Method and system for monitoring the transport state of a nose assembly based on multi-source sensor fusion

Abstract

The application relates to the technical field of rail transit equipment, and discloses a nose assembly transportation state monitoring method and system based on multi-source sensing fusion, which comprises the following steps: collecting interval data, vibration signals and real-time position data of a transportation device of a nose assembly transportation scene, fusing original data sets after time sequence alignment and abnormality cleaning; obtaining a smooth interval data set through state prediction and error correction iteration optimization; extracting time-frequency joint features to calculate time sequence fluctuation values, and identifying a continuous fluctuation abnormal interval; correcting a real-time offset vector of the device and mapping the real-time offset vector to a transportation space coordinate system to complete collision detection, combining device operation parameters to grade evaluate collision risks, generating a warning instruction sequence when the collision risks exceed a threshold, and dynamically adjusting a transportation device posture and an interval. The method can realize global accurate real-time monitoring and active risk prevention and control of a nose assembly transportation state, and effectively guarantees the safety and stability of rail transit equipment transportation.

Description

Technical Field

[0001] This invention relates to the field of rail transit equipment technology, and in particular to a method and system for monitoring the transportation status of a front-end component based on multi-source sensor fusion. Background Technology

[0002] Currently, in the rail transit equipment industry, the safety status monitoring of the transportation of the nose section components is a key aspect of preventing equipment collisions and avoiding transportation risks, and has become a major research focus in the field.

[0003] In existing technologies, single-sensor monitoring or manual inspection is mainly used to ensure the safety of the machine head assembly during transportation. This relies heavily on fixed distance thresholds or simple vibration amplitude assessments, such as using preset safety distances to determine the transportation status and using basic displacement data to determine equipment deviation. However, in complex operating environments, transportation conditions can fluctuate dynamically, such as road bumps. Fixed distance threshold monitoring is easily affected by road bumps, environmental noise, and vibration interference. It is also affected by narrow transportation paths, variable equipment postures, and multi-equipment collaborative transportation, resulting in insufficient accuracy in transportation status monitoring and delayed collision risk identification. Furthermore, existing technologies lack the ability to fuse multi-source sensor data and analyze abnormal fluctuations in real time. Especially in complex scenarios such as tunnel transportation and parallel operation of multiple devices, it is difficult to quickly identify deviation anomalies and trigger warnings, leading to frequent transportation safety hazards.

[0004] In summary, existing technologies are insufficient to achieve accurate monitoring of the entire transportation status of the nose section components, and cannot meet the core requirements of the rail transit equipment industry for equipment transportation safety and real-time status monitoring when facing complex transportation scenarios. Summary of the Invention

[0005] This invention provides a method and system for monitoring the transportation status of the nose section components based on multi-source sensor fusion, so as to achieve accurate monitoring of the transportation status of the nose section components across the entire domain, and meet the core requirements of the rail transit equipment industry for equipment transportation safety and real-time status monitoring when facing complex transportation scenarios.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a method for monitoring the transportation status of a nose assembly based on multi-source sensor fusion, comprising: Acquire spacing data, vibration signals, and real-time location data of the transportation equipment in the transportation scenario of the nose assembly; By fusing the spacing data with the vibration signal, the original dataset is obtained; The predicted spacing state and the actual spacing measurement value are calculated based on the original dataset. The original dataset is then iteratively optimized based on the predicted spacing state and the actual spacing measurement value to obtain a smoothed spacing dataset. Extract the dynamic change features of the smooth spacing dataset and the amplitude-frequency change features of the vibration signal, calculate the time-series fluctuation value based on the dynamic change features and the amplitude-frequency change features, and filter the data range where the time-series fluctuation value exceeds the preset fluctuation judgment threshold to obtain the fluctuation abnormal range; If the length of the fluctuation abnormal interval exceeds the preset interval abnormal threshold, then the real-time offset vector is extracted from the real-time location data, and the real-time offset vector is error-corrected to obtain the corrected offset vector. The corrected offset vector is mapped to a preset transportation space coordinate system and collision detection is performed to obtain potential collision data; Based on the potential collision data and the pre-acquired operating parameters of the transportation equipment, the collision risk level is determined. If the collision risk level is higher than the preset risk judgment threshold, a warning instruction sequence is generated. Adjust the posture and spacing of the transport equipment according to the sequence of warning instructions.

[0007] Secondly, the present invention provides a nose assembly transportation status monitoring system based on multi-source sensor fusion, comprising: The data acquisition module is used to acquire spacing data, vibration signals, and real-time location data of the transportation equipment in the transportation scenario of the nose assembly. The data fusion module is used to fuse the spacing data and the vibration signal to obtain the original dataset; The data optimization module is used to calculate the predicted spacing state value and the actual spacing measurement value based on the original dataset, and to perform iterative optimization processing on the original dataset based on the predicted spacing state value and the actual spacing measurement value to obtain a smoothed spacing dataset. The interval identification module is used to extract the dynamic change features of the smooth interval dataset and the amplitude-frequency change features of the vibration signal, calculate the time-series fluctuation value based on the dynamic change features and the amplitude-frequency change features, filter the data intervals where the time-series fluctuation value exceeds the preset fluctuation judgment threshold, and obtain the fluctuation abnormal interval. The vector processing module is used to extract a real-time offset vector from the real-time location data if the length of the fluctuation abnormal interval exceeds a preset interval abnormal threshold, and to perform error correction on the real-time offset vector to obtain a corrected offset vector. The collision detection module is used to map the correction offset vector to a preset transportation space coordinate system and perform collision detection to obtain potential collision data. The risk assessment module is used to determine the collision risk level based on the potential collision data and the pre-acquired operating parameters of the transportation equipment. If the collision risk level is higher than the preset risk judgment threshold, a warning instruction sequence is generated. The calibration and control module is used to adjust the attitude and spacing of the transport equipment according to the warning command sequence.

[0008] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention collects spacing data, vibration signals and real-time position data of the equipment in the transportation scenario of the machine head assembly, aligns the time sequence nodes to complete the fusion of multi-source data, and generates a smooth spacing dataset through outlier removal and iterative optimization filtering. It breaks through the limitations of traditional single sensor monitoring and the limited coverage of manual inspection, explores the multi-dimensional dynamic characteristics of the equipment transportation process, eliminates the influence of road bumps, environmental noise and vibration interference, provides high-precision basic data support for transportation status monitoring, effectively improves the reliability of multi-source sensor data acquisition, and solves the problems of data time sequence misalignment and insufficient monitoring accuracy in traditional methods.

[0009] (2) This invention extracts the dynamic change features of the smooth spacing dataset and the amplitude and frequency change features of the vibration signal, weighted and fused to calculate the time-series fluctuation values ​​to screen the abnormal fluctuation range. When the threshold is exceeded, the real-time offset vector is extracted and the error correction is completed to obtain the corrected offset vector. This invention breaks through the limitations of traditional fixed threshold judgment which is easily affected by working conditions and the lag in abnormal fluctuation identification. It accurately captures the correlation features between equipment offset abnormality and working condition fluctuation, provides multi-dimensional judgment basis for risk prediction, significantly improves the accuracy of abnormal identification in complex scenarios in narrow alleyways, and makes up for the shortcomings of existing technologies that cannot distinguish between occasional fluctuations and continuous offsets.

