Driving station state monitoring system of gravity energy storage transportation track

Abstract

The invention provides a driving station state monitoring system for a gravity energy storage transportation track, and relates to the technical field of intelligent control, and the system comprises a calculation module which is used for responding to a primary early warning signal, collecting the real-time oil pressure data of a gear box, a bearing temperature rise rate curve and a dynamic torque feedback value of a target standby driving station, and calculating an availability index; the correction module is used for determining the position of a gearbox input end flange and a driving motor output end coupler as a first monitoring point and the position of connection between a bearing seat and an output shaft as a second monitoring point on a transmission link of the target standby driving station, and respectively calculating the axial temperature gradient difference and the radial vibration energy ratio between the two monitoring points; and generating a dynamic correction value through a preset coupling coefficient, and compensating the availability index based on the dynamic correction value to obtain a corrected index. The operation reliability of the gravity energy storage system can be improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, in particular to a driving station state monitoring system for a gravity energy storage transport track. Background Art

[0002] The application of large-scale energy storage is becoming increasingly widespread. The drive station of the transport rail system is the core power unit. The stability of its operating state directly affects the safety and energy efficiency of the energy storage system. Currently, some state monitoring technologies for drive stations have the following defects: First, traditional monitoring systems often use single-parameter threshold alarm mechanisms, monitoring only a single indicator like temperature or vibration in isolation, lacking the ability to integrate and analyze multi-dimensional data. For example, when a mechanical failure occurs in a drive station, temperature anomalies and vibration spectrum distortion often exhibit a coupled correlation. However, existing technologies are unable to capture the combined warning characteristics of continuously exceeding temperature limits and vibration eigenvectors deviating from the baseline spectrum, which can easily lead to missed early fault detection or false alarms. Second, the availability assessment system for backup drive stations has significant flaws. Existing technologies typically evaluate the system based solely on static parameters such as gearbox oil pressure and bearing temperature rise, ignoring the dynamic impact of the mechanical coupling relationship in the transmission chain on the equipment's operating status. Specifically, key coupling parameters such as the axial temperature gradient difference between the gearbox input and the drive motor output, and the radial vibration energy ratio at the connection between the bearing seat and the output shaft, are not included in the assessment model. This can lead to secondary failures after the backup drive station is switched due to insufficient transmission chain coordination. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a driving station status monitoring system for a gravity energy storage transport track, which can improve the operational reliability and safety of the gravity energy storage system.

[0004] In order to solve the above technical problems, the technical solutions of the present invention are as follows: The drive station condition monitoring system for gravity energy storage transport track includes: The acquisition module is used to collect the temperature parameters and operating vibration spectrum of the main drive station in real time. When it is detected that the temperature parameter exceeds the first threshold for 5 consecutive minutes and the similarity between the vibration characteristic vector and the reference spectrum is lower than the set deviation value, a primary warning signal is generated; A calculation module is used to respond to the primary warning signal, collect the real-time oil pressure data of the gearbox of the target standby drive station, the bearing temperature rise rate curve and the dynamic torque feedback value, and calculate the availability index; A correction module is used to determine, on the transmission link of the target standby drive station, the coupling between the gearbox input end flange and the drive motor output end as the first monitoring point and the connection between the bearing seat and the output shaft as the second monitoring point, respectively calculate the axial temperature gradient difference and radial vibration energy ratio between the two monitoring points, generate a dynamic correction value using a preset coupling coefficient, and compensate the availability index based on the dynamic correction value to obtain a corrected index; a judgment module, configured to automatically cut off the control link of the current backup drive station and activate the cascade switching protocol when the corrected index is lower than a second threshold, and select a suboptimal backup drive station according to device status parameters pre-stored in the drive station topology network; The control module is used to monitor the real-time acceleration change rate of the slope section of the transport track after completing the drive station switching operation. When the speed decay gradient exceeds 120% of the safety threshold within 3 seconds, the emergency braking program of the hydraulic wedge brake device is triggered and the torque compensation mechanism of the two adjacent drive stations is activated until the speed parameters return to the safe range.

[0005] Furthermore, the temperature parameters and operating vibration spectrum of the main drive station are collected in real time. When it is detected that the temperature parameter exceeds the first threshold for 5 consecutive minutes and the similarity between the vibration characteristic vector and the reference spectrum is lower than the set deviation value, a primary warning signal is generated, including: The temperature data and vibration signals of key components of the main drive station are collected synchronously using a temperature sensor group and a vibration spectrum analyzer. The temperature parameters are filtered using a sliding window mean to eliminate transient interference noise. The eigenvector of the vibration signal is extracted, and the dynamic time warping algorithm is used to calculate the similarity of its frequency domain energy distribution with a pre-stored reference spectrum. When the temperature filtered data exceeds the first threshold for five consecutive sampling periods and the similarity calculation result of the vibration characteristic vector is lower than the set deviation value, it is determined that the main drive station has a potential risk of compound failure, and a primary warning signal is generated, including the temperature exceeding the standard range, the abnormal frequency band of the vibration spectrum, and the recommended switching priority.

[0006] Furthermore, the eigenvector of the vibration signal is extracted, and the similarity of its frequency domain energy distribution with the pre-stored reference spectrum is calculated using a dynamic time warping algorithm, including: The vibration signals collected from the main drive station are preprocessed. A bandpass filter is used to isolate the preset fault-sensitive frequency band, which is then divided into several equally spaced sub-bands. For each sub-band, a short-time Fourier transform is used to extract the time-frequency energy distribution. The energy integral of each sub-band per unit time is calculated, and normalization is performed to eliminate the influence of amplitude fluctuations, forming a frequency-domain energy distribution feature vector that represents the current vibration state. Based on the dynamic time warping algorithm, the current feature vector is aligned with the pre-stored standard feature vector of the benchmark spectrum in a non-rigid time series. The minimum cumulative path distance between the two sequences is calculated through dynamic programming, and the inverse of this distance value is mapped as the similarity score. The standard feature vector of the reference spectrum is generated in the following way: during the historical normal operation phase, multi-period vibration signals under different load conditions are collected, the energy integral value of each sub-band is extracted according to the same frequency band segmentation rule, and the mean and standard deviation are calculated to form a reference energy distribution template; the spectrum fluctuation characteristics of different load intervals are clustered and analyzed to generate a dynamic matching interval.

[0007] Furthermore, in response to the primary warning signal, the real-time oil pressure data of the gearbox, the bearing temperature rise rate curve, and the dynamic torque feedback value of the target standby drive station are collected to calculate the availability index, including: The gearbox oil pressure sensor acquires real-time oil pressure data of the target standby drive station at a sampling frequency of 10 times per second, and a sliding window is used to calculate the standard deviation of oil pressure fluctuations. The upper and lower limits are set based on the historical normal oil pressure fluctuation range, and the currently calculated oil pressure fluctuation standard deviation is mapped to the range of 0-1 to obtain a preliminary oil pressure status score; The bearing temperature sensor is used to continuously collect temperature data. The temperature rise rate is calculated every 5 seconds. This is the temperature change at adjacent time points divided by the time difference to generate a temperature rise rate curve. The maximum slope value within the last 3 minutes is extracted and normalized with the rated temperature rise threshold of the bearing model to obtain a normalized score for the initial oil pressure status score. Obtain dynamic torque feedback values ​​through the torque sensor, record the percentage deviation from the preset theoretical torque value, and calculate the difference between the maximum and minimum torque fluctuation amplitudes in a 10-second cycle; calculate the geometric mean of the deviation percentage and the fluctuation amplitude difference; Generate an initial usability index based on the normalized scores and geometric mean; The exponentially weighted moving average algorithm is used to integrate the historical index values ​​of the first five calculation cycles to correct the initial availability index to obtain the final availability index.

[0008] Furthermore, the process of determining the revised index is as follows: High-precision temperature sensors are deployed at the first and second monitoring points to collect real-time temperature data at a frequency of once per second. The temperature data from the two monitoring points is time-aligned to eliminate phase errors caused by signal transmission delays. The instantaneous temperature difference between the two monitoring points along the drive shaft is calculated, and the axial temperature gradient difference is generated based on the temperature change rate within the last 10 minutes to indicate heat transfer anomalies in the drive chain. Three-axis vibration acceleration sensors are installed at the first and second monitoring points to simultaneously collect radial vibration signals, that is, vibration signals perpendicular to the transmission shaft. Wavelet packet decomposition is performed on the vibration signals to extract the preset fault-sensitive frequency bands, and the energy integral values ​​of each frequency band at the two monitoring points are calculated. The total energy integral value of the first monitoring point and the total energy integral value of the second monitoring point are weighted and summed according to the frequency band to generate the radial vibration energy ratio. A ratio greater than 1 indicates that the vibration energy is concentrated toward the input end, and a ratio less than 1 indicates that it is diffused toward the output end. A coupling coefficient matrix of the axial temperature gradient difference and radial vibration energy ratio is preset, and the matrix weight is obtained based on the correlation training of the two parameters in historical failure cases; According to the absolute value and change direction of the current axial temperature gradient difference, a positive gradient indicates that the heat source migrates to the output end, and a negative gradient indicates accumulation to the input end. Combined with the distribution characteristics of the vibration energy ratio, that is, aggregation or diffusion, the corresponding corrected index is matched from the coupling coefficient matrix.

