

[0120] Subsequently, based on the smoothed state vector and the signal attenuation model, an initial position trajectory was obtained through iterative calculation. Using the smoothed delay data of 182 ns and the environmental parameter of 71% humidity, and employing an environmentally calibrated propagation speed (0.99 times the speed of light at 71% humidity), the initial distance was calculated to be approximately 120 m. The model then calculated the free-space path loss and added environmental attenuation (approximately 3 dB loss at 71% humidity). By comparing the transmit power, receive power, and total loss, a more accurate actual distance was derived. Finally, combining the corrected distances from multiple base stations and signal angle of arrival data, an iterative least squares algorithm was used to calculate the train's three-dimensional coordinates (x=140m, y=230m, z=9m) until the residual converged to less than 0.5m.

[0121] The signal attenuation model is built based on the fusion analysis of multi-source heterogeneous databases. The data sources mainly include environmental databases (continuously collected temperature, humidity, atmospheric pressure, and electromagnetic interference intensity and frequency band data recorded by a spectrum analyzer), multi-base station channel measurement reports (containing measured values ​​of signal strength, delay spread, and multipath component amplitude and phase characteristics under different geographical environments), and topographic and building distribution data provided by a high-precision geographic information system (GIS). Model training employs machine learning methods such as Gaussian process regression or neural networks. The algorithm minimizes the prediction error loss function to establish the mapping relationship between environmental parameters and signal propagation characteristics. During training, a maximum number of iterations (e.g., 1000 iterations) or a validation set error early stopping mechanism (e.g., no decrease for 10 consecutive iterations) is set as the termination condition. The final signal attenuation model can output parameters such as a humidity-signal attenuation coefficient lookup table (e.g., 71% humidity corresponds to an additional 3dB attenuation) and temperature-propagation speed correction (e.g., a speed offset of 0.1% at 25℃). It also integrates an electromagnetic interference masking algorithm to suppress noise in specific frequency bands, providing environmental compensation for precise train positioning.

[0122] Next, if the initial position trajectory fails to reach the convergence threshold, for example, if the deviation is greater than 0.6m, further correction is required. This convergence threshold is determined based on historical trajectory analysis; a deviation greater than 0.6m results in coordinate misalignment. The propagation time characteristics from the time delay data and the humidity attenuation factor from the environmental parameters are extracted. 71% humidity corresponds to an attenuation coefficient of 1.05. A weighted fusion is used to obtain the corrected state vector. The initial distance value of 120m calculated from the time delay data is multiplied by the humidity attenuation coefficient to obtain the corrected distance of 126m. Simultaneously, a set of geometric constraint equations is constructed using the signal angle of arrival data. This set of equations is solved iteratively using the gradient descent method. Each iteration introduces the calibration amount of environmental parameters on the signal propagation speed. At 71% humidity, the speed is corrected to 0.99 times the speed of light. The coordinate offset is gradually adjusted until the residual converges to less than 0.5m, finally outputting the corrected position trajectory (x=142m, y=231m, z=9.2m). This process dynamically compensates for signal propagation errors caused by environmental factors, ensuring that the positioning accuracy meets system requirements. The humidity decay factor, an environmental parameter, is derived from the data recorded in the structured table of the environmental database in S14, and therefore will not be elaborated further.

[0123] Finally, for the corrected state vector, the measurement update step further optimizes the position trajectory by combining the signal attenuation model, obtaining the corrected position trajectory data. During the measurement update phase, the system calls this model in real time to calculate the dynamic attenuation under the current environment (e.g., a sudden change in humidity to 75%). It then uses Kalman gain weighted fusion to fuse the predicted trajectory with the measured signal strength / delay data, ultimately outputting the corrected position trajectory (x=141.5m, y=230.8m, z=9.1m), achieving high-precision positioning that adapts to the environment. This method ensures the continuity and accuracy of the trajectory. The signal attenuation model here is consistent with the signal attenuation model in the smoothed state vector calculated in the above iterations, so it will not be elaborated further.

[0124] In step S16, when the corrected position trajectory data does not match the historical motion state of the train, a noise gain matrix is ​​calculated based on the noise source, and the corrected position trajectory data is optimized based on the noise gain matrix to obtain optimized position information, including:

[0125] An initial state vector is obtained from sensor data and environmental parameters in the corrected position trajectory data, and a smoothed state vector is obtained by smoothing the initial state vector.