[0010] (3) This invention obtains potential collision data by mapping the correction offset vector to the transportation space coordinate system, and determines the collision risk level by combining the equipment operation parameters. When the threshold is exceeded, a warning instruction sequence is generated and the attitude and spacing of the transportation equipment are dynamically adjusted. This invention breaks through the limitations of traditional methods that lack multi-dimensional risk closed-loop management and dynamic control capabilities, and provides real-time and accurate anti-collision warning basis for the transportation of the head assembly. It solves the problems of inaccurate collision risk identification and lagging safety management in the roadway transportation scenario, and takes into account both real-time monitoring and transportation safety, thus meeting the core requirements of the rail transit equipment industry for the transportation safety of the head assembly. Attached Figure Description

[0011] Figure 1 This is a schematic flowchart of a method for monitoring the transportation status of a nose assembly based on multi-source sensor fusion, provided in the first embodiment of the present invention. Figure 2 This is a schematic diagram of a nose assembly transportation status monitoring system based on multi-source sensor fusion provided in the second embodiment of the present invention. Detailed Implementation

[0012] 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.

[0013] Reference Figure 1 The first embodiment of the present invention provides a method for monitoring the transportation status of a nose assembly based on multi-source sensor fusion, comprising the following steps: S101, acquire spacing data, vibration signals and real-time position data of the transportation equipment in the transportation scenario of the nose assembly; S102, the spacing data and the vibration signal are fused to obtain the original dataset; S103, calculate the predicted spacing state value and the actual spacing measurement value based on the original dataset, and perform iterative optimization processing on the original dataset based on the predicted spacing state value and the actual spacing measurement value to obtain a smooth spacing dataset; S104, extract the dynamic change features of the smooth spacing dataset and the amplitude-frequency change features of the vibration signal, calculate the time-series fluctuation value based on the dynamic change features and the amplitude-frequency change features, filter the data range where the time-series fluctuation value exceeds the preset fluctuation judgment threshold, and obtain the fluctuation abnormal range. S105, if the length of the fluctuation abnormal interval exceeds the preset interval abnormal threshold, then extract the real-time offset vector from the real-time location data, perform error correction on the real-time offset vector, and obtain the corrected offset vector. S106, the correction offset vector is mapped to a preset transportation space coordinate system and collision detection is performed to obtain potential collision data; S107, Based on the potential collision data and combined with the pre-acquired operating parameters of the transportation equipment, determine the collision risk level. If the collision risk level is higher than the preset risk judgment threshold, generate a warning instruction sequence. S108, adjust the posture and spacing of the transport equipment according to the warning instruction sequence.

[0014] In step S101, acquiring the spacing data, vibration signal, and real-time position data of the transportation equipment for the nose assembly transportation scenario includes: Collect real-time distance data between the transport equipment and the tunnel wall to obtain spacing data; The vibration amplitude values ​​during the operation of the transportation equipment are collected to obtain vibration signals; Collect the spatial coordinate values ​​of the transportation equipment to obtain the real-time location data of the transportation equipment.

[0015] It should be noted that, firstly, real-time distance values ​​between the transport equipment and the tunnel wall are collected to obtain spacing data. Data acquisition is accomplished using laser rangefinders deployed at multiple points. These sensors are installed at key locations on the front, rear, and both sides of the transport equipment, achieving a measurement accuracy of ±1 mm. The sampling frequency is set according to the real-time monitoring requirements of the tunnel transport scenario. The sampling frequency threshold is set based on historical operational data of the machine head assembly tunnel transport over the past year, with a base sampling frequency of 100 Hz. For high-precision transport scenarios in narrow tunnels, the frequency can be increased to 200 Hz, while for conventional transport scenarios in wide tunnels, it can be decreased to 50 Hz. All collected distance values ​​are mapped to the 0-1 range using the min-max normalization method. The data is complete in time sequence and free from acquisition bias, and can be directly used for subsequent multi-source data fusion processing.

[0016] For example, during the tunnel transportation of the machine head assembly, the distance values ​​between the equipment and the left and right tunnel walls are collected in real time at a frequency of 100 Hz by laser rangefinders on both sides of the machine body, and standardized spacing data are generated after normalization processing.

[0017] Subsequently, vibration amplitude values ​​during the operation of the transportation equipment were collected to obtain vibration signals. Data acquisition was accomplished using a triaxial accelerometer, which was installed on the transport equipment's carrying platform and traveling mechanism. The sensitivity was set to 0.01 grams, and the sampling frequency was set according to the monitoring requirements of road surface bumps. The sampling frequency was set based on statistical data of road surface vibration interference over the past year, with a base sampling frequency of 200 Hz. For high-frequency vibration scenarios on bumpy roads, the frequency could be increased to 500 Hz, while for routine transportation scenarios on smooth roads, it could be decreased to 100 Hz. The collected vibration amplitude values ​​were simultaneously processed using minimum-maximum normalization to maintain a consistent data range with the spacing data, enabling accurate capture of subtle vibrations and road surface bumps during equipment operation.

[0018] For example, by using an accelerometer on the carrying platform, the vertical and horizontal vibration amplitudes during transportation are collected at a frequency of 200 Hz, generating a continuous vibration signal sequence.

[0019] The spatial coordinates of the transportation equipment are collected to obtain its real-time location data. Data acquisition relies on a positioning network composed of a BeiDou high-precision positioning module and UWB positioning base stations within the tunnel, achieving centimeter-level positioning accuracy. The location data update frequency is synchronized with the sampling frequency of the spacing data, ensuring temporal alignment of multi-source data. The update frequency of the positioning data can be adaptively adjusted according to the operating speed of the transportation equipment; the faster the equipment operates, the higher the update frequency, ensuring the real-time and continuous nature of the location data. All spatial coordinate values ​​are uniformly mapped to the preset transportation spatial coordinate system of the tunnel, eliminating coordinate misalignment and data delay issues. For example, through UWB positioning base stations and vehicle-mounted positioning modules, the three-dimensional spatial coordinates of the transportation equipment are synchronously collected at a frequency of 100 Hz, generating real-time location data that is temporally aligned with the spacing data.

[0020] In step S102, fusing the spacing data with the vibration signal to obtain the original dataset includes: Align the spacing data with the timing nodes of the vibration signal to obtain timing synchronization data; Data exceeding a preset amplitude threshold is removed from the timing synchronization data to obtain cleaned data; The cleaned data is integrated to generate the original dataset.

[0021] It should be noted that, firstly, the timing nodes of the spacing data and vibration signal are aligned to obtain time-synchronized data. Timing alignment employs a precise timestamp matching algorithm, using the unified clock of the transportation equipment control system as a reference. The sampling time points of the spacing data and vibration signal are matched point-by-point, with matching accuracy controlled at the millisecond level. This completely eliminates timing misalignment issues caused by differences in sampling frequencies between the two types of sensors. The matched data is then arranged continuously in the order of sampling. For example, the spacing value corresponding to each millisecond-level timestamp is paired with the vibration amplitude to form time-synchronized data without any timing misalignment.