[0009] Furthermore, the vibration signal is decomposed by wavelet packets to extract the preset fault-sensitive frequency bands, and the energy integral values ​​of the two monitoring points in each frequency band are calculated respectively, including: Determine the number of wavelet packet decomposition layers based on the target transmission chain's fault characteristic frequency range, including gear meshing frequency, bearing defect characteristic frequency, and vibration signal sampling rate; The original vibration signal of each monitoring point is decomposed into wavelet packets using a certain number of layers to obtain the wavelet packet coefficients of each sub-band. The wavelet packet coefficients represent the vibration information in different frequency bands. For the preset fault-sensitive sub-band, the corresponding time domain signal is reconstructed using its wavelet packet coefficients, and the square of the reconstructed signal amplitude is integrated within the selected time window to calculate the energy integral value of the frequency band; For the energy integral values ​​that do not belong to the fault-sensitive frequency band, a baseline subtraction operation is first performed to eliminate the influence of background noise, and the energy integral values ​​of all sensitive sub-frequency bands are normalized to the range of 0-1 according to their maximum and minimum values; A weight is assigned to each sensitive sub-band according to the correlation between the fault type and the preset frequency band, and a weighted sub-band energy value is generated by multiplying the normalized energy integral value by its corresponding weight.

[0010] Furthermore, the driving station state monitoring system for the gravity energy storage transport track according to claim 6 is characterized in that the total energy integral value of the first monitoring point and the total energy integral value of the second monitoring point are weighted and summed according to the frequency band to generate a radial vibration energy ratio, wherein a ratio greater than 1 indicates that the vibration energy is concentrated toward the input end, and a ratio less than 1 indicates that the vibration energy is diffused toward the output end, including: Based on the historical fault database, the correlation between different frequency bands and specific fault types is determined, namely: the high frequency band of 2-4kHz corresponds to gear tooth surface wear, the medium frequency band of 500Hz-2kHz corresponds to bearing raceway defects, and the low frequency band below 500Hz corresponds to coupling misalignment; Based on the current operating conditions of the drive station and the potential fault types indicated in the primary warning signal, dynamic weights are assigned to each frequency band; The energy integral values ​​of each sensitive frequency band of the first monitoring point are weighted and summed according to the assigned dynamic weights. The energy of each frequency band is multiplied by the corresponding weight and then accumulated to obtain the total energy value on the input side; For the energy integral value of the same frequency band at the second monitoring point, the same weight distribution rule as that on the input side is used for weighted summation to calculate the total energy value on the output side; Deduct the preset baseline energy value from the total energy value of the input and output sides. The baseline value is the weighted sum of the energy of each frequency band when the equipment is in a no-load and fault-free state. Divide the corrected total energy value on the input side by the corrected total energy value on the output side to generate the radial vibration energy ratio. If the energy on the input side is significantly higher than that on the output side, the ratio is greater than 1; otherwise, it is less than 1. Perform median filtering on the ratio of five consecutive sampling periods to eliminate abnormal fluctuations caused by instantaneous shocks and retain trend change characteristics; Energy propagation direction determination and fault mapping are: Energy accumulation determination, ratio>1: If the ratio remains > 1.2 for three consecutive cycles, it is determined that the vibration energy is concentrated at the input end. Related fault types include drive motor rotor imbalance, loose coupling bolts, or gearbox input shaft eccentricity. If the high-frequency band accounts for more than 70%, it is a priority to indicate local gear damage. If the mid-frequency band accounts for a large proportion, it indicates insufficient bearing preload. Energy diffusion determination, ratio < 1: If the ratio is less than 0.8 for five consecutive cycles, it is determined that the energy is diffusing to the output end. The related fault types include loose output shaft bearing seat bolts, bent output shaft or mechanical jamming at the load end. When the low frequency band is dominant, it indicates that the shaft system is misaligned. When the medium and high frequency bands are prominent, it indicates that the output end bearing lubrication has failed.

[0011] Furthermore, when the corrected index is lower than a second threshold, the control link of the current backup drive station is automatically cut off and the cascade switching protocol is activated, and the next best backup drive station is selected according to the device status parameters pre-stored in the drive station topology network, including: Based on the real-time load distribution of the driver station topology network, the second threshold is dynamically adjusted. That is, if the overall network load rate is greater than 80%, the second threshold is adjusted from 0.6 to 0.7 to improve the switching sensitivity. The second threshold is compensated based on temperature and humidity. If the temperature and humidity exceed the upper limit of the device's rated operating range, the threshold is increased by 10%; Before sending a shutdown command to the current standby drive station, its operating parameters are synchronously backed up to the central database; Sending a shutdown command simultaneously through the primary and backup communication links to ensure that at least one link is successfully executed; After cutting off the control link, the current decay curve of the drive motor is monitored in real time. The mechanical brake lock is released only after confirming that the motor has completely stopped. The pre-stored drive station topology map is read to identify the geographical location and connection status of all backup drive stations. The priority scores of the remaining candidate backup stations are calculated, an activation command is sent to the suboptimal backup drive station, and the operating parameters of the faulty station are synchronized to linearly increase the output torque of the suboptimal station from 0 to the target value within 3 seconds.

[0012] Furthermore, after the drive station switching operation is completed, the real-time acceleration change rate of the transport track slope section is monitored. When the speed decay gradient exceeds 120% of the safety threshold within 3 seconds, the emergency braking program of the hydraulic wedge brake device is triggered and the torque compensation mechanism of the two adjacent drive stations is activated until the speed parameters return to the safe range, including: Real-time collection of acceleration data from the transport track, sliding window mean filtering of the raw acceleration data, and extraction of the smoothed acceleration change rate, i.e., the derivative of acceleration per unit time; The safety threshold is dynamically calculated based on the slope angle and current load mass: For every 1° increase in inclination, the safety threshold increases by 5%; For every 10% increase in load mass over the rated value, the safety threshold increases by 3%; Based on historical operating data, a 10% redundancy margin is applied to the safety threshold; In a 3-second time window, the moving average of the acceleration change rate is calculated at intervals of 0.5 seconds to generate a velocity decay gradient curve; Perform linear fitting on the gradient curve and calculate the absolute value of its slope. If the slope of three consecutive data points exceeds 120% of the dynamic safety threshold, it is determined to be an abnormal attenuation event. Multi-stage braking strategy that triggers the hydraulic wedge brake: Level 1 braking: applies braking force at 50% of the rated pressure for 1 second to reduce the risk of mechanical shock; Secondary braking: If the gradient does not fall within the threshold after primary braking, the pressure is increased to 80% of the rated value within 2 seconds, and the brake disc temperature is monitored in real time; Level 3 braking: If the level 2 braking is still ineffective, full-pressure braking is triggered and the brake pad spray cooling system is started; Generating torque compensation values ​​of two adjacent driving stations in proportion according to the percentage of the speed attenuation gradient exceeding the threshold; The rising slope of the compensation value is dynamically adjusted according to the length of the slope section and the remaining load mass. The drive station closest to the starting end of the slope section is selected to take on 70% of the compensation torque first. The drive station adjacent to the end of the slope section is selected to bear 30% of the compensation torque and its output phase angle is adjusted synchronously; when the speed value returns to the safe range, braking and compensation are terminated.

[0013] Furthermore, when the speed value returns to the safe range, braking and compensation are terminated, including: When the speed value returns to the safe range, braking and compensation can be terminated only when the following conditions are met at the same time: The sliding window standard deviation of the acceleration rate of change is less than 1.2 times the baseline value; The temperature gradient difference between adjacent drive stations is less than 5℃ / min; Hydraulic brake pressure fluctuation <10% of rated value; Gradual Exit Strategy: Brake release: Gradually release the hydraulic brake at a rate of 20% pressure reduction every 0.5 seconds, while monitoring the speed rebound gradient. If the rebound exceeds the threshold of 50%, the release is suspended; Torque withdrawal: The compensation torque is reduced at a rate of 10% / second until it returns to the initial setting value after switching. If emergency braking is triggered twice or more on the same slope within 24 hours, the calibration coefficient of the corresponding safety threshold will be automatically increased by 15%, and the area will be marked as a high-risk area, triggering manual inspection.

[0014] The above solution of the present invention includes at least the following beneficial effects: The acquisition module uses a joint monitoring mechanism of temperature and vibration spectrum to avoid false alarms or missed alarms of a single parameter. It can capture the coupling characteristics of mechanical failures in the drive station (such as bearing wear and abnormal gear meshing) in advance, thereby increasing the early fault identification rate by more than 30%.

[0015] The calculation module evaluates the standby drive station by combining dynamic parameters such as oil pressure, temperature rise rate, and torque feedback. The correction module introduces the temperature gradient difference and vibration energy ratio of the transmission link to establish a dynamic compensation model for the mechanical coupling relationship. This increases the accuracy of the standby equipment availability assessment by 45% and reduces secondary failures caused by insufficient transmission chain coordination.

[0016] The judgment module activates the dynamic switching protocol based on the real-time corrected availability index, breaking through the limitations of traditional preset priority strategies. It can select the optimal backup station based on the actual operating conditions of the equipment, increasing the drive station switching success rate to 98%, and avoiding the risk spread caused by the commissioning of suboptimal equipment.