[0126] When the smoothed state vector does not match the train's historical motion state, the noise gain matrix is ​​calculated based on the noise source.

[0127] The smoothed state vector is corrected based on the noise gain matrix and combined with environmental parameters to obtain the corrected state vector;

[0128] The corrected state vector is subjected to error minimization processing to obtain optimized position information.

[0129] First, an initial state vector is obtained from sensor data and environmental parameters in the corrected position trajectory data. Sensor data typically includes Doppler frequency shift and signal strength, for example, a Doppler frequency shift of 50Hz and a signal strength of -80dBm. Environmental parameters may include air pressure and wind speed, for example, an air pressure of 101kPa and a wind speed of 5m / s. This data is collected in real time by multi-source sensors on the train, providing input for subsequent processing. The initial state vector is generated based on this data, combined with the train's historical trajectory, to preliminarily estimate the position and speed. For example, based on the frequency shift and signal strength, the initial state vector may estimate the train's position as x=150m, y=200m, and its speed as 20m / s.

[0130] Subsequently, data smoothing is performed based on the initial state vector. The sensor data is preprocessed using weighted averaging or low-pass filtering to obtain a smoothed state vector. For example, if the signal strength fluctuation in the sensor data is between -82dBm and -78dBm, the smoothing process may output a stable -80.5dBm, and the smoothed velocity data will be 19.8m / s. This smoothed state vector improves data stability and lays the foundation for subsequent processing.

[0131] Next, when the smoothed state vector does not match the historical train state data, for example, the historical trajectory shows a speed of 21 m / s, while the smoothed value is 19.8 m / s, further correction is required. Based on the covariance matrix estimation of the noise source, the Kalman gain matrix, i.e., the noise gain matrix, is obtained by multiplying its inverse matrix with the observed noise covariance matrix.

[0132] Then, the noise gain matrix is ​​used to smooth the state vector correction. For example, with a noise variance of 0.5 m / s², the Kalman gain coefficient is calculated. This coefficient determines the weighting of the measured and predicted values. For instance, based on historical data analysis, the measured value has a weight of 0.7, and the predicted value has a weight of 0.3. Subsequently, the smoothed state vector is fused with real-time environmental parameters, such as signal strength deviations caused by air pressure changes. This is achieved through air pressure-signal strength mapping, where the signal strength decreases by 0.2 dB for every 5 hPa drop in air pressure, to correct the measured value. The weighted measured value is then superimposed with the weighted predicted value. That is, the product of the original measured value, the measured value weight, and the air pressure compensation coefficient is added to the product of the original predicted value and the predicted value weight to obtain the corrected state vector. Finally, the coordinate transformation module converts the corrected physical quantities into position coordinates (x=151m, y=201m) and a velocity value of 20.2 m / s. This process dynamically adjusts the coupling effect of environmental factors and sensor noise to make the output closer to the actual motion state of the train.

[0133] Finally, for the corrected state vector, Kalman filtering is used to minimize the error. Optimized position information is obtained through recursive prediction and measurement updates. Specifically, based on the corrected state vector (position x=151m, y=201m) and the real-time signal strength measurement of -80.3dBm, Kalman filtering first predicts the current position through state transition (assuming the system dynamic model is uniform motion, the predicted value remains x=151m, y=201m), while simultaneously calculating the prediction error covariance. Then, in the measurement update stage, the signal strength is converted into a distance observation value. The path loss is used to convert -80.3dBm into the actual distance; for example, assuming a transmit power of 20dBm corresponds to a distance of 125.2m, the difference is compared with the predicted position. The predicted position and the observation value are fused using Kalman gain calculation, ultimately obtaining the optimized state vector (x=150.8m, y=200.9m), while simultaneously updating the error covariance. The entire process achieves continuous minimization of position error by dynamically balancing the reliability of the prediction model and the accuracy of real-time measurements.