[0022] It is worth noting that the amplitude judgment threshold is set based on the statistical data of normal sensor acquisition during the transportation of machine head components in tunnels over the past year. The basic judgment range for spacing values ​​is set to 0.3 meters to 1.5 meters, and the basic judgment range for vibration amplitude is set to 0 grams to 0.5 grams. In narrow tunnel transportation scenarios, the spacing range can be tightened to 0.5 meters to 1.2 meters, and in bumpy road transportation scenarios, the vibration amplitude range can be widened to 0 grams to 0.8 grams. Data exceeding the preset amplitude judgment threshold in the time-series synchronous data is removed to obtain cleaned data. The removal operation directly removes out-of-bounds sampling points, retaining the continuous time series and complete features of valid data. For example, removing out-of-bounds data with a spacing value of 0.2 meters and a vibration amplitude of 0.9 grams yields cleaned data that retains only reasonable values.

[0023] Finally, the cleaned data is integrated to generate the original dataset. The integration operation concatenates the spacing data and vibration signals according to their sampling time sequence, forming a structured dataset containing time-series identifiers, spacing values, and vibration amplitudes. The dataset format is adapted to the input requirements of subsequent Kalman filter iterative optimization. All data retains the previously completed min-max normalization results, maintaining a uniform value range of 0 to 1, without repeating the normalization operation. For example, by concatenating all valid cleaned data in chronological order, an original dataset that can be directly used for spacing state prediction is generated.

[0024] In step S103, the step of calculating the predicted spacing state value and the actual spacing measurement value based on the original dataset, and iteratively optimizing the original dataset based on the predicted spacing state value and the actual spacing measurement value to obtain a smoothed spacing dataset includes: Retrieve pre-acquired historical spacing data, extract the average spacing value and spacing change rate of the historical spacing data, and construct spacing status parameters; Perform a time-series recursive calculation based on the spacing state parameters to generate a spacing state prediction value; Extract the spacing values ​​from the original dataset to determine the actual spacing measurements; Calculate the error covariance between the predicted spacing state value and the actual spacing measurement value; Based on the error covariance, a gain coefficient is calculated by combining the percentage of the numerical deviation between the predicted spacing state value and the measured spacing value, and an error correction parameter is generated based on the gain coefficient. The spacing acquisition values ​​are corrected according to the error correction parameters to obtain corrected spacing data; Calculate the numerical deviation between the predicted spacing state value and the corrected spacing data to obtain the iterative deviation value; The iteration deviation value is optimized until it is lower than a preset error judgment threshold to obtain a smooth spacing dataset.

[0025] It should be noted that, firstly, pre-acquired historical spacing data is retrieved, and the average spacing value and spacing change rate of the historical spacing data are extracted to construct spacing status parameters. The historical spacing data comes from stable operation data collection sequences of the same type of transport equipment in the same tunnel environment over the past year. The average spacing value is obtained by performing an arithmetic mean operation on the valid historical data. The spacing change rate is obtained by calculating the ratio of the numerical difference between adjacent sampling points in the historical sequence to the sampling time interval. The final constructed spacing status parameters include two dimensions: the baseline mean spacing value and the dynamic change rate. During parameter initialization, an initial state covariance matrix is ​​simultaneously set. The initial value of the matrix is ​​set according to the dispersion of the historical data to ensure that the status parameters can be directly used for subsequent time-series recursive calculations. For example, retrieving historical spacing data from tunnel transport over the past year yields an average spacing value of 0.8 meters and a spacing change rate of 0.02 meters per second, completing the construction of the two-dimensional spacing status parameters.

[0026] Subsequently, a time-series recursive calculation is performed based on the spacing state parameters to generate the spacing state prediction value. The time-series recursive calculation is implemented using a Kalman filter state prediction equation. The recursive step size is kept completely consistent with the sampling period of the spacing data. A process noise covariance matrix is ​​introduced into the calculation process. The matrix value is set according to the statistical data of the bumpiness of the transportation road surface. The base value is set to 0.005, which can be adjusted upward to 0.01 for bumpy roads and downward to 0.002 for smooth roads. After the recursive calculation is completed, the spacing state prediction value at the next sampling time is output. The predicted value is time-aligned with the actual measured value.

[0027] For example, based on the constructed spacing state parameters, the state prediction operation is performed with a recursive step size of 10 milliseconds, generating a spacing state prediction value of 0.78 meters for the next moment.

[0028] Next, the spacing data collected from the original dataset are extracted to determine the actual spacing measurements. The extraction process strictly matches the time nodes of the time-series recursion, selecting valid spacing data collected at the same time as the predicted value from the original dataset. These values ​​have already undergone outlier removal and normalization, eliminating the need for repeated data cleaning. The final determined actual spacing measurements and the predicted spacing status values ​​are within the same time dimension and numerical range. For example, extracting the valid spacing data corresponding to 10 milliseconds from the original dataset determines the actual spacing measurement to be 0.81 meters.

[0029] Then, the error covariance between the predicted spacing state value and the actual measured spacing value is calculated. The error covariance calculation is implemented using the Kalman filter covariance update equation. The calculation process simultaneously combines the process noise covariance and the measurement noise covariance. The measurement noise covariance is set based on the measurement accuracy statistics of the laser rangefinder sensor, with a base value of 0.002. For high-precision sensors, the value can be lowered to 0.001, and for conventional precision sensors, it can be raised to 0.003. After the calculation is completed, the iteratively updated error covariance matrix is ​​output for subsequent gain coefficient calculation.

[0030] Subsequently, based on the error covariance and the proportion of numerical deviation between the predicted and measured spacing values, a gain coefficient is calculated. Error correction parameters are then generated based on this gain coefficient. The gain coefficient is a Kalman gain, calculated using the error covariance as the core input. The proportion of the absolute difference between the predicted and measured spacing values ​​is also incorporated as a weighted correction term. The resulting gain coefficient dynamically adjusts to a range between 0.4 and 0.6. The error correction parameters are then generated based on the product of the gain coefficient and the prediction residual. These parameters directly adapt to the correction requirements of the collected spacing values. For example, by combining the error covariance and the proportion of numerical deviation, a gain coefficient of 0.52 is obtained, and the corresponding error correction parameters are calculated simultaneously.

[0031] Next, the spacing data is corrected based on the error correction parameters to obtain the corrected spacing data. The correction calculation is achieved by weighted fusion of predicted and measured values, using the error correction parameters as weights to dynamically correct the original spacing data. The correction process fully preserves the temporal characteristics of the data while effectively filtering out random noise interference caused by road surface bumps. The corrected spacing data significantly improves the fit with the actual tunnel spacing. For example, correcting the original spacing data of 0.81 meters using the error correction parameters yields a corrected spacing data of 0.795 meters.

[0032] Next, the numerical deviation between the predicted spacing and the corrected spacing data is calculated to obtain the iterative deviation value. The numerical deviation is the absolute difference between the two sets of data. The difference result is mapped to the interval between 0 and 1 using the minimum-maximum normalization method to eliminate the judgment bias caused by the magnitude of the numerical values. The final iterative deviation value can directly reflect the correction effect of this iteration; the smaller the value, the higher the correction accuracy. For example, calculating the absolute difference between the predicted value of 0.78 meters and the corrected spacing data of 0.795 meters yields an iterative deviation value of 0.015 meters.