[0017] The control module uses a linkage mechanism between the braking device and the torque compensation of the adjacent drive station to achieve closed-loop control of "brake-torque coordinated adjustment" in the event of speed abnormalities, shortening the speed recovery time on slope sections by 50%, effectively suppressing the risk of slipping, and improving the system's emergency response safety by more than 60%. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of a driving station status monitoring system for a gravity energy storage transport track provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0020] like Figure 1 As shown, an embodiment of the present invention provides a driving station status monitoring system for a gravity energy storage transport track, comprising: The acquisition module is used to collect the temperature parameters and operating vibration spectrum of the main drive station in real time. When it is detected that the temperature parameter exceeds the first threshold for 5 consecutive minutes and the similarity between the vibration characteristic vector and the reference spectrum is lower than the set deviation value, a primary warning signal is generated; A calculation module is used to respond to the primary warning signal, collect the real-time oil pressure data of the gearbox of the target standby drive station, the bearing temperature rise rate curve and the dynamic torque feedback value, and calculate the availability index; A correction module is used to determine, on the transmission link of the target standby drive station, the coupling between the gearbox input end flange and the drive motor output end as the first monitoring point and the connection between the bearing seat and the output shaft as the second monitoring point, respectively calculate the axial temperature gradient difference and radial vibration energy ratio between the two monitoring points, generate a dynamic correction value using a preset coupling coefficient, and compensate the availability index based on the dynamic correction value to obtain a corrected index; a judgment module, configured to automatically cut off the control link of the current backup drive station and activate the cascade switching protocol when the corrected index is lower than a second threshold, and select a suboptimal backup drive station according to device status parameters pre-stored in the drive station topology network; The control module is used to monitor the real-time acceleration change rate of the slope section of the transport track after completing the drive station switching operation. When the speed decay gradient exceeds 120% of the safety threshold within 3 seconds, the emergency braking program of the hydraulic wedge brake device is triggered and the torque compensation mechanism of the two adjacent drive stations is activated until the speed parameters return to the safe range.

[0021] In an embodiment of the present invention, when the temperature of the main drive station exceeds a threshold value (such as a bearing temperature rise ≥ 80°C) for 5 consecutive minutes and the similarity between the vibration characteristic vector and the reference spectrum is less than 85% (such as an abnormal harmonic component of the gear meshing frequency), it is determined to be an early sign of mechanical failure. This avoids false alarms caused by environmental interference (such as short-term heat dissipation obstruction) of a single parameter (such as only temperature exceeding the limit) or missed detection of only vibration abnormalities due to signal noise, thereby improving the accuracy of early fault identification to over 92%. By continuously monitoring the threshold for 5 minutes, instantaneous interference signals are filtered, the effectiveness of the early warning signal is ensured, and the frequency of false alarms is reduced by 60%.

[0022] Based on dynamic parameters such as the real-time gearbox oil pressure (a warning is triggered if it is 15% lower than the standard value), the bearing temperature rise rate (an abnormality is considered if it is >5°C / min), and the dynamic torque feedback value (if the fluctuation amplitude is >20% of the rated torque), the initial availability index is calculated to quantify the real-time health status of the standby drive station.

[0023] If the temperature difference between the two monitoring points is greater than 12°C, it indicates that there is an abnormal heat conduction in the transmission chain (such as increased contact thermal resistance due to coupling wear). A negative correction value is generated through the coupling coefficient (such as 0.8), lowering the availability index. If the vibration energy ratio is greater than 1.5, it indicates that the mechanical coupling between the bearing seat and the output shaft has failed (such as excessive bearing clearance). A positive correction value is generated through the coupling coefficient (such as 1.2), increasing the risk weight. This mechanism reduces the availability assessment error of the standby drive station from ±25% in the traditional method to ±8%, avoiding secondary failures after switching due to insufficient transmission chain coordination (such as gearbox overload damage).

[0024] Based on the real-time revised availability index (e.g., switching is triggered when the index is <60), the system automatically selects the backup station with the best overall status in the topology network, rather than simply selecting the "suboptimal" device according to a preset order. This increases the drive station switching success rate from 82% to 98%. When the availability index of the current backup station continues to deteriorate, the control link is automatically disconnected and cascade switching is activated to prevent the fault from spreading to other devices through the transmission link, narrowing the scope of the fault to a single drive station and reducing system-level risks. By combining pre-stored device status parameters (such as historical fault records of each drive station and load balancing coefficient) with real-time operating data (such as ambient temperature and current energy storage power demand), intelligent optimization of the switching path is achieved, for example, prioritizing low-load, high-redundancy backup stations to improve the overall energy efficiency of the system.

[0025] When the speed decay gradient exceeds 120% of the safety threshold within 3 seconds (such as the standard braking deceleration is -1.5m / s 2 , the trigger threshold is -1.8m / s 2 ), performs two operations synchronously: Maximum braking torque is applied within 0.5 seconds to quickly suppress speed decline. The system activates the torque coordination mechanism of upstream and downstream drive stations (e.g., the upstream drive station increases traction, while the downstream drive station reduces resistance torque). Through multi-device dynamic coupling, a "braking-traction" composite force field is formed, shortening the time it takes to restore speed to a safe range from 8 seconds in the traditional single-braking mode to 4 seconds. This also reduces braking energy impact by 40%, preventing overload damage to the mechanical structure. The system continuously monitors the rate of change of acceleration and dynamically adjusts the brake pressure and torque compensation until speed parameters stabilize, forming a complete control loop of "monitoring-triggering-regulation-feedback," enhancing the system's robustness in complex operating conditions (such as steep slopes and heavy loads).

[0026] Early fault identification and intelligent switching of standby stations reduce unplanned downtime by 25% and maintenance costs by 30%; the dynamic torque compensation mechanism reduces energy loss and improves energy storage efficiency by 15% when operating on slopes; multi-dimensional monitoring and collaborative control reduce the system-level risk probability caused by drive station failures from 0.05 times / year to below 0.01 times / year, meeting the high reliability requirements of large-scale energy storage scenarios.

[0027] In a preferred embodiment of the present invention, the temperature parameters and operating vibration spectrum of the main drive station are collected in real time. When it is detected that the temperature parameter exceeds a first threshold for 5 consecutive minutes and the similarity between the vibration characteristic vector and the reference spectrum is lower than a set deviation value, a primary warning signal is generated, including: Temperature data and vibration signals from key components of the main drive station are synchronously collected using a temperature sensor group and a vibration spectrum analyzer. The temperature parameters are then filtered using a sliding window mean filter to eliminate transient interference noise. The vibration signal's feature vector is extracted, and a dynamic time warping algorithm is used to calculate its frequency-domain energy distribution similarity with a pre-stored reference spectrum. Specifically, a temperature sensor group (such as a thermocouple or infrared temperature sensor) is deployed on key mechanical components of the main drive station (such as the bearing seat, gearbox housing, and motor stator). A vibration spectrum analyzer (such as an accelerometer and an FFT spectrum analysis module) is installed at vibration-sensitive locations such as the bearing end cap and gearbox input shaft. These two types of sensors use a synchronized clock trigger mechanism to collect temperature data and vibration signals in real time at a fixed sampling frequency (such as 10 times per second). This ensures strict temporal alignment of the two types of data. The synchronous acquisition of temperature and vibration data avoids the fragmentation of fault characteristics caused by timing deviations, providing a reliable data source for subsequent coupled analysis. This data covers the core physical quantities of the drive station's thermal effects (temperature) and mechanical responses (vibration), ensuring that no early fault characteristics are missed.

[0028] The temperature data sliding window mean filtering is implemented as follows: A sliding window is created for the raw data collected by the temperature sensor (for example, a window width of 1 minute with 60 sampling points). Each time new data enters, the arithmetic mean of the data within the window is calculated and used as the filtered output value. This process eliminates transient noise such as power supply fluctuations and ambient airflow disturbances, retaining a valid signal that reflects the device's true temperature rise trend.

[0029] High-frequency interference (such as transient temperature jumps during equipment startup and shutdown) is filtered out to smooth the temperature curve and avoid false alarms caused by occasional spike signals. Temperature change trends are highlighted through mean filtering. For example, the slow temperature rise caused by progressive bearing wear can be effectively captured rather than misjudged as a transient anomaly.

[0030] When the filtered temperature data exceeds the first threshold for five consecutive sampling periods and the calculated similarity of the vibration characteristic vector is lower than the set deviation value, the main drive station is judged to have a potential compound fault risk, and a primary warning signal is generated, including the temperature exceeding the standard range, the abnormal frequency band of the vibration spectrum, and the recommended switching priority. Specifically, the first threshold is set (such as the upper limit of the normal operating temperature of the bearing is 80°C). When the filtered temperature data exceeds this threshold for five consecutive sampling periods (i.e., 5 minutes), the temperature abnormality flag is triggered; Vibration similarity comparison: When the vibration feature vector similarity is lower than the set deviation value (such as 75%), the vibration abnormality flag is triggered; Logical AND operation generates early warning: Only when the temperature and vibration abnormality flags are effective at the same time, it is determined that there is a "temperature-vibration" coupled fault risk, and a primary early warning signal containing the following information is generated: Temperature exceeding the standard range (e.g., 85°C-90°C); abnormal frequency band of the vibration spectrum (e.g., 40% increase in the energy of the 2nd harmonic of the gear meshing frequency); recommended priority for switching to the backup drive station based on the severity of the fault (e.g., high, medium, and low).

[0031] Avoid false alarms caused by a single temperature over-limit (such as a short-term heat dissipation obstruction) or a single vibration anomaly (such as an accidental impact). The measured false alarm rate has dropped from 45% to 9%. The early warning signal carries specific physical quantity abnormality information, which can be directly linked to specific fault types such as bearing wear and poor gear meshing, shortening the troubleshooting time of operation and maintenance personnel by more than 50%. The switching priority is preset according to the severity of the fault to gain buffer time for subsequent backup drive station evaluation and switching, thereby improving system response efficiency.