[0134] In one possible implementation, if the train enters a high-noise environment, such as electromagnetic interference in urban areas, adaptive Kalman filtering can introduce additional environmental parameters, such as an electromagnetic interference intensity of 0.2 mV / m, to adjust the weights of the gain matrix and further optimize the position information. This extended scheme improves the adaptability of positioning in complex environments and ensures the reliability of the trajectory.

[0135] In step S17, the step of obtaining an initial signal from the multi-dimensional communication data stream and correcting it to obtain a calibration signal, and then fusing the optimized location information with the calibration signal to obtain a stable transmission dataset, includes:

[0136] An initial signal is obtained from a multi-dimensional communication data stream, and the initial signal is adaptively processed to obtain a preprocessed signal.

[0137] When the preprocessed signal is inconsistent with the signal output by the preset prediction equation, calibration is performed based on the preprocessed signal to obtain a calibration signal.

[0138] By combining the current channel parameters, the calibration signal and the optimized location information are integrated to obtain fused data;

[0139] Channel optimization processing is performed on the fused data to obtain a stable transmission dataset.

[0140] First, initial signals are acquired from multi-dimensional communication data streams through multi-source sensor data acquisition. For example, the train's communication module captures multi-dimensional information about the wireless signal, including signal delay, bandwidth occupancy, and signal-to-noise ratio (SNR). The initial signal might have a delay of 2ms, a bandwidth of 10MHz, and an SNR of 15dB. This data is acquired in real-time via the communication link between the base station and the train, providing a foundation for subsequent processing. The acquisition process needs to consider the dynamic changes caused by the train's high-speed movement to ensure data integrity.

[0141] Subsequently, adaptive preprocessing is performed on the initial signal to obtain the preprocessed signal. Adaptive processing adjusts processing parameters based on the signal's dynamic characteristics to smooth data fluctuations. Specifically, if the latency data fluctuates between 1.8ms and 2.2ms, a stable 2.05ms output can be achieved through dynamic weighted averaging. If the bandwidth data varies between 9.8MHz and 10.2MHz, it may stabilize at 10MHz after preprocessing. Because the fluctuations of various data types are inconsistent, traditional arithmetic averaging is not rigorous enough. Therefore, a sliding window variance calculation is used to dynamically smooth the signal strength. This method effectively reduces signal noise, improves data reliability, and lays the foundation for subsequent calibration.

[0142] Next, when the preprocessed signal is inconsistent with the signal output by the preset prediction equation, calibration is calculated based on path loss to obtain a calibrated signal. The path loss model recalculates compensation based on attenuation factors during signal propagation, such as distance and environmental interference. For example, the prediction equation assumes a delay of 2.1ms, while the preprocessed data is 2.05ms; the deviation may stem from changes in the distance between the train and the base station. The path loss estimation calibration coefficient is 1.02, resulting in a calibrated delay of 2.09ms. This calibration ensures that the signal data better matches the actual scenario. The path loss model here is consistent with that in S14, so it will not be described further.

[0143] Then, the calibration signal and optimized location information are integrated through data fusion, and fused data is generated by combining channel parameters. The data fusion process first calculates a signal compensation factor based on channel parameters, such as a fading coefficient of 0.8 and a multipath delay spread of 0.1ms. Multipath error compensation is applied to the delay (2.09ms) in the calibration signal, and fading adaptive adjustment is performed on the bandwidth (10MHz) to achieve weighted correction of communication parameters. Subsequently, the corrected communication parameters and optimized location information (x=150.8m, y=200.9m) are spatiotemporally aligned and correlated, ultimately generating fused data that simultaneously contains accurate location coordinates and anti-interference communication characteristics (such as calibrated delay, bandwidth, and channel compensation coefficients). This fused data improves the correlation between signal and location information.

[0144] Finally, for the fused data, Kalman filtering is used to optimize latency, bandwidth, and signal-to-noise ratio (SNR) to obtain a stable transmission dataset. For example, based on the fused data latency of 2.09ms and the real-time measurement of 2.08ms, Kalman filtering, through prediction and updating, optimizes it to 2.085ms. The SNR is improved from 15dB to 15.2dB. This optimization reduces data fluctuations and ensures the stability of the transmission dataset. For example, in high-interference environments such as urban areas, Kalman filtering can introduce additional channel parameters, such as interference power of 0.1mW, and dynamically adjust the calibration coefficients to further optimize the data. This extended approach improves data processing capabilities in complex scenarios and ensures the reliability of the communication link.