[0033] Finally, the iteration deviation value is optimized until it falls below a preset error judgment threshold, resulting in a smoothed spacing dataset. The error judgment threshold is set based on the accuracy requirements of the head assembly transportation spacing monitoring over the past year. The basic threshold is set to 0.001, which can be lowered to 0.0005 for high-precision roadway transportation scenarios and raised to 0.002 for regular transportation scenarios. The maximum number of iterations is set to 100 to avoid infinite loops. When the iteration deviation value falls below the threshold, the iteration is terminated and all corrected time-series data are output, forming a smoothed spacing dataset.

[0034] For example, when the iteration reaches the 8th iteration, the iteration deviation value drops to 0.0008, which is lower than the basic threshold of 0.001. The iteration is then terminated and the complete smooth spacing dataset is output.

[0035] In step S104, the dynamic change features of the smoothing spacing dataset and the amplitude-frequency change features of the vibration signal are extracted. Based on these dynamic change features and amplitude-frequency change features, time-series fluctuation values ​​are calculated. Data intervals where the time-series fluctuation values ​​exceed a preset fluctuation judgment threshold are filtered to obtain abnormal fluctuation intervals, including: The amplitude and frequency variation features of the smooth interval dataset are extracted from the time series to obtain the dynamic variation features; Extract the amplitude-frequency variation characteristics of the vibration signal; Calculate time-series statistical values ​​based on the amplitude-frequency variation characteristics and the dynamic variation characteristics, and determine the short-term variation trend of the data based on the time-series statistical values; Based on the short-term change trend, calculate the time-series fluctuation value of the smooth interval dataset, and filter out sequence segments whose time-series fluctuation values ​​exceed the preset fluctuation judgment threshold. Locate the time interval corresponding to the sequence segment and perform interval marking to obtain the fluctuation anomaly interval.

[0036] It should be noted that, firstly, amplitude and frequency variation features are extracted from the time series of the smoothed spacing dataset to obtain dynamic variation features. Amplitude variation features are extracted using a sliding window statistical method. The window size is statistically set based on the time series variation patterns of the machine head assembly's roadway transportation over the past year. The basic window length is set to 10 seconds, and the sliding step size is set to 1 second. For high-precision monitoring scenarios, the window length can be shortened to 5 seconds, and for regular transportation scenarios, it can be extended to 15 seconds. The mean, standard deviation, peak value, and range are calculated for the time series data within each window to form an amplitude variation feature set. Frequency variation features are extracted using Fast Fourier Transform (FFT) technology. Frequency domain transformation is performed on the time series data within each sliding window to extract the dominant frequency values ​​and spectral energy distribution within the range of 0.5 Hz to 10 Hz, forming a frequency variation feature set. All feature values ​​are mapped to the 0-1 range using the min-max normalization method, and finally concatenated to form a complete dynamic variation feature set. The feature dimensions are completely aligned with the time series, with no data misalignment or feature loss issues.

[0037] For example, a 10-second window is used to perform sliding statistics on the smooth interval time series, and the amplitude standard deviation and the dominant frequency value are calculated. After normalization, a standardized dynamic change characteristic is formed.

[0038] Subsequently, the amplitude-frequency variation characteristics of the vibration signal are extracted. These characteristics include time-domain amplitude features and frequency-domain distribution features. Time-domain amplitude features are extracted using a sliding window statistical method, with window parameters identical to those of the spacing data to ensure temporal alignment. The peak effective value kurtosis and variance of the vibration signal within each window are calculated to form the time-domain amplitude features. Frequency-domain distribution features are extracted using Fast Fourier Transform (FFT) technology. A frequency domain transformation is performed on the vibration signal within the window, locking onto the road surface bump-sensitive frequency band from 0.5 Hz to 10 Hz. The peak energy percentage and dominant frequency within this band are extracted to form the frequency-domain distribution features. All feature values ​​undergo simultaneous min-max normalization to maintain a consistent numerical range with the dynamic variation features. The resulting concatenation forms the complete amplitude-frequency variation characteristics of the vibration signal.

[0039] In this implementation case, time-series statistical values ​​are calculated based on amplitude-frequency variation characteristics and dynamic variation characteristics, and the short-term trend of the data is determined based on these statistical values. The time-series statistical values ​​are calculated using a multi-feature weighted fusion algorithm. The fusion weights are set based on the accuracy data statistics of historical transportation anomaly identification. The weight for dynamic variation characteristics is set to 0.6, and the weight for vibration signal amplitude-frequency variation characteristics is set to 0.4. This weighting is suitable for the anomaly identification needs of roadway transportation scenarios. Dynamic spacing changes are more indicative of offset anomalies and are given a higher weight, while vibration signals reflect road surface bumps and interference and are given a lower weight. After weighted fusion, continuous time-series statistical values ​​are obtained. Then, an autoregressive moving average model is used to fit the time-series statistical values. The autoregression order of the model is set to 2, and the moving average order is set to 1. The fitting yields the short-term trend of the data within a 5-second prediction step. The trend results are divided into three categories: stable, rising, and falling, which can intuitively reflect the joint change law of spacing and vibration.

[0040] Next, based on short-term trends, the temporal fluctuation values ​​of the smoothed interval dataset are calculated, and sequence segments with temporal fluctuation values ​​exceeding a preset fluctuation threshold are filtered out. The temporal fluctuation values ​​are calculated by combining the short-term trends and the temporal dispersion of the smoothed interval dataset. The value is a weighted sum of the absolute value of the trend slope and the standard deviation of the data within the window, with weights of 0.7 and 0.3 respectively. A higher value indicates a more severe fluctuation in the interval data. The fluctuation threshold is set based on the statistical analysis of temporal fluctuation data under normal transportation conditions over the past year. The basic threshold is set at 0.8 mm, which can be lowered to 0.5 mm for high-precision tunnel transportation scenarios and raised to 1.2 mm for regular transportation scenarios. The temporal fluctuation values ​​corresponding to each sliding window are compared point-by-point with the preset fluctuation threshold, and all data points with values ​​exceeding the threshold are filtered out. Data points continuously exceeding the threshold are combined to form sequence segments, which retain complete temporal identifiers and numerical characteristics.

[0041] For example, if the calculated temporal fluctuation value of a certain window is 1.1 mm, which exceeds the basic threshold of 0.8 mm, the continuous data point is marked as an abnormal sequence segment.

[0042] Finally, the time intervals corresponding to the sequence segments are located and interval marking is performed to obtain the fluctuation anomaly intervals. Based on the timestamps of the start and end data points of the sequence segments, the time intervals are converted to millisecond-level time intervals. The interval boundaries accurately match the sampling time sequence, without interval truncation or extension issues. Each interval is then uniquely identified, with the marking content including the start and end times of the interval, the maximum fluctuation value, and the corresponding trend. All marked intervals are arranged in chronological order to form a complete fluctuation anomaly interval. For example, locating the 12-second to 18-second time interval corresponding to the abnormal sequence segment, and generating the corresponding fluctuation anomaly interval after interval marking.