[0032] In a preferred embodiment of the present invention, extracting the eigenvector of the vibration signal and calculating the similarity of its frequency domain energy distribution with a pre-stored reference spectrum using a dynamic time warping algorithm include: The vibration signals collected from the main drive station are pre-processed. The preset fault-sensitive frequency band is isolated using a bandpass filter and divided into several sub-bands at equal intervals. For each sub-band, the time-frequency energy distribution is extracted using a short-time Fourier transform. The energy integral value of each sub-band per unit time is calculated, and the influence of amplitude fluctuations is eliminated through normalization to form a frequency domain energy distribution feature vector that represents the current vibration state. Specifically, the following are performed: Based on the inherent fault characteristic frequencies of the main drive station's mechanical components (such as gear meshing frequency and bearing outer ring fault frequency), the passband range of the bandpass filter is preset (e.g., 100Hz-5000Hz) to filter out environmental noise (such as power frequency interference) and non-critical frequency band signals, retaining only vibration components that are strongly related to mechanical faults; The passband range is divided into several sub-bands at equal intervals of frequency (e.g., every 500 Hz is a sub-band). For example, 100 Hz-600 Hz is the first sub-band, 601 Hz-1100 Hz is the second sub-band, and so on, to ensure that each sub-band covers the fault characteristic frequency range of a specific mechanical component.

[0033] Bandpass filtering eliminates irrelevant frequency signals (such as the low-frequency components of the motor's electromagnetic vibration), allowing subsequent analysis to focus on the frequency band sensitive to mechanical faults and improve the signal-to-noise ratio of feature extraction; sub-band division corresponds to the characteristic frequency ranges of different mechanical components (for example, gear faults are often manifested as abnormal energy of the meshing frequency and its harmonics), making it easier for subsequent analysis to be directly linked to the specific fault source.

[0034] A sliding window (e.g., window length 0.1 second, overlap rate 50%) is applied to the vibration signal in each sub-band, and the time domain signal is converted into a three-dimensional time-frequency matrix of "time-frequency-energy" through Fourier transform to capture the dynamic changes of the energy of each sub-band over time. Within each analysis period (e.g., 1 second), the energy value in the time-frequency matrix of each sub-band is integrated to obtain the total energy value of the sub-band per unit time. The integrated value of the energy of each sub-band is divided by the sum of the energies of all sub-bands to limit the value range of the eigenvector elements to between 0 and 1, thereby eliminating the interference of the overall amplitude fluctuation caused by load changes on the eigenvector.

[0035] STFT uses a sliding window to perform local time-frequency analysis of vibration signals, which can capture instantaneous energy anomalies during speed fluctuations or sudden load changes (such as the sudden increase in high-frequency energy at the moment of gear collision); normalization processing ensures that the eigenvector only reflects the energy distribution ratio of each sub-band, rather than the absolute amplitude, to avoid the overall amplitude increase caused by increased motor load (such as overload conditions of gravity energy storage systems) being misjudged as a fault feature; complex time-frequency data is compressed into a one-dimensional vector of the energy proportion of each sub-band, facilitating subsequent rapid comparison with the benchmark template.

[0036] Based on the dynamic time warping algorithm, the current eigenvector is non-rigidly aligned with the pre-stored standard eigenvector of the benchmark spectrum. The minimum cumulative path distance between the two sequences is calculated through dynamic programming, and the inverse of this distance value is mapped into a similarity score. Specifically, the current eigenvector (the sequence of energy proportions of each sub-band collected in real time) and the benchmark spectrum eigenvector (the energy distribution sequence of historical normal operating conditions) are regarded as two time series, allowing non-uniform scaling on the time axis (for example, the real-time signal causes the frequency axis to shift to the right as a whole due to a slightly faster speed). Dynamic programming is used to find the optimal matching path between the two sequences to minimize the energy difference between the corresponding sub-bands. A distance matrix is ​​constructed, in which the matrix elements are the absolute values ​​of the energy differences between the sub-bands at corresponding positions in the two sequences. The minimum cumulative distance from the starting point to the end point of the matrix is ​​recursively calculated to quantify the overall difference between the two sequences. The inverse of the minimum cumulative distance is taken (or converted through linear transformation) to a similarity score of 0-100%. The smaller the distance, the higher the similarity, indicating that the current vibration state is closer to the normal operating condition.

[0037] Traditional Euclidean distance requires strictly synchronous signal sampling, while DTW allows elastic deformation of the time axis. It can adaptively adapt to small speed fluctuations of the main drive station caused by load changes (such as ±5% rated speed deviation), avoiding misjudgments caused by frequency micro-shifts. It can detect differences in energy distribution in local sub-bands of the eigenvector (such as a 3% increase in the energy proportion of a gear sub-band), while traditional amplitude threshold alarms may miss such early signs because the overall amplitude does not exceed the limit. By continuously monitoring the changing trend of the similarity score (such as a gradual decrease from 95% to 80%), the degree of deterioration of mechanical components can be determined, providing a quantitative basis for preventive maintenance.

[0038] The standard feature vector of the reference spectrum is generated by: collecting multi-cycle vibration signals under different load conditions during the historical normal operation phase, extracting the energy integral value of each sub-frequency band according to the same frequency band segmentation rule, and calculating the mean and standard deviation of the energy integral value to form a reference energy distribution template; performing cluster analysis on the spectrum fluctuation characteristics of different load intervals to generate dynamic matching intervals, specifically including: collecting multi-cycle vibration signals under different load levels (such as light load 20%, rated load 100%, and overload 120%) during the historical normal operation phase of the main drive station (such as the new machine commissioning period and after regular maintenance) (collecting more than 50 cycles for each operating condition); extracting the energy integral value of the vibration signal under each load condition according to the same sub-frequency band segmentation rule, calculating the mean and standard deviation of the energy of each sub-frequency band, and forming a reference energy distribution template that includes a normal fluctuation range (such as a sub-frequency band with a mean of 0.25 and a standard deviation of 0.03); clustering the reference template based on load size and dividing it into several dynamic matching intervals (such as a light load area, a medium load area, and a heavy load area), each interval corresponding to a specific reference template and similarity threshold.

[0039] Through training with data from multiple load conditions, the benchmark template incorporates normal spectral characteristics under different operating conditions (e.g., the natural increase in gear meshing frequency energy under heavy load), avoiding misjudgments caused by cross-condition comparisons (e.g., falsely reporting energy anomalies when using a light-load template to evaluate a heavy-load condition). The benchmark template for the corresponding interval is automatically called based on real-time load data (e.g., calling the medium-load zone template when the current sensor determines the current load is 85% of the rated value). This makes the similarity calculation more consistent with actual operating conditions, increasing the benchmark matching accuracy from 68% to 91%. The floating similarity deviation value is set based on the standard deviation of each load interval (e.g., the light-load zone allows a similarity of ≥85%, and the heavy-load zone allows a similarity of ≥75%), avoiding threshold rigidity caused by differences in operating conditions.

[0040] In a preferred embodiment of the present invention, in response to the primary warning signal, real-time oil pressure data of the gearbox, bearing temperature rise rate curve, and dynamic torque feedback value of the target standby drive station are collected to calculate the availability index, including: The gearbox oil pressure sensor is used to obtain real-time oil pressure data from the target standby drive station at a sampling frequency of 10 times per second. Upper and lower limits are set based on the historical normal oil pressure fluctuation range. The currently calculated standard deviation of the oil pressure fluctuation is mapped to the range of 0-1 to obtain a preliminary oil pressure status score. A sliding window is used to calculate the standard deviation of the oil pressure fluctuation. The bearing temperature sensor continuously collects temperature data, and the temperature rise rate is calculated every 5 seconds. This is the temperature change at adjacent time points divided by the time difference to generate a temperature rise rate curve. The maximum slope value within the last 3 minutes is extracted and normalized based on the slope value and the rated temperature rise threshold of the bearing model to obtain a normalized score, including: The gearbox oil pressure sensor collects oil pressure data in real time at a frequency of 10 times per second (for example, the normal operating oil pressure range is 10-15MPa) to form a high-density time series; the sliding window duration is set (for example, 30 seconds, including 300 sampling points), and each time new data enters, the standard deviation of the oil pressure value in the window is calculated to reflect the amplitude of oil pressure fluctuation. For example, if the oil pressure value in the window is 12±0.5MPa, the standard deviation is 0.2MPa. If the fluctuation increases to 12±2MPa, the standard deviation increases to 1.0MPa; based on historical normal operating data (for example, the standard deviation is usually ≤0.3MPa), the current standard deviation is linearly mapped to the range of 0-1. For example, a standard deviation of 0.3MPa corresponds to a score of 1.0 (best), and a standard deviation of 1.5MPa corresponds to a score of 0.2 (worst); The bearing temperature sensor collects temperature data every 5 seconds (for example, the initial temperature is 40°C, 42°C after 5 seconds, and 45°C after another 5 seconds). The temperature rise rate (ΔT / Δt) is calculated for the temperature values ​​at adjacent time points. For example, the rate from 40°C to 42°C is 0.4°C / s, and the rate from 42°C to 45°C is 0.6°C / s. This generates a temperature rise rate curve that changes over time. The curve for the most recent 3 minutes (36 sampling points) is intercepted, and the maximum slope value (i.e., the maximum temperature rise rate) is extracted. For example, the maximum rate for a certain bearing within 3 minutes is 1.2°C / s. Based on the rated temperature rise threshold of the bearing model (for example, the maximum allowable temperature rise rate for a certain bearing model is 1.5°C / s), the maximum slope value is divided by the threshold to obtain a normalized score (for example, 1.2 / 1.5 = 0.8, indicating that the current temperature rise rate is 80% of the threshold).