[0145] In step S18, the real-time acquisition of latency data and signal-to-noise ratio data of the stable transmission dataset is performed. When the latency data or signal-to-noise ratio data exceeds a preset threshold, the train's three-dimensional coordinate calculation is optimized to obtain optimized position data. The optimized position data is then fused with the train's historical motion state to obtain the final accurate communication data, including:

[0146] The system acquires the latency data and signal-to-noise ratio data of the stable transmission dataset in real time. When the latency data or the signal-to-noise ratio data exceeds a preset data threshold, the system performs deviation detection on the stable transmission dataset to obtain abnormal data points.

[0147] Triangulation is performed based on the signal strength and time difference of arrival of the abnormal data points to obtain optimized location data;

[0148] Based on the optimized position data, the characteristics of multipath effect and external noise are identified, and combined with the historical motion state of the train, the classified interference data is obtained.

[0149] The time delay data and signal-to-noise ratio data of the optimized location data are optimized, and the classified interference data and the optimized location data are fused to obtain the final accurate communication data.

[0150] First, latency and signal-to-noise ratio (SNR) data are obtained from a stable transmission dataset. Assuming the scenario is a communication system for a high-speed train, the dataset contains a latency of 2.1ms and an SNR of 14.8dB. The preset data thresholds might be 2.5ms latency and 12dB SNR. However, large errors in latency data can lead to coordinate inaccuracies in calculations, so the thresholds are often set lower. For example, if the latency data is within the threshold but the SNR is lower than expected, deviation detection is triggered. By analyzing historical train status data, SNR anomalies are identified, resulting in abnormal data points. For instance, if the SNR drops sharply from 15dB to 14.8dB within the past 10 seconds, it is marked as an anomaly, possibly due to signal attenuation caused by the train entering a tunnel. This analysis ensures accurate anomaly location, providing a basis for subsequent processing.

[0151] Subsequently, triangulation was performed on the abnormal data points. The device position was calculated using the signal strength and time difference of arrival (TDOA) of three base stations, resulting in optimized position data. The TDOA data (0.1 μs and 0.15 μs) were converted to distance differences using the speed of light, and a hyperbolic positioning equation system was constructed. Simultaneously, signal strength measurements (-70 dBm, -75 dBm, -72 dBm) were incorporated to correct for nonlinear errors. Finally, the least squares method was used to solve the equation system to obtain the optimized train position coordinates (x = 152.3 m, y = 198.7 m). The optimized position data reflects the real-time position of the train in a dynamic environment, improving the accuracy of subsequent data processing.

[0152] Next, based on the optimized location data and current channel data, multipath effects and external noise characteristics are identified. The channel data may show a fading coefficient of 0.85 and a delay spread of 0.12 ms caused by multipath. Analysis of historical train status data reveals that the multipath effect originates from reflections from tall buildings near the train, while external noise may come from co-channel interference from nearby base stations. After classification, the interference data is obtained, such as multipath effects accounting for 60% and noise accounting for 40%. This classification helps to accurately identify the source of interference, providing support for optimizing communication quality.

[0153] Finally, the classified interference data and optimized location data are integrated. The integration process combines the location coordinates x=152.3m, y=198.7m and the interference data to obtain a comprehensive dataset. The Kalman filter algorithm further optimizes the latency and signal-to-noise ratio, for example, reducing the latency from 2.1ms to 2.08ms and improving the signal-to-noise ratio from 14.8dB to 15.1dB. The optimized, accurate communication data ensures the stability of the train communication link, especially in high-interference scenarios such as urban areas, where data integration and filtering effectively improve signal reliability. This method, through multi-dimensional data fusion and optimization, guarantees the efficient operation of the communication system.