[0043] In step S105, if the length of the fluctuation anomaly interval exceeds a preset interval anomaly threshold, a real-time offset vector is extracted from the real-time location data, and error correction is performed on the real-time offset vector to obtain a corrected offset vector, including: Calculate the continuous displacement difference of the real-time position data to generate a real-time offset vector; The interference error in the real-time offset vector is filtered out to obtain the corrected offset vector.

[0044] It should be noted that the interval anomaly threshold is set based on historical data of abnormal fluctuations in the transportation of machine head components in the tunnel over the past year. The shortest abnormal interval duration that could trigger equipment offset collision risks is statistically analyzed, and the base threshold is set to 5 seconds. This can be lowered to 3 seconds for high-precision transportation scenarios in narrow tunnels, and increased to 8 seconds for conventional transportation scenarios in wide tunnels. This threshold has been verified in multiple scenarios and can reliably distinguish between occasional fluctuations and persistent offset anomalies. If the length of an abnormal fluctuation interval exceeds this preset threshold, the subsequent offset vector extraction and correction process is initiated. For example, if the duration of an abnormal fluctuation interval is 6 seconds, exceeding the base threshold of 5 seconds, the system automatically triggers the offset vector extraction and correction operation.

[0045] Subsequently, the continuous displacement difference of the real-time location data is calculated to generate a real-time offset vector. The real-time location data comes from a positioning network composed of UWB positioning base stations and vehicle-mounted Beidou high-precision positioning modules within the tunnel. The data update frequency is kept completely consistent with the interval data sampling frequency to ensure perfect timing alignment. The continuous displacement difference calculation is based on the preset transportation space coordinate system of the tunnel. The numerical differences of the equipment's spatial coordinates in the horizontal, vertical, and vertical dimensions are calculated at adjacent sampling times. The horizontal dimension is perpendicular to the direction of equipment movement, the vertical dimension is the direction of equipment movement, and the vertical dimension is the equipment height direction. The displacement rate in each dimension is then calculated in conjunction with the sampling time interval. Finally, a real-time offset vector containing three-dimensional displacement components and offset direction angles is generated. The calculation accuracy of the offset direction angle is controlled within ±0.5 degrees. All vector data are mapped to the interval of 0 to 1 using the minimum-maximum normalization method to maintain a unified standard with the previously processed data.

[0046] For example, by calculating the difference in device spatial coordinates between adjacent 10-millisecond sampling times, a lateral displacement of 0.02 meters and a longitudinal displacement of 0.01 meters are obtained. Combined with the orientation angle, a standardized real-time offset vector is generated.

[0047] In this implementation, interference errors in the real-time offset vector are filtered out to obtain the corrected offset vector. Interference error filtering and correction are implemented using a Kalman filter algorithm. The process noise covariance matrix of the filter is set to 0.01, and the measurement noise covariance matrix is ​​set to 0.05. The real-time offset vector is smoothed and optimized through iterative calculations. During the correction process, the state estimation error is updated in real time. When the error converges to within 0.1 mm, the correction is considered effective. If the convergence condition is not met, the filter parameters are automatically adjusted, and the iterative calculation is re-executed. Before correction, jump data exceeding the physical limits of the equipment's operation are removed from the vector. The reasonable displacement range is set to ±0.05 meters every 10 milliseconds based on the maximum operating speed of the equipment. The final corrected offset vector completely retains the three-dimensional displacement components and orientation angle information, and the timing and fluctuation anomaly ranges are perfectly aligned, allowing it to be directly used for subsequent collision detection calculations. For example, performing Kalman filter iterative correction on the real-time offset vector filters out the 0.03-meter jump error caused by road bumps, resulting in a corrected offset vector with an error convergence to 0.08 mm.

[0048] In step S106, mapping the corrected offset vector to a preset transport space coordinate system and performing collision detection to obtain potential collision data includes: The correction offset vector is mapped to a preset transportation space coordinate system to obtain the vector projection direction and the equipment forward direction; Calculate the angle between the vector projection direction and the device's forward direction, and compare the angle with a preset angle threshold. If the angle exceeds the angle threshold, it is marked as a potential collision direction. Obtain the distance to the obstacle in the potential collision direction, and calculate the collision risk index based on the obstacle distance and the correction offset vector; If the collision risk index exceeds a preset risk threshold, the potential collision direction is recorded as potential collision data.

[0049] It should be noted that, firstly, the correction offset vector is mapped to a preset transport space coordinate system to obtain the vector projection direction and the equipment's forward direction. The preset transport space coordinate system uses the tunnel entrance as the origin, the tunnel extension direction as the longitudinal axis, the horizontal direction perpendicular to the tunnel wall as the transverse axis, and the direction perpendicular to the ground as the height axis. The spatial range of the coordinate system completely covers the entire transport tunnel and equipment operating area. Vector mapping is implemented using a spatial coordinate transformation algorithm, mapping the three-dimensional displacement components of the correction offset vector to the three coordinate axes of the coordinate system. The projection direction of the vector in the horizontal plane is obtained through vector space projection calculation. Simultaneously, the equipment's forward direction is determined by the tangent direction of the equipment's trajectory. Both the projection direction and the forward direction are quantified using angle values ​​in the coordinate system, with accuracy controlled within ±0.5 degrees.

[0050] For example, by mapping the corrected offset vector containing lateral and longitudinal displacement components to the roadway transport space coordinate system, the calculated vector projection direction is 15 degrees to the left laterally and the equipment forward direction is 0 degrees longitudinally.

[0051] Subsequently, the angle between the vector projection direction and the device's forward direction is calculated, and this angle is compared with a preset angle threshold. If the angle exceeds the threshold, it is marked as a potential collision direction. The angle is the absolute angle between the two directions in the horizontal plane, calculated in degrees, and the calculation accuracy is consistent with the direction angle accuracy.

[0052] The angle threshold is set based on historical collision accident data from the transportation of machine head components in tunnels over the past year. The minimum offset angle of all collision accidents is statistically analyzed, and the base threshold is set at 30 degrees. This can be lowered to 20 degrees for narrow tunnel transportation scenarios and raised to 40 degrees for wide tunnel conventional transportation scenarios. This threshold has been verified in multiple scenarios and can accurately identify offset directions with collision risk. The comparison process is performed moment-by-moment to ensure no risky directions are missed.

[0053] For example, if the calculated angle between the vector projection direction and the device's forward direction is 35 degrees, which exceeds the basic threshold of 30 degrees, then the direction that is 15 degrees to the left is marked as the potential collision direction.

[0054] Subsequently, the distance to obstacles in the potential collision direction is obtained, and the collision risk index is calculated based on the obstacle distance and the corrected offset vector. The obstacle distance is acquired in real time by a laser rangefinder in the corresponding direction, with the sampling frequency perfectly consistent with the spacing data sampling frequency, achieving a measurement accuracy of ±1 mm. The collision risk index is obtained by weighted summation of obstacle distance and offset rate, where the obstacle distance has a weight of 0.6, and the offset rate corresponding to the corrected offset vector has a weight of 0.4. This weighting combination is based on historical statistical data of lane collision risks. Obstacle distance directly determines the time window for collision occurrence and has a more direct impact on risk, thus receiving a higher weight. The offset rate characterizes how quickly the equipment approaches the obstacle and has a lower weight. Before calculation, all values ​​are mapped to the range of 0 to 1 using the min-max normalization method. The higher the final collision risk index value, the greater the probability of a collision.