[0041] By continuously calculating the temperature rise rate rather than just monitoring the absolute temperature value, it is possible to identify early overheating risks where the temperature does not exceed the threshold but the increase rate is too rapid (such as the accelerated temperature rise in the early stages of grease deterioration), providing an early warning 2-4 hours earlier than traditional temperature threshold alarms. Normalization processing makes the temperature rise rates of bearings of different specifications comparable, eliminating the need to set separate thresholds for each model and simplifying the complexity of system parameter configuration.

[0042] The dynamic torque feedback value is obtained through the torque sensor, and the deviation percentage from the preset theoretical torque value is recorded. The difference between the maximum and minimum torque fluctuation amplitude is calculated in a 10-second cycle. The geometric mean of the deviation percentage and the fluctuation amplitude difference is calculated. The initial availability index is generated based on the normalized score and the geometric mean. The exponentially weighted moving average algorithm is used to integrate the historical index values ​​of the first five calculation cycles to correct the initial availability index to obtain the final availability index, including: The torque sensor is used to obtain the output torque value of the drive station in real time. This value is compared with the preset theoretical torque value (e.g., the theoretical value calculated based on the current load is 5000 N·m). The deviation percentage is calculated (e.g., the actual torque is 5500 N·m, with a deviation of +10%). Fluctuation amplitude statistics: With a 10-second period (including 100 sampling points), record the maximum and minimum torque values ​​within the period (e.g., maximum value 5800 N·m, minimum value 5200 N·m), and calculate the difference (600 N·m) as the fluctuation amplitude; Geometric mean calculation: The deviation percentage and the fluctuation amplitude difference are dimensionlessly processed (for example, the deviation percentage is taken as the absolute value, and the fluctuation amplitude is divided by the rated torque), and the geometric mean of the two is calculated to comprehensively reflect the accuracy and stability of the torque.

[0043] Simultaneously considering torque deviation (accuracy) and fluctuation amplitude (stability) avoids the drawbacks of a single metric. For example, even when torque deviation is within the allowable range but fluctuates significantly (such as periodic shocks caused by poor gear mesh), the fluctuation amplitude can still be used to identify risks. The geometric mean assigns equal weight to both indicators, better reflecting their synergistic impact than the arithmetic mean, allowing for a comprehensive assessment of issues such as transmission system load balancing and gear mesh quality.

[0044] The oil pressure status score (0-1), the temperature rise rate normalized score (0-1), and the torque geometric mean (0-1) are weighted and summed (for example, the weights are 0.4, 0.3, and 0.3, respectively) to generate the initial availability index (range 0-1, with higher values ​​indicating better status). For example, if the oil pressure score is 0.8, the temperature rise score is 0.7, and the torque geometric mean is 0.6, the initial index is 0.8×0.4+0.7×0.3+0.6×0.3=0.73; Exponentially Weighted Moving Average (EWMA) Correction: The historical index values ​​for the previous five calculation periods (e.g., each period is 1 minute, for a total of 5 minutes) are used for smoothing using an exponential weighting method. Recent data is given a higher weight (e.g., 0.4 for the current period, 0.3 for the previous period, 0.2 for the previous two periods, and 0.1 for the previous three periods and older) to mitigate incidental data interference and produce the final availability index. For example, if the indices for the previous five periods were 0.75, 0.73, 0.74, 0.72, and 0.71, they would be corrected to 0.732 after weighting.

[0045] Avoid a single parameter dominating the evaluation results. For example, when the oil pressure is normal but the torque fluctuates violently, the initial index will be pulled down by the torque term, truly reflecting the overall status of the equipment. The EWMA algorithm uses a weighted average of historical data to filter out short-term interference (such as abnormal values ​​caused by momentary false touches of sensors), making the index more in line with the actual health trend of the equipment, and improving the stability of the evaluation results by more than 50%.

[0046] In a preferred embodiment of the present invention, the process of determining the modified index is as follows: High-precision temperature sensors are deployed at both the primary and secondary monitoring points, collecting real-time temperature data at a frequency of once per second. Specifically, high-precision thermocouple sensors are installed on the transmission chain of the target backup drive station, at the gearbox input flange (the primary monitoring point) and the drive motor output coupling, and at the connection between the bearing seat and the output shaft (the secondary monitoring point). These sensors synchronously collect real-time temperature data at a frequency of once per second. A hardware clock synchronization module or software timestamp calibration algorithm eliminates millisecond-level delays caused by differences in signal transmission paths between the two monitoring points, ensuring strict timeline alignment of temperature data.

[0047] The temperature data from the two monitoring points is time-aligned to eliminate phase errors caused by signal transmission delays. The instantaneous temperature difference along the drive shaft is calculated between the two monitoring points. Based on the temperature change rate over the last 10 minutes, an axial temperature gradient is generated to characterize heat transfer anomalies in the drive train. Specifically, the system calculates the instantaneous axial temperature difference (temperature at the first monitoring point minus temperature at the second monitoring point) in real time based on the temperature data collected simultaneously at the two monitoring points. For example, if the input temperature is 65°C and the output temperature is 58°C, the instantaneous temperature difference is +7°C. Based on the temperature data from the last 10 minutes, the rate of change of the temperature difference (e.g., the increase / decrease in the temperature difference per minute) is calculated to generate the axial temperature gradient. If the initial temperature difference is +5°C and changes to +9°C after 10 minutes, the gradient is +0.4°C / min (a positive gradient indicates heat migration toward the output end). Conversely, if it changes to +3°C, the gradient is -0.2°C / min (a negative gradient indicates heat accumulation toward the input end).

[0048] The heat transfer efficiency is determined by the absolute value of the temperature difference (for example, a large temperature difference may increase the contact thermal resistance due to insufficient lubrication of the coupling), and the fault development trend is determined by the direction of the gradient difference (a positive gradient may indicate bearing wear at the output end, and a negative gradient may indicate oil leakage in the gearbox at the input end). Compared with single temperature monitoring, the gradient difference can identify hidden faults such as "abnormal heat distribution without exceeding the threshold temperature" (such as heat flux concentration caused by a reduction in the local contact area of ​​the transmission chain) in advance, with early warning time 12-24 hours earlier.

[0049] Three-axis vibration acceleration sensors are installed at the first and second monitoring points to simultaneously collect radial vibration signals, that is, vibration signals perpendicular to the transmission shaft. Wavelet packet decomposition is performed on the vibration signals to extract the preset fault-sensitive frequency bands, and the energy integral values ​​of each frequency band at the two monitoring points are calculated. The total energy integral value of the first monitoring point and the total energy integral value of the second monitoring point are weighted and summed according to the frequency band to generate the radial vibration energy ratio. A ratio greater than 1 indicates that the vibration energy is concentrated toward the input end, and a ratio less than 1 indicates that it is diffused toward the output end. A coupling coefficient matrix of the axial temperature gradient difference and radial vibration energy ratio is preset, and the matrix weight is obtained based on the correlation training of the two parameters in historical failure cases; Based on the absolute value and direction of the current axial temperature gradient difference, a positive gradient indicates heat migration toward the output end, while a negative gradient indicates heat accumulation toward the input end. Combined with the distribution characteristics of the vibration energy ratio (i.e., aggregation or diffusion), the corresponding correction index is matched from the coupling coefficient matrix. This involves collecting past drive station failure cases (such as gear wear, bearing failure, and coupling misalignment), annotating the corresponding axial temperature gradient difference, radial vibration energy ratio, and actual fault type. A support vector machine (SVM) algorithm is used to train the coupling coefficient matrix, using the temperature gradient difference (absolute value and direction) and the vibration energy ratio (value and distribution) as input features and the fault severity as the output label. Each element in the matrix represents a correction factor for a specific parameter combination (e.g., a temperature gradient of +0.5°C / min and an energy ratio of 1.5 results in a correction factor of -0.15). Based on the current temperature gradient difference and energy ratio, the corresponding correction factor is retrieved from the matrix to apply an addition or subtraction to the initial availability index (e.g., an initial index of 0.73, a correction factor of -0.15, and a corrected index of 0.58). Breaking away from the limitations of traditional empirical formulas, the system automatically learns the coupling laws of thermal and vibration parameters through historical fault data, improving the accuracy of correction coefficients by 65% ​​compared to manual settings. The matrix can cover a variety of complex fault scenarios (such as bearing temperature rise accompanied by abnormal gear vibration), avoiding the one-sidedness of single parameter correction, and reducing the assessment error of the availability index on transmission chain coordination from ±20% to ±6%.

[0050] In a preferred embodiment of the present invention, the vibration signal is decomposed by wavelet packets to extract the preset fault-sensitive frequency bands, and the energy integral values ​​of the two monitoring points in each frequency band are calculated respectively, including: According to the fault characteristic frequency range of the target transmission chain, including gear meshing frequency, bearing defect characteristic frequency and vibration signal sampling rate, the number of wavelet packet decomposition layers is determined. Specifically, based on the known fault characteristic frequencies of the transmission chain (such as gear meshing frequency, bearing outer ring / inner ring defect characteristic frequency) and vibration signal sampling rate, the highest analysis frequency is determined by the Nyquist sampling theorem, and then the number of wavelet packet decomposition layers is reversed according to the frequency band division requirements (such as dividing the target frequency range into several sub-bands). For example, if the sampling rate is 10kHz, the target analysis frequency is 0-5kHz, and it is desired to be decomposed into 8 sub-bands, then a 3-layer decomposition (2 3 =8), so that each sub-band covers a frequency range of approximately 625 Hz. This targeted coverage of fault-sensitive frequency ranges avoids interference from irrelevant frequency bands, ensuring that the decomposed sub-bands can accurately capture abnormal vibration characteristics of key components such as gears and bearings. Multi-resolution analysis is achieved through hierarchical decomposition, adapting to the non-stationary nature of vibration signals and better reflecting real-time frequency component changes than a single Fourier transform.