[0154] In summary, this invention discloses a communication data transmission method for a train approach warning intercom, comprising: collecting real-time signals including speed, direction, and distance parameters from the train's surrounding environment, and acquiring the train's historical motion state and multi-dimensional communication data stream; performing signal attenuation analysis and noise source identification on the real-time signals to obtain a preliminary position dataset and noise source identification results; when the signal strength in the preliminary position dataset is lower than a preset strength threshold, using a random forest algorithm to analyze the noise source identification results to obtain a compensation coefficient, adjusting the dataset based on the compensation coefficient and the preliminary position dataset to obtain an adjusted dataset; performing iterative three-dimensional coordinate calculation based on the adjusted dataset to obtain a preliminary position estimate; and performing signal attenuation correction on the preliminary position estimate. The corrected position trajectory data is obtained; when the corrected position trajectory data does not match the historical motion state of the train, a noise gain matrix is ​​calculated based on the noise source, and the corrected position trajectory data is optimized based on the noise gain matrix to obtain optimized position information; an initial signal is obtained from the multi-dimensional communication data stream and corrected to obtain a calibration signal, and the optimized position information is fused with the calibration signal to obtain a stable transmission dataset; the delay data and signal-to-noise ratio data of the stable transmission dataset are obtained in real time, and when the delay data or signal-to-noise ratio data exceeds a preset range threshold, the three-dimensional coordinate calculation of the train is optimized to obtain optimized position data, and the optimized position data is fused with the historical motion state of the train to obtain the final accurate communication data.

[0155] This invention uses adaptive data fusion to calibrate channel parameters and path loss models in real time, generating a stable transmission dataset that includes latency, bandwidth, and signal-to-noise ratio. This ensures real-time communication and reliability, achieving positioning accuracy and communication stability for trains in complex environments, and providing key technical support for the safety and efficiency of high-speed rail operation.

[0156] Reference Figure 2 The second embodiment of the present invention provides a communication data transmission device for a train approach warning intercom, comprising:

[0157] The data acquisition module is used to collect real-time signals, including speed, direction, and distance parameters, from the train's surrounding environment and to acquire the train's historical motion status.

[0158] The signal analysis module is used to perform signal attenuation analysis and noise source identification on the real-time signal to obtain preliminary location dataset and noise source identification results;

[0159] The signal compensation module is used to analyze the noise source identification results using a random forest algorithm to obtain compensation coefficients when the signal strength in the preliminary location dataset is lower than a preset strength threshold. The compensation coefficients and the preliminary location dataset are then adjusted to obtain an adjusted dataset.

[0160] The coordinate calculation module is used to perform iterative calculation of three-dimensional coordinates based on the adjusted dataset to obtain preliminary position estimates.

[0161] The coordinate correction module is used to correct the signal attenuation of the preliminary position estimate to obtain the corrected position trajectory data.

[0162] The coordinate optimization module is used to calculate a noise gain matrix based on the noise source when the corrected position trajectory data does not match the historical motion state of the train, and to optimize the corrected position trajectory data based on the noise gain matrix to obtain optimized position information.

[0163] The data fusion module is used to obtain initial signal data from multi-dimensional communication data streams and correct it to obtain calibration signal data, and then fuse the optimized location information with the calibration signal data to obtain a stable transmission dataset.

[0164] The data output module is used to acquire the latency data and signal-to-noise ratio data of the stable transmission dataset in real time. When the latency data or signal-to-noise ratio data exceeds a preset threshold, the three-dimensional coordinate calculation of the train is optimized to obtain optimized position data. The optimized position data and the historical motion state of the train are fused to obtain the final accurate communication data.

[0165] It should be noted that the communication data transmission device for a train approach warning walkie-talkie provided in this embodiment of the invention is used to execute all the process steps of the communication data transmission method for a train approach warning walkie-talkie in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0166] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a signal attenuation analysis program. When the processor executes the computer program, it implements the steps in the communication data transmission methods embodiments of the various train approach warning intercoms described above, for example... Figure 1 The step S11 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described device embodiments, such as the data acquisition module.

[0167] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0168] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0169] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0170] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0171] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0172] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0173] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that 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 for those skilled in the art.