[0055] For example, if the distance to an obstacle in the potential collision direction is 0.8 meters, and the offset rate of the correction offset vector is used to perform a weighted calculation, the collision risk index is obtained as 0.72.

[0056] If the collision risk index exceeds the preset risk threshold, the potential collision direction is recorded as potential collision data. The risk threshold is set based on statistical data of effective collision risk identification in roadway transportation over the past year. The lowest risk index of all successfully warned collision events is calculated, and the basic threshold is set to 0.6. For high-precision collision avoidance scenarios, this can be lowered to 0.5, and for regular transportation scenarios, it can be raised to 0.7. This threshold has been verified under multiple operating conditions and can reliably distinguish between safe and high-risk collision states. When the risk exceeds the limit, the synchronously recorded information includes the potential collision direction angle, obstacle distance, collision risk index, corresponding time, and equipment position coordinates. All data is arranged continuously in chronological order, forming complete potential collision data that can be directly used for subsequent risk level assessment calculations.

[0057] For example, if the collision risk index of 0.72 exceeds the basic threshold of 0.6, the system records complete information about the potential collision direction and generates corresponding potential collision data.

[0058] In step S107, the collision risk level is determined based on the potential collision data and pre-acquired transportation equipment operating parameters. If the collision risk level is higher than a preset risk assessment threshold, a warning instruction sequence is generated, including: Extract the pre-acquired equipment operating speed and safety distance values ​​to obtain the transportation equipment operating parameters; Calculate the collision assessment value based on the potential collision data and the operating parameters of the transportation equipment; The collision assessment values ​​are compared with preset grading intervals to determine the collision risk level; If the collision risk level is higher than the preset risk assessment threshold, a warning instruction sequence is generated.

[0059] It should be noted that, firstly, the pre-acquired equipment operating speed and safety clearance values ​​are extracted to obtain the operating parameters of the transport equipment. The equipment operating speed is collected in real time by an incremental encoder on the transport equipment's walking mechanism, with the collection frequency completely consistent with the sampling frequency of the clearance data to ensure timing synchronization. The safety clearance value is set based on the braking performance test data of the head assembly in tunnel transport over the past year and statistical analysis of equipment size parameters. The basic safety clearance is set at 1.5 meters, which can be increased to 2.0 meters for narrow tunnel transport scenarios and decreased to 1.0 meter for wide tunnel conventional transport scenarios. The extracted operating parameters also include two auxiliary parameters: equipment braking response time and steering adjustment accuracy. All parameter values ​​are mapped to the range of 0 to 1 using the minimum-maximum normalization method to maintain a consistent standard with the pre-processed data.

[0060] For example, the real-time operating speed of 1.2 meters per second and the basic safety distance of 1.5 meters are extracted from the transportation equipment control system, and the braking response time parameter is obtained synchronously. After normalization, standardized transportation equipment operating parameters are generated.

[0061] Subsequently, collision assessment values ​​are calculated based on potential collision data and transportation equipment operating parameters. These values ​​are derived using a multi-dimensional weighted fusion algorithm. The fusion weights are set based on historical collision warning accuracy data, with the collision risk index in the potential collision data assigned a weight of 0.4, equipment operating speed assigned a weight of 0.3, and the difference between obstacle distance and safe clearance assigned a weight of 0.3. This weighting is suitable for the collision risk assessment needs of roadway transportation. The collision risk index directly reflects the degree of deviation anomaly and has a stronger indicative role in risk prediction, thus receiving a higher weight. The difference between operating speed and clearance represents the time window for collision occurrence, with a relatively balanced impact, thus receiving a lower weight. During the calculation process, the difference between obstacle distance and safe clearance is first normalized, and then weighted and summed with the other two types of parameters. The final collision assessment value ranges from 0 to 1, with higher values ​​indicating a more severe collision risk.

[0062] For example, by combining a collision risk index of 0.72, an operating speed of 1.2 meters per second, and a distance difference of 0.7 meters to perform a weighted fusion calculation, a collision assessment value of 0.68 is obtained.

[0063] The collision risk level is determined by comparing the collision assessment value with a preset grading interval. The collision risk grading interval is set based on the severity statistics of roadway transport collision events over the past year. The basic grading interval is set as follows: 0 to 0.3 corresponds to low risk, 0.3 to 0.7 corresponds to medium risk, and 0.7 to 1.0 corresponds to high risk. For high-precision collision avoidance scenarios, the interval can be tightened, lowering the lower limit of medium risk to 0.25 and the lower limit of high risk to 0.65. For regular transport scenarios, the interval can be widened, raising the lower limit of medium risk to 0.35 and the lower limit of high risk to 0.75. The comparison process is performed hourly, matching the real-time calculated collision assessment value with the grading interval to directly determine the corresponding collision risk level. The grading results can intuitively reflect the severity and urgency of the collision event. For example, if the collision assessment value of 0.68 falls within the 0.3 to 0.7 interval, the collision risk level of the current transport status is determined to be medium risk.

[0064] If the collision risk level exceeds the preset risk assessment threshold, a warning instruction sequence is generated. The risk assessment threshold is statistically set based on the safety management requirements for the transportation of nose components over the past year. The basic threshold is set to medium risk level, which can be lowered to low risk level for high-precision nose component transportation scenarios and raised to high risk level for routine transportation scenarios. When the collision risk level exceeds the preset threshold, the system generates a warning instruction sequence according to the urgency level corresponding to the risk level. The instruction sequence includes four types of instructions: equipment deceleration ratio, steering fine-tuning angle, air valve opening adjustment, and audible and visual warning trigger. The instructions are prioritized from high to low risk level. High-risk levels trigger all emergency instructions, medium-risk levels trigger deceleration and steering fine-tuning instructions, and low-risk levels only trigger audible and visual warning instructions. The timing of the instruction sequence is perfectly matched with the equipment response delay parameters to ensure the synchronization and effectiveness of instruction execution. For example, if the collision risk level is medium risk and exceeds the preset low-risk assessment threshold, the system generates a warning instruction sequence including a 30% deceleration and a 5-degree steering fine-tuning.

[0065] In step S108, the attitude of the transport equipment and the distance between the equipment are adjusted according to the warning instruction sequence.

[0066] It should be noted that the warning command sequence is first analyzed to separate equipment attitude adjustment commands and equipment spacing adjustment commands. Simultaneously, the execution priority and parameter thresholds corresponding to these commands are extracted. Command parsing employs a timing alignment algorithm, using the unified clock of the equipment control system as a reference. The command sequence is split according to the order of execution, with execution priorities set based on the collision risk level. Emergency commands corresponding to high-risk levels have the highest priority, followed by adjustment commands for medium-risk levels, and warning commands for low-risk levels have the lowest priority. The transmission delay of command issuance is controlled within 50 milliseconds to ensure real-time command execution.