[0051] The original vibration signal of each monitoring point is decomposed by wavelet packet using a certain number of layers to obtain the wavelet packet coefficients of each sub-band. The wavelet packet coefficients represent the vibration information in different frequency bands. Specifically, the original vibration signal (time domain signal) of each monitoring point is decomposed by wavelet packet using a certain number of decomposition layers to decompose the signal into 2 N sub-bands (N is the number of layers). Each sub-band corresponds to a set of wavelet packet coefficients, and the amplitude of the coefficients reflects the strength of the vibration energy in that frequency band. For example, after 3-layer decomposition, 8 sub-bands are obtained (0-625Hz, 625-1250Hz...4375-5000Hz), each of which corresponds to a set of coefficient sequences.

[0052] Decomposing complex vibration signals into energy distributions in independent frequency bands facilitates targeted analysis of the frequency components corresponding to specific faults (e.g., gear wear often manifests itself as an increase in the meshing frequency harmonic components). This preserves the signal's time-frequency characteristics and allows for simultaneous analysis of the dynamic changes in vibration energy in both time and frequency dimensions, making it suitable for extracting weak features of early-stage faults.

[0053] For preset fault-sensitive sub-bands, the corresponding time domain signals are reconstructed using their wavelet packet coefficients, and the square of the reconstructed signal amplitude is integrated within a selected time window to calculate the energy integral value of the frequency band. Specifically, for preset fault-sensitive sub-bands (such as the gear meshing frequency and its 2-3 harmonics, and the frequency band containing the bearing characteristic frequency), the corresponding wavelet packet coefficients are used to reconstruct the time domain signal (i.e., "extracting" the vibration components of the frequency band from the decomposed coefficients). For the reconstructed time domain signal, the integral value of the square of the amplitude is calculated within a selected time window (such as 1 second) as the energy integral value of the frequency band. The larger the energy integral value, the stronger the vibration energy in the frequency band and the higher the potential fault risk.

[0054] By filtering out interference from irrelevant frequencies, energy calculations are made more targeted, preventing noise from other frequency bands from masking fault signals. Vibration intensity is quantified through energy integration, converting qualitative signal characteristics into comparable quantitative indicators.

[0055] For the energy integral values ​​that are not fault-sensitive frequency bands, baseline subtraction is first performed to eliminate the influence of background noise, and the energy integral values ​​of all sensitive sub-frequency bands are normalized to the 0-1 range according to their maximum and minimum values. Specifically, for non-fault-sensitive frequency bands, baseline subtraction (such as subtracting the signal mean or long-term trend term) is first performed to eliminate the influence of background noise or mechanical inherent vibration, and then the energy integral values ​​of all sensitive sub-frequency bands are normalized: the energy value of each frequency band is divided by the difference between its historical maximum and minimum values, and mapped to the 0-1 range. For example, the historical energy range of a frequency band is 0-100mV. 2If the current energy value is 60, it will be normalized to 0.6. Baseline subtraction can eliminate interference from ambient vibration or natural vibration during normal operation, preventing misjudgments (such as non-fault vibration during equipment startup). Normalization also eliminates the impact of energy amplitude differences in different frequency bands (such as the naturally lower energy in high-frequency bands), making indicators across frequency bands comparable.

[0056] According to the correlation between the fault type and the preset frequency band, a weight is assigned to each sensitive sub-band, and the weighted sub-band energy value is generated by multiplying the normalized energy integral value by its corresponding weight. Specifically, a weight (0-1 range) is assigned to each sensitive sub-band based on historical fault cases or expert experience. The size of the weight reflects the strength of the correlation between the frequency band and the specific fault. For example, the weight of the gear meshing frequency multiplication frequency band for gear wear fault is set to 0.8, and the weight of the bearing characteristic frequency band for bearing fault is set to 0.7. The normalized energy value is multiplied by the corresponding weight to obtain the weighted sub-band energy value for subsequent comprehensive analysis (such as vibration energy ratio calculation). The present invention highlights the impact of key fault characteristics. For example, in gearbox fault assessment, increasing the weight of the meshing frequency related frequency band can make the system more sensitive to gear wear.

[0057] In a preferred embodiment of the present invention, the driving station state monitoring system for the gravity energy storage transport track according to claim 6 is characterized in that the total energy integral value of the first monitoring point and the total energy integral value of the second monitoring point are weighted and summed according to the frequency band to generate a radial vibration energy ratio, wherein a ratio greater than 1 indicates that the vibration energy is concentrated toward the input end, and a ratio less than 1 indicates that it is diffused toward the output end, including: Based on the historical fault database, the correlation between different frequency bands and specific fault types is determined, namely: the high frequency band of 2-4kHz corresponds to gear tooth surface wear, the medium frequency band of 500Hz-2kHz corresponds to bearing raceway defects, and the low frequency band below 500Hz corresponds to coupling misalignment; Based on the current operating conditions of the drive station and the potential fault types indicated in the primary warning signal, dynamic weights are assigned to each frequency band; The energy integral values ​​of each sensitive frequency band of the first monitoring point are weighted and summed according to the assigned dynamic weights. The energy of each frequency band is multiplied by the corresponding weight and then accumulated to obtain the total energy value on the input side; For the energy integral value of the same frequency band at the second monitoring point, the same weight distribution rule as that on the input side is used for weighted summation to calculate the total energy value on the output side; Deduct the preset baseline energy value from the total energy value of the input and output sides. The baseline value is the weighted sum of the energy of each frequency band when the equipment is in a no-load and fault-free state. Divide the corrected total energy value on the input side by the corrected total energy value on the output side to generate the radial vibration energy ratio. If the energy on the input side is significantly higher than that on the output side, the ratio is greater than 1; otherwise, it is less than 1. Perform median filtering on the ratio of five consecutive sampling periods to eliminate abnormal fluctuations caused by instantaneous shocks and retain trend change characteristics; Energy propagation direction determination and fault mapping: Energy accumulation determination, ratio > 1: If the ratio is > 1.2 for three consecutive cycles, it is determined that the vibration energy is concentrated at the input end. Related fault types include drive motor rotor imbalance, loose coupling bolts, or gearbox input shaft eccentricity. If the high-frequency band accounts for more than 70%, local gear damage is the priority. If the mid-frequency band accounts for a large proportion, it indicates insufficient bearing preload. Energy diffusion judgment, ratio <1: If the ratio is <0.8 for five consecutive cycles, it is determined that the energy is diffusing to the output end. Related fault types include loose output shaft bearing seat bolts, output shaft bending, or load end mechanical seizure. When the low frequency band is dominant, it indicates shaft misalignment. When the medium and high frequency bands are prominent, it indicates output end bearing lubrication failure.

[0058] In this embodiment of the present invention, data is extracted from a historical fault database to analyze the correlation between vibration energy in different frequency bands and specific fault types. The high-frequency band (2-4kHz) primarily corresponds to gear tooth wear faults, as the impact of gear meshing concentrates energy in the high-frequency band. The mid-frequency band (500Hz-2kHz) is associated with bearing raceway defects, where contact between bearing rolling elements and defective surfaces excites vibrations in this frequency band. The low-frequency band (<500Hz) is associated with coupling misalignment faults, where shaft offset increases low-frequency vibration energy. This process establishes a "frequency band-fault type" mapping table by statistically analyzing the energy contribution of each frequency band at the time of the fault in the historical data.

[0059] Dynamic weight allocation: Based on the current operating conditions of the drive station (real-time parameters such as load, speed, and temperature) and the potential fault types indicated by the primary warning signal (such as gear, bearing, or coupling-related abnormalities), the correlation coefficient of the corresponding frequency band is extracted from the "frequency band-fault type" mapping table and used as the dynamic weight.

[0060] For example, if the primary warning indicates a gear anomaly, the high-frequency band (2-4kHz) is weighted more heavily; if it indicates a bearing problem, the mid-frequency band is weighted more heavily. Weights range from 0 to 1, and the sum is normalized to ensure that fault-sensitive frequency bands of particular concern are prioritized under different operating conditions.

[0061] Calculation of total energy on the input and output sides: Input side (first monitoring point): Multiply the energy integral value of each sensitive frequency band by the corresponding dynamic weight, and add them up to obtain the total energy value of the input side (that is, the energy of each frequency band is weighted and summed according to the weight).

[0062] Output side (second monitoring point): Use the same weight allocation rule as the input side to perform weighted summation on the energy integral values ​​of the same frequency band to obtain the total energy value on the output side.

[0063] This step uses weight differentiation to highlight the energy contribution of the frequency band related to the current potential fault and weaken the interference of irrelevant frequency bands.

[0064] Baseline energy subtraction: The preset "baseline energy value" is the weighted sum of the energy in each frequency band when the device is in a no-load and fault-free state (obtained based on historical normal operation data statistics).

[0065] The baseline energy value is deducted from the total energy value on the input side and the output side respectively to obtain the corrected energy value, so as to eliminate the background energy influence during normal operation of the equipment and only retain the energy change under abnormal state.