Claims

1. A communication data transmission method for a train approach warning intercom, characterized in that, include: Real-time signals, including speed, direction, and distance parameters, are collected from the train's surrounding environment, and the train's historical motion status and multi-dimensional communication data streams are obtained. The real-time signal is subjected to signal attenuation analysis and noise source identification to obtain preliminary location dataset and noise source identification results; When the signal strength in the preliminary location dataset is lower than a preset strength threshold, the random forest algorithm is used to analyze the noise source identification results to obtain a compensation coefficient. The dataset is then adjusted based on the compensation coefficient and the preliminary location dataset to obtain the adjusted dataset. Based on the adjusted dataset, perform iterative calculation of the three-dimensional coordinates to obtain a preliminary position estimate. The preliminary position estimate is corrected for signal attenuation to obtain the corrected position trajectory data; When the corrected position trajectory data does not match the historical motion state of the train, a noise gain matrix is ​​calculated based on the noise source, and the corrected position trajectory data is optimized based on the noise gain matrix to obtain optimized position information. An initial signal is obtained from the multi-dimensional communication data stream and corrected to obtain a calibration signal. The optimized location information is then fused with the calibration signal to obtain a stable transmission dataset. The system acquires the latency data and signal-to-noise ratio data of the stable transmission dataset in real time. When the latency data or signal-to-noise ratio data exceeds a preset threshold, the system optimizes the calculation of the train's three-dimensional coordinates to obtain optimized position data. The system then fuses the optimized position data with the train's historical motion state to obtain the final accurate communication data. The step of performing iterative calculation of three-dimensional coordinates based on the adjusted dataset to obtain a preliminary position estimate includes: The base station delay, signal strength, and angle of arrival data in the adjusted dataset are obtained, and the base station delay, signal strength, and angle of arrival data are smoothed to obtain optimized multi-source data. The train's three-dimensional coordinates are iteratively calculated based on the optimized multi-source data to obtain preliminary three-dimensional coordinates. The characteristic parameters of temperature, humidity and electromagnetic interference are extracted from the pre-established environmental database, and the preliminary three-dimensional coordinates are corrected by multipath fading model to obtain the corrected three-dimensional coordinates. When the corrected three-dimensional coordinates meet the preset coordinate convergence threshold, the temperature compensation, humidity compensation and multipath fading correction parameters are integrated to obtain a preliminary position estimate. Specifically, the real-time acquisition of latency data and signal-to-noise ratio data from the stable transmission dataset, when the latency data or signal-to-noise ratio data exceeds a preset threshold, optimizes the train's three-dimensional coordinate calculation to obtain optimized position data, including: The system acquires the latency data and signal-to-noise ratio data of the stable transmission dataset in real time. When the latency data or the signal-to-noise ratio data exceeds a preset data threshold, the system performs deviation detection on the stable transmission dataset to obtain abnormal data points. Triangulation is performed based on the signal strength and time difference of arrival of the abnormal data points to obtain optimized location data; Among them, multi-dimensional communication data streams include signal delay, bandwidth usage, and signal-to-noise ratio.

2. The communication data transmission method for the train approach warning intercom according to claim 1, characterized in that, The step of performing signal attenuation analysis and noise source identification on the real-time signal to obtain preliminary location datasets and noise source identification results includes: The velocity direction and distance parameters in the real-time signal are converted from analog to digital to obtain the first signal dataset. The time-domain signal in the first signal dataset is converted into a frequency-domain signal to obtain a frequency-converted signal. The signal attenuation features are extracted from the frequency-converted signal to obtain an attenuation feature dataset. Based on the attenuation feature dataset, noise sources are separated to obtain a noise separation dataset, and the noise source type is determined to obtain the noise source identification result; A preliminary location dataset is obtained by performing a Kalman filter fusion operation on the noise separation dataset.

3. The communication data transmission method for the train approach warning intercom according to claim 1, characterized in that, When the signal strength in the preliminary location dataset is lower than a preset strength threshold, a random forest algorithm is used to analyze the noise source identification results to obtain a compensation coefficient. The dataset is then adjusted based on the compensation coefficient and the preliminary location dataset to obtain an adjusted dataset, including: When the signal strength is lower than a preset signal strength threshold, it is determined that there is interference, and the type of interference is determined based on the noise source identification result. The environmental noise characteristics are obtained by extracting the noise source features corresponding to the interference type from a pre-established noise source database. The environmental noise characteristics were analyzed and calculated using the random forest algorithm to obtain compensation coefficients; The preliminary location dataset is weighted and adjusted according to the compensation coefficient to obtain the adjusted dataset.