[0067] Subsequently, the operating posture of the transport equipment is adjusted according to the posture adjustment command. The adjustment includes three parameters: steering angle, vehicle level, and driving direction. The steering angle is adjusted through the equipment's steering actuator, with an adjustment accuracy controlled within ±0.5 degrees. The vehicle level is adjusted by the leveling cylinders of the support platform, with a level deviation controlled within ±0.2 degrees. The driving direction is corrected through the closed-loop control of the navigation system. The correction process matches the preset transport trajectory of the tunnel in real time. The adjustment step size is adaptively matched with the equipment's operating speed. The faster the speed, the smaller the adjustment step size, avoiding secondary deviations caused by posture adjustment.

[0068] For example, based on the instruction to fine-tune the steering by 5 degrees, the steering actuator is controlled to correct the equipment's travel direction by 5 degrees toward the center of the roadway, and the vehicle's levelness is adjusted simultaneously to ensure a smooth and shock-free attitude adjustment process.

[0069] The spacing between the transport equipment and the parallel transport equipment on the tunnel wall is adjusted according to the spacing adjustment command. The adjustment process uses real-time data collected by the laser rangefinder as feedback and adopts a proportional-integral control algorithm to achieve closed-loop regulation. The target value of the spacing adjustment is based on the preset safe spacing value, combined with dynamic correction according to the collision risk level. Under the high risk level, the target spacing is increased to 1.5 times the basic safe spacing; under the medium risk level, it is increased to 1.2 times; and under the low risk level, the basic safe spacing remains unchanged. During the adjustment process, the change of the spacing value is monitored in real time to ensure that the adjusted spacing is stable within the range of ±0.1 meters of the target value.

[0070] For example, according to the spacing adjustment instruction of 30% reduction, the equipment operating speed is reduced from 1.2 meters per second to 0.84 meters per second, and the target spacing with the left side of the tunnel wall is simultaneously increased from 1.5 meters to 1.8 meters, completing the closed-loop adjustment of the spacing.

[0071] During attitude and spacing adjustments, real-time data on equipment position, spacing, and vibration are collected to continuously monitor the adjustment effect and synchronously update the execution progress of the early warning command sequence. The stability threshold for the adjustment effect is set based on statistical data from stable equipment operation over the past year. The basic stability threshold is set as spacing fluctuation less than ±0.05 meters and attitude angle fluctuation less than ±0.2 degrees within 5 consecutive seconds. For high-precision transportation scenarios, this can be tightened to fluctuations less than ±0.03 meters and ±0.1 degrees within 10 consecutive seconds, while for conventional transportation scenarios, it can be relaxed to fluctuations less than ±0.1 meters and ±0.3 degrees within 3 consecutive seconds. If the monitored values ​​do not meet the stability threshold, a secondary correction command is automatically generated to fine-tune the equipment attitude and spacing until the stability requirements are met.

[0072] Once the device's posture and spacing meet the stability threshold, the adjustment is confirmed as complete. Simultaneously, the adjusted device operating status data is collected and archived along with the data from the early warning command sequence adjustment process. This data is used for subsequent training and optimization of the collision risk assessment model. The model training employs a support vector machine algorithm. The training set is constructed from valid data from historical adjustment processes and collision risk event data, with a minimum of 5000 data entries. A radial basis function kernel is used during training, with a learning rate of 0.01 and an maximum of 200 iterations. Training stops when the classification accuracy fluctuation is less than 0.5% for 10 consecutive iterations. The optimized model can further improve the accuracy of subsequent early warnings and adjustments.

[0073] For example, after confirming that the equipment posture and spacing have reached a stable standard, the entire process data of this early warning adjustment is archived to the local database and simultaneously input into the risk assessment model to complete iterative optimization.

[0074] In summary, this invention discloses a method for monitoring the transportation status of a nose section component based on multi-source sensor fusion. The method includes collecting spacing data, vibration signals, and real-time position data of the transportation equipment in the nose section component transportation scenario; fusing these data after time-series alignment and anomaly cleaning to obtain an original dataset; obtaining a smoothed spacing dataset through iterative optimization of state prediction and error correction; extracting time-frequency joint features to calculate time-series fluctuation values ​​and identifying persistent fluctuation anomaly intervals; correcting the equipment's real-time offset vector and mapping it to the transportation space coordinate system to complete collision detection; and combining equipment operating parameters to grade and assess collision risk. When a threshold is exceeded, a warning command sequence is generated, and the attitude and spacing of the transportation equipment are dynamically adjusted. This invention solves the problems of traditional single-sensor monitoring being susceptible to environmental interference, lagging anomaly identification, and insufficient accuracy in collision risk prediction. It achieves comprehensive, accurate, real-time monitoring and proactive risk prevention of the nose section component transportation status, effectively ensuring the safety and stability of rail transit equipment transportation.

[0075] Reference Figure 2 The second embodiment of the present invention provides a nose assembly transportation status monitoring system based on multi-source sensor fusion, comprising: The data acquisition module is used to acquire spacing data, vibration signals, and real-time location data of the transportation equipment in the transportation scenario of the nose assembly. The data fusion module is used to fuse the spacing data and the vibration signal to obtain the original dataset; The data optimization module is used to calculate the predicted spacing state value and the actual spacing measurement value based on the original dataset, and to perform iterative optimization processing on the original dataset based on the predicted spacing state value and the actual spacing measurement value to obtain a smoothed spacing dataset. The interval identification module is used to extract the dynamic change features of the smooth interval dataset and the amplitude-frequency change features of the vibration signal, calculate the time-series fluctuation value based on the dynamic change features and the amplitude-frequency change features, filter the data intervals where the time-series fluctuation value exceeds the preset fluctuation judgment threshold, and obtain the fluctuation abnormal interval. The vector processing module is used to extract a real-time offset vector from the real-time location data if the length of the fluctuation abnormal interval exceeds a preset interval abnormal threshold, and to perform error correction on the real-time offset vector to obtain a corrected offset vector. The collision detection module is used to map the correction offset vector to a preset transportation space coordinate system and perform collision detection to obtain potential collision data. The risk assessment module is used to determine the collision risk level based on the potential collision data and the pre-acquired operating parameters of the transportation equipment. If the collision risk level is higher than the preset risk judgment threshold, a warning instruction sequence is generated. The calibration and control module is used to adjust the attitude and spacing of the transport equipment according to the warning command sequence.

[0076] It should be noted that the multi-source sensor fusion-based nose assembly transportation status monitoring system provided in this embodiment of the invention is used to execute all the process steps of the multi-source sensor fusion-based nose assembly transportation status monitoring method in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0077] 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.

[0078] 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 method for monitoring the transport status of a nose section component based on multi-source sensor fusion, characterized in that, include: Acquire spacing data, vibration signals, and real-time location data of the transportation equipment in the transportation scenario of the nose assembly; By fusing the spacing data with the vibration signal, the original dataset is obtained; The predicted spacing state and the actual spacing measurement value are calculated based on the original dataset. The original dataset is then iteratively optimized based on the predicted spacing state and the actual spacing measurement value to obtain a smoothed spacing dataset. Extract the dynamic change features of the smooth spacing dataset and the amplitude-frequency change features of the vibration signal, calculate the time-series fluctuation value based on the dynamic change features and the amplitude-frequency change features, and filter the data range where the time-series fluctuation value exceeds the preset fluctuation judgment threshold to obtain the fluctuation abnormal range; If the length of the fluctuation abnormal interval exceeds the preset interval abnormal threshold, then the real-time offset vector is extracted from the real-time location data, and the real-time offset vector is error-corrected to obtain the corrected offset vector. The corrected offset vector is mapped to a preset transportation space coordinate system and collision detection is performed to obtain potential collision data; Based on the potential collision data and the pre-acquired operating parameters of the transportation equipment, the collision risk level is determined. If the collision risk level is higher than the preset risk judgment threshold, a warning instruction sequence is generated. Adjust the posture and spacing of the transport equipment according to the sequence of warning instructions.