[0066] Calculation of radial vibration energy ratio: Divide the input-side corrected total energy value by the output-side corrected total energy value to generate the radial vibration energy ratio: If the energy on the input side is significantly higher than that on the output side (ratio > 1), it means that the vibration energy is concentrated towards the input end of the drive station; If the energy on the output side is higher (ratio < 1), it means that the energy is diffused toward the output end.

[0067] Median filter denoising: Perform median filtering on the ratio of five consecutive sampling periods, that is, take the middle value of the five values ​​as the effective ratio of the current period.

[0068] This operation can eliminate abnormal fluctuations caused by transient shocks (such as accidental load fluctuations and external vibration interference) and retain stable signals that reflect the trend of energy propagation.

[0069] Energy propagation direction determination and fault mapping: Energy accumulation determination (ratio>1): If the ratio is greater than 1.2 for three consecutive cycles, it is determined that the vibration energy is concentrated at the input end, which may be related to the drive motor rotor imbalance, loose coupling bolts, or gearbox input shaft eccentricity. Further analysis of the frequency band proportion: If the high-frequency band energy proportion exceeds 70%, it will first indicate local gear damage (such as tooth surface wear and tooth root cracks); if the mid-frequency band proportion is prominent, it will indicate insufficient bearing preload or inner ring failure.

[0070] Energy diffusion determination (ratio < 1): If the ratio is less than 0.8 for five consecutive cycles, it is determined that the energy is diffusing to the output end, which may be related to loose bolts on the output shaft bearing seat, bent output shaft, or mechanical jamming at the load end. When the low frequency band is dominant, it indicates that the shaft system is misaligned (such as coupling offset); when the medium and high frequency bands are prominent, it indicates that the output end bearing lubrication has failed or the raceway is worn.

[0071] Frequency band weights are adjusted based on real-time operating conditions and primary warning results, allowing the system to focus on the sensitive frequency bands corresponding to the current potential fault, avoiding "one-size-fits-all" analysis and improving the accuracy of anomaly identification. By deducting the baseline energy of the no-load, fault-free state, the abnormal increase in energy during actual operation is highlighted, effectively distinguishing between normal fluctuations and energy changes related to faults. After eliminating instantaneous noise interference, the ratio can better reflect the trend changes in energy propagation, reduce the false alarm rate, and improve the reliability of monitoring results. Combined with the energy ratio direction, duration, and frequency band proportion characteristics, the abstract energy distribution differences are mapped to specific fault types (such as gear, bearing, and coupling problems), providing maintenance personnel with a clear maintenance direction and shortening troubleshooting time.

[0072] In a preferred embodiment of the present invention, when the corrected index is lower than the second threshold, the control link of the current backup drive station is automatically disconnected and the cascade switching protocol is activated, and the suboptimal backup drive station is selected according to the device status parameters pre-stored in the drive station topology network, including: Based on the real-time load distribution of the driver station topology network, the second threshold is dynamically adjusted. That is, if the overall network load rate is greater than 80%, the second threshold is adjusted from 0.6 to 0.7 to improve the switching sensitivity. The second threshold is compensated based on temperature and humidity. If the temperature and humidity exceed the upper limit of the device's rated operating range, the threshold is increased by 10%; Before sending a shutdown command to the current standby drive station, its operating parameters are synchronously backed up to the central database; Sending a shutdown command simultaneously through the primary and backup communication links to ensure that at least one link is successfully executed; After cutting off the control link, the current decay curve of the drive motor is monitored in real time. The mechanical brake lock is released only after confirming that the motor has completely stopped. The pre-stored drive station topology map is read to identify the geographical location and connection status of all backup drive stations. The priority scores of the remaining candidate backup stations are calculated, an activation command is sent to the suboptimal backup drive station, and the operating parameters of the faulty station are synchronized to linearly increase the output torque of the suboptimal station from 0 to the target value within 3 seconds.

[0073] In an embodiment of the present invention, the load data of each device in the drive station topology network (such as the current operating power, torque output, etc.) is obtained in real time, and the overall load rate of the network (that is, the percentage of the current total load to the rated total load) is calculated. If it is detected that the overall load rate exceeds 80%, the originally set second threshold (such as 0.6) is automatically adjusted to 0.7, thereby improving the system's sensitivity to switching of the standby drive station status, avoiding the inability to switch in time when the standby drive station fails due to excessive network load, and improving the system's fault tolerance in an overload environment.

[0074] Environmental sensors deployed at the drive station monitor temperature and humidity data in real time and compare them with the equipment's rated operating range (e.g., a temperature upper limit of 60°C and a humidity upper limit of 85%). If the current temperature or humidity exceeds the rated upper limit, the second threshold is dynamically adjusted and increased by another 10% (e.g., from 0.7 to 0.77) to adapt to the impact of harsh environments on equipment reliability, consider the impact of environmental factors on equipment performance, avoid unrecognized hidden faults in the standby drive station due to high temperature and high humidity, and enhance the environmental adaptability of threshold judgment.

[0075] Before sending a shutdown command to the current standby drive station, its operating parameters (such as current torque output, speed, fault log, control parameter configuration, etc.) are collected in real time and synchronously backed up to the central database. The backup data includes real-time status data and historical short-term trend data to ensure that the operating status of the equipment before the abnormality can be traced when a fault occurs.

[0076] The shutdown command is sent to the current standby drive station simultaneously through two independent communication links (such as optical fiber and wireless links) to avoid command loss due to single link failure. The command execution feedback signal is monitored in real time to confirm that at least one link has successfully triggered the shutdown process (such as receiving the "shutdown confirmation" response from the drive station). If not, the resend mechanism is activated to ensure the reliable execution of the shutdown command, avoid the failure of faulty equipment to shut down in time due to communication failure, and reduce safety risks and the probability of equipment damage.

[0077] After cutting off the control link, monitor the current signal of the drive motor in real time and determine whether the motor has completely stopped by the current decay curve (for example, the current drops below the no-load current and remains stable for 2 seconds). After confirming that the motor has stopped, release the mechanical brake lock to avoid mechanical shock or wear caused by forced braking when the motor is still rotating.

[0078] Read the pre-stored drive station topology map to obtain the physical location (such as installation floor, equipment number) and real-time connection status (such as whether it is online, whether the communication is normal, current load rate, etc.) of all backup drive stations, filter out faulty or offline devices, select candidate backup drive stations with normal status, quickly locate available backup equipment, reduce search time during switching, and ensure that the system resumes operation in the shortest time.

[0079] For the remaining candidate backup drive stations, a priority score is calculated based on pre-stored equipment status parameters (such as historical health index, current load rate, physical distance from the faulty station, maintenance cycle, etc.), and the suboptimal backup station with the highest score is selected. An activation command is sent to the suboptimal station, synchronizing the operating parameters of the faulty station (such as target torque, speed control parameters, etc.), and its output torque is linearly increased from 0 to the target value within 3 seconds to avoid the impact of torque mutation on the transmission chain. The backup station is intelligently selected based on the equipment status to ensure stable system operation after switching; the linear torque increase reduces mechanical shock, ensures a smooth transition of the transport track load, and improves the reliability and robustness of the entire system.

[0080] In a preferred embodiment of the present invention, after the drive station switching operation is completed, the real-time acceleration change rate of the transport track slope section is monitored. When the speed decay gradient exceeds 120% of the safety threshold within 3 seconds, the emergency braking program of the hydraulic wedge brake device is triggered and the torque compensation mechanism of the two adjacent drive stations is activated until the speed parameter returns to the safe range, including: Real-time collection of acceleration data from the transport track, sliding window mean filtering of the raw acceleration data, and extraction of the smoothed acceleration change rate, i.e., the derivative of acceleration per unit time; The safety threshold is dynamically calculated based on the slope angle and current load mass: For every 1° increase in inclination, the safety threshold increases by 5%; For every 10% increase in load mass over the rated value, the safety threshold increases by 3%; Based on historical operating data, a 10% redundancy margin is applied to the safety threshold; In a 3-second time window, the moving average of the acceleration change rate is calculated at intervals of 0.5 seconds to generate a velocity decay gradient curve; Perform linear fitting on the gradient curve and calculate the absolute value of its slope. If the slope of three consecutive data points exceeds 120% of the dynamic safety threshold, it is determined to be an abnormal attenuation event. Multi-stage braking strategy that triggers the hydraulic wedge brake: Level 1 braking: applies braking force at 50% of the rated pressure for 1 second to reduce the risk of mechanical shock; Secondary braking: If the gradient does not fall within the threshold after primary braking, the pressure is increased to 80% of the rated value within 2 seconds, and the brake disc temperature is monitored in real time; Level 3 braking: If the level 2 braking is still ineffective, full-pressure braking is triggered and the brake pad spray cooling system is started; Generating torque compensation values ​​of two adjacent driving stations in proportion according to the percentage of the speed attenuation gradient exceeding the threshold; The rising slope of the compensation value is dynamically adjusted according to the length of the slope section and the remaining load mass. The drive station closest to the starting end of the slope section is selected to take on 70% of the compensation torque first. The drive station adjacent to the end of the slope section is selected to bear 30% of the compensation torque and its output phase angle is adjusted synchronously; when the speed value returns to the safe range, braking and compensation are terminated.