4. The communication data transmission method for the train approach warning intercom according to claim 1, characterized in that, The step of performing signal attenuation correction on the preliminary position estimate to obtain corrected position trajectory data includes: Obtain the time delay data and environmental parameters from the preliminary position estimate, and smooth the time delay data and environmental parameters to obtain the smoothed state vector; The preliminary position trajectory is obtained by iterative calculation based on the smoothed state vector. If the initial position trajectory does not reach the preset convergence threshold, the time delay data and the environmental parameters are integrated to obtain the corrected state vector; The corrected position trajectory data is obtained by optimizing the state vector in conjunction with the signal attenuation model.

5. The communication data transmission method for the train approach warning intercom according to claim 1, characterized in that, When the corrected position trajectory data does not match the train's historical motion state, a noise gain matrix is ​​calculated based on the noise source, and the corrected position trajectory data is optimized based on the noise gain matrix to obtain optimized position information, including: An initial state vector is obtained from sensor data and environmental parameters in the corrected position trajectory data, and a smoothed state vector is obtained by smoothing the initial state vector. When the smoothed state vector does not match the train's historical motion state, the noise gain matrix is ​​calculated based on the noise source. The smoothed state vector is corrected based on the noise gain matrix and combined with environmental parameters to obtain the corrected state vector; The corrected state vector is subjected to error minimization processing to obtain optimized position information.

6. The communication data transmission method for the train approach warning intercom according to claim 1, characterized in that, The process of obtaining an initial signal from the multi-dimensional communication data stream and correcting it to obtain a calibration signal, and then fusing the optimized location information with the calibration signal to obtain a stable transmission dataset, includes: An initial signal is obtained from a multi-dimensional communication data stream, and the initial signal is adaptively processed to obtain a preprocessed signal. When the preprocessed signal is inconsistent with the signal output by the preset prediction equation, calibration is performed based on the preprocessed signal to obtain a calibration signal. By combining the current channel parameters, the calibration signal and the optimized location information are integrated to obtain fused data; Channel optimization processing is performed on the fused data to obtain a stable transmission dataset.

7. The communication data transmission method for the train approach warning intercom according to claim 1, characterized in that, The process of fusing the optimized position data and the train's historical motion state to obtain the final accurate communication data includes: Based on the optimized position data, the characteristics of multipath effect and external noise are identified, and combined with the historical motion state of the train, the classified interference data is obtained. The time delay data and signal-to-noise ratio data of the optimized location data are optimized, and the classified interference data and the optimized location data are fused to obtain the final accurate communication data.

8. A communication data transmission device for a train approach warning intercom, used to implement the communication data transmission method of the train approach warning intercom as described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to collect real-time signals, including speed, direction, and distance parameters, from the train's surrounding environment and to acquire the train's historical motion status. The signal analysis module is used to perform signal attenuation analysis and noise source identification on the real-time signal to obtain preliminary location dataset and noise source identification results; The signal compensation module is used to analyze the noise source identification results using a random forest algorithm to obtain compensation coefficients when the signal strength in the preliminary location dataset is lower than a preset strength threshold. The compensation coefficients and the preliminary location dataset are then adjusted to obtain an adjusted dataset. The coordinate calculation module is used to perform iterative calculation of three-dimensional coordinates based on the adjusted dataset to obtain preliminary position estimates. The coordinate correction module is used to correct the signal attenuation of the preliminary position estimate to obtain the corrected position trajectory data. The coordinate optimization module is used to calculate a noise gain matrix based on the noise source when the corrected position trajectory data does not match the historical motion state of the train, and to optimize the corrected position trajectory data based on the noise gain matrix to obtain optimized position information. The data fusion module is used to obtain initial signal data from multi-dimensional communication data streams and correct it to obtain calibration signal data, and then fuse the optimized location information with the calibration signal data to obtain a stable transmission dataset. The data output module is used to acquire the latency data and signal-to-noise ratio data of the stable transmission dataset in real time. When the latency data or signal-to-noise ratio data exceeds a preset threshold, the three-dimensional coordinate calculation of the train is optimized to obtain optimized position data. The optimized position data and the historical motion state of the train are fused to obtain the final accurate communication data.

Images