2. The method for monitoring the transportation status of a nose assembly based on multi-source sensor fusion according to claim 1, characterized in that, The acquisition of spacing data, vibration signals, and real-time position data of the transportation equipment in the nose assembly transportation scenario includes: Collect real-time distance data between the transport equipment and the tunnel wall to obtain spacing data; The vibration amplitude values ​​during the operation of the transportation equipment are collected to obtain vibration signals; Collect the spatial coordinate values ​​of the transportation equipment to obtain the real-time location data of the transportation equipment.

3. The method for monitoring the transportation status of a nose assembly based on multi-source sensor fusion according to claim 1, characterized in that, The fusion of the spacing data and the vibration signal yields the original dataset, which includes: Align the spacing data with the timing nodes of the vibration signal to obtain timing synchronization data; Data exceeding a preset amplitude threshold is removed from the timing synchronization data to obtain cleaned data; The cleaned data is integrated to generate the original dataset.

4. The method for monitoring the transportation status of a nose assembly based on multi-source sensor fusion according to claim 1, characterized in that, The step of calculating the predicted spacing state value and the actual spacing measurement value based on the original dataset, and iteratively optimizing the original dataset based on the predicted spacing state value and the actual spacing measurement value to obtain a smoothed spacing dataset includes: Retrieve pre-acquired historical spacing data, extract the average spacing value and spacing change rate of the historical spacing data, and construct spacing status parameters; Perform a time-series recursive calculation based on the spacing state parameters to generate a spacing state prediction value; Extract the spacing values ​​from the original dataset to determine the actual spacing measurements; Calculate the error covariance between the predicted spacing state value and the actual spacing measurement value; Based on the error covariance, a gain coefficient is calculated by combining the percentage of the numerical deviation between the predicted spacing state value and the measured spacing value, and an error correction parameter is generated based on the gain coefficient. The spacing acquisition values ​​are corrected according to the error correction parameters to obtain corrected spacing data; Calculate the numerical deviation between the predicted spacing state value and the corrected spacing data to obtain the iterative deviation value; The iteration deviation value is optimized until it is lower than a preset error judgment threshold to obtain a smooth spacing dataset.

5. The method for monitoring the transportation status of a nose assembly based on multi-source sensor fusion according to claim 1, characterized in that, The process involves extracting the dynamic change features of the smoothed spacing dataset and the amplitude-frequency change features of the vibration signal, calculating the time-series fluctuation value based on the dynamic change features and the amplitude-frequency change features, and filtering out data intervals where the time-series fluctuation value exceeds a preset fluctuation judgment threshold to obtain the fluctuation abnormal interval, including: The amplitude and frequency variation features of the smooth interval dataset are extracted from the time series to obtain the dynamic variation features; Extract the amplitude-frequency variation characteristics of the vibration signal; Calculate time-series statistical values ​​based on the amplitude-frequency variation characteristics and the dynamic variation characteristics, and determine the short-term variation trend of the data based on the time-series statistical values; Based on the short-term change trend, calculate the time-series fluctuation value of the smooth interval dataset, and filter out sequence segments whose time-series fluctuation values ​​exceed the preset fluctuation judgment threshold. Locate the time interval corresponding to the sequence segment and perform interval marking to obtain the fluctuation anomaly interval.

6. The method for monitoring the transportation status of a nose assembly based on multi-source sensor fusion according to claim 1, characterized in that, If the length of the abnormal fluctuation range exceeds a preset abnormal range threshold, a real-time offset vector is extracted from the real-time location data, and error correction is performed on the real-time offset vector to obtain a corrected offset vector, including: Calculate the continuous displacement difference of the real-time position data to generate a real-time offset vector; The interference error in the real-time offset vector is filtered out to obtain the corrected offset vector.

7. The method for monitoring the transportation status of a nose assembly based on multi-source sensor fusion according to claim 1, characterized in that, The step of mapping the corrected offset vector to a preset transportation space coordinate system and performing collision detection to obtain potential collision data includes: The correction offset vector is mapped to a preset transportation space coordinate system to obtain the vector projection direction and the equipment forward direction; Calculate the angle between the vector projection direction and the device's forward direction, and compare the angle with a preset angle threshold. If the angle exceeds the angle threshold, it is marked as a potential collision direction. Obtain the distance to the obstacle in the potential collision direction, and calculate the collision risk index based on the obstacle distance and the correction offset vector; If the collision risk index exceeds a preset risk threshold, the potential collision direction is recorded as potential collision data.

8. The method for monitoring the transportation status of a nose assembly based on multi-source sensor fusion according to claim 1, characterized in that, The collision risk level is determined based on the potential collision data and pre-acquired transportation equipment operating parameters. If the collision risk level is higher than a preset risk assessment threshold, a warning instruction sequence is generated, including: Extract the pre-acquired equipment operating speed and safety distance values ​​to obtain the transportation equipment operating parameters; Calculate the collision assessment value based on the potential collision data and the operating parameters of the transportation equipment; The collision assessment values ​​are compared with preset grading intervals to determine the collision risk level; If the collision risk level is higher than the preset risk assessment threshold, a warning instruction sequence is generated.

9. A nose section component transportation status monitoring system based on multi-source sensor fusion, characterized in that, include: The data acquisition module is used to acquire spacing data, vibration signals, and real-time location data of the transportation equipment in the transportation scenario of the nose assembly. The data fusion module is used to fuse the spacing data and the vibration signal to obtain the original dataset; The data optimization module is used to calculate the predicted spacing state value and the actual spacing measurement value based on the original dataset, and to perform iterative optimization processing on the original dataset based on the predicted spacing state value and the actual spacing measurement value to obtain a smoothed spacing dataset. The interval identification module is used to extract the dynamic change features of the smooth interval dataset and the amplitude-frequency change features of the vibration signal, calculate the time-series fluctuation value based on the dynamic change features and the amplitude-frequency change features, filter the data intervals where the time-series fluctuation value exceeds the preset fluctuation judgment threshold, and obtain the fluctuation abnormal interval. The vector processing module is used to extract a real-time offset vector from the real-time location data if the length of the fluctuation abnormal interval exceeds a preset interval abnormal threshold, and to perform error correction on the real-time offset vector to obtain a corrected offset vector. The collision detection module is used to map the correction offset vector to a preset transportation space coordinate system and perform collision detection to obtain potential collision data. The risk assessment module is used to determine the collision risk level based on the potential collision data and the pre-acquired operating parameters of the transportation equipment. If the collision risk level is higher than the preset risk judgment threshold, a warning instruction sequence is generated. The calibration and control module is used to adjust the attitude and spacing of the transport equipment according to the warning command sequence.

Images