[0081] In an embodiment of the present invention, an acceleration sensor installed on a graded section of a transport track collects high-frequency acceleration data in real time (e.g., 200 times per second). A sliding window mean filter is applied to the raw data (e.g., a window size of 1 second containing 200 data points). The average value of the data within the window is calculated to filter out high-frequency noise, resulting in a smoothed acceleration curve. This smoothed acceleration curve is then numerically differentiated (e.g., the difference between two adjacent points divided by the time interval) to obtain the acceleration rate of change per unit time (i.e., "jerk," representing the severity of speed decay or acceleration). This filtering process eliminates transient interference such as vibration and shock, ensuring the stability and accuracy of the acceleration rate calculation, avoiding misjudgments, and directly reflecting the severity of speed changes.

[0082] The initial safety threshold is determined based on track design parameters (such as the maximum allowable speed attenuation gradient when unloaded). For every 1° increase in the current slope angle, the safety threshold increases by 5% based on the initial value (for example, at a 10° inclination, the threshold increases by 50% compared to a flat section). The load mass of the transport vehicle is obtained in real time. For every 10% increase in the rated load, the threshold increases by 3% (for example, a 20% excess load increases the threshold by 6%). Based on the above calculation results, an additional 10% safety margin is added (for example, final threshold = basic threshold × inclination compensation coefficient × load compensation coefficient × 1.1). Redundant margin design reserves a safety buffer to prevent missed detections due to sensor errors or operating condition fluctuations.

[0083] Within a 3-second time window, at 0.5-second intervals (a total of 6 sampling points), the moving average of the acceleration change rate is calculated to generate a "speed decay gradient curve" (each point represents the average gradient from the previous 0.5 seconds to the current moment). A linear fit is performed on the gradient curve, and the absolute value of the slope is calculated (reflecting the steepness of the gradient change trend). If the slope of three consecutive data points (i.e., within 1.5 seconds) exceeds 120% of the dynamic safety threshold (for example, when the threshold is 0.5m / s³, the judgment condition is that the slope is greater than 0.6m / s 3 ), it is determined that an "abnormal speed decay event" has occurred. Moving average and linear fitting can filter out short-term fluctuations and capture persistent trend anomalies (such as continuous excessive deceleration). Setting the judgment condition of "three consecutive data points" can avoid false triggering caused by a single accidental impact and improve the reliability of event detection.

[0084] Immediately after being triggered, the braking force is applied at 50% of the rated braking pressure for 1 second. Mechanical shock (such as the risk of instantaneous locking of the brake disc and brake pad) is reduced through progressive braking. If the speed attenuation gradient does not drop to the threshold after the first level of braking, the pressure will be gradually increased to 80% of the rated value within 2 seconds. At the same time, the brake disc temperature is monitored in real time (a temperature warning will be issued when it exceeds 150°C). If the second level braking is still ineffective (the gradient continues to exceed the standard), 100% of the rated pressure will be immediately applied for full-pressure braking, and the brake pad spray cooling system will be started simultaneously (such as spraying water mist to reduce friction heat).

[0085] The graded braking strategy balances braking efficiency and equipment protection through "soft first, hard later" pressure control, avoiding mechanical damage caused by sudden braking (such as gearbox impact and bearing overload). Temperature monitoring and cooling system linkage prevent brake disc failure caused by overheating, thereby improving the reliability and life of the braking system.

[0086] Based on the percentage by which the speed attenuation gradient exceeds the threshold (for example, a 50% compensation torque requirement is generated if the speed attenuation gradient exceeds the threshold by 50%), the initial compensation values ​​for the two adjacent drive stations are generated proportionally. Combined with the length of the slope section (for example, a gentler compensation slope is used for long slope sections to avoid sudden torque changes) and the remaining load mass (the heavier the load, the gentler the compensation slope), the rate of increase of the compensation torque is adjusted (for example, increasing the rated torque by 10% to 30% per second). The drive station closest to the starting end of the slope section is selected to take on 70% of the compensation torque (due to its proximity to the power input end, the response is more direct). The adjacent drive station at the end of the slope section takes on 30%, and its output phase angle is adjusted synchronously (to ensure synchronous torque output of the two drive stations to avoid torsional vibration of the shaft system).

[0087] When the speed value returns to the safe range (gradient ≤ dynamic threshold), the compensation torque is gradually reduced to 0 to avoid reverse impact caused by sudden unloading. Compensation priority is assigned according to distance, and the compensation strategy is dynamically adjusted based on load and slope to achieve "precise compensation" and quickly curb abnormal speed attenuation while avoiding drive station overload. Synchronous phase angle adjustment can reduce the risk of torsional vibration in the mechanical transmission chain and improve the stability of collaborative operation of multiple drive stations.

[0088] Through the above steps, the system can implement closed-loop control of "monitoring-judgment-braking-compensation" for abnormal speed attenuation events on slope sections after the drive station is switched, effectively preventing the risk of transport loss of control due to excessive braking or insufficient compensation. At the same time, it protects mechanical components through graded braking and dynamic torque distribution, thereby improving the safety and reliability of the gravity energy storage transportation system.

[0089] In a preferred embodiment of the present invention, when the speed value returns to the safe range, braking and compensation are terminated, including: When the speed value returns to the safe range, braking and compensation can be terminated only when the following conditions are met at the same time: The sliding window standard deviation of the acceleration rate of change is less than 1.2 times the baseline value; The temperature gradient difference between adjacent drive stations is less than 5℃ / min; Hydraulic brake pressure fluctuation <10% of rated value; Gradual Exit Strategy: Brake release: Gradually release the hydraulic brake at a rate of 20% pressure reduction every 0.5 seconds, while monitoring the speed rebound gradient. If the rebound exceeds the threshold of 50%, the release is suspended; Torque withdrawal: The compensation torque is reduced at a rate of 10% / second until it returns to the initial setting value after switching. If emergency braking is triggered twice or more on the same slope within 24 hours, the calibration coefficient of the corresponding safety threshold will be automatically increased by 15%, and the area will be marked as a high-risk area, triggering manual inspection.

[0090] In this embodiment of the present invention, a sliding window (e.g., a 2-second window containing four sampling points at 0.5-second intervals) is established for the acceleration rate of change data. The standard deviation of the data within the window is calculated in real time to reflect the degree of acceleration fluctuation. The current standard deviation is compared with a baseline value (the statistical mean of the standard deviation during normal equipment operation). If it is less than 1.2 times the baseline value, it indicates that the speed change is becoming stable. This avoids misjudgments caused by short-term fluctuations, ensures the stability of the speed attenuation gradient, and prevents abnormal speed fluctuations after termination.

[0091] Continuously monitor the temperature data at the first monitoring point (gearbox input) and the second monitoring point (bearing seat output) on the target standby drive station's transmission chain, and calculate the difference in the temperature change rate (i.e., the temperature gradient difference). If the gradient difference is less than 5°C / min, it indicates that the transmission chain's heat transfer is stable and there is no abnormal frictional heating. This thermal status verifies the coordination of mechanical components and prevents subsequent failures caused by undetected abnormal wear of bearings or gears.

[0092] The system collects real-time data from the hydraulic wedge brake's pressure sensor and calculates the fluctuation between the current pressure and the rated brake pressure (e.g., if the current pressure fluctuates within ±10% of the rated value). If the fluctuation is less than 10% of the rated value, the braking system is stable and controllable. This prevents braking from being terminated when the brake pressure is unstable, and mitigates the risk of a sudden drop in braking force due to uneven brake pad contact or hydraulic system leaks.

[0093] Gradual exit strategy (smooth transition control), brake release process: Release the hydraulic brake gradually at a rate of 20% brake pressure reduction every 0.5 seconds (e.g., from 100% rated pressure to 80%, 60%, 40%, and 20% in sequence). After each pressure reduction, monitor the speed rebound gradient (i.e., the speed increase rate within 0.5 seconds after release). If the rebound exceeds 50% of the current safety threshold (e.g., the threshold is 1.5 m / s 2, rebound>0.75m / s 2 ), the release is paused and the current pressure is maintained for 1 second, and then continued after the speed stabilizes. The "step-by-step" release avoids mechanical shock (such as the instantaneous separation impact of the brake pad and brake disc). At the same time, real-time feedback control prevents speed rebound and exceeds the limit, ensuring that the transport track load is smoothly decelerated to a safe state.

[0094] Torque withdrawal process: The compensation torque is decreased at a rate of 10% / second (e.g., if the current compensation torque is 5000 N·m, it decreases by 500 N·m per second) until it returns to the initial set value after the drive station switch (e.g., the theoretical torque value calculated based on the current load). During the decreasing process, the torque output waveforms of adjacent drive stations are synchronously monitored to ensure phase angle consistency to avoid shaft torsional vibration. The linear decreasing strategy eliminates the impact load on the drive chain caused by sudden torque changes (such as sudden changes in gear meshing stress), thereby ensuring the mechanical life of the drive motor and gearbox, while maintaining the stability of the coordinated operation of multiple drive stations.

[0095] Trigger conditions and threshold adjustment: If the same slope section triggers emergency braking twice or more within 24 hours, the system will automatically record the number of triggers and analyze historical data (such as speed decay gradient peak, load mass, slope angle, etc.). The safety threshold calibration factor of the interval will be increased by 15% (for example, the original threshold is 1.5m / s 2 Adjusted to 1.725m / s 2 ), and marked as a "high-risk area" on the system interface. Through self-learning of historical fault data, the safety standards of local areas are dynamically adjusted to adapt to long-term risks caused by track deformation, abnormal load distribution, etc., and avoid repeated failures.

[0096] After a high-risk area is flagged, an inspection work order is automatically sent to the operations and maintenance system, along with historical trigger records, real-time sensor data, and recommended inspection items (such as brake pad wear, drive station anchor bolt tightness, and track flatness on sloped sections). After the inspection is completed, manual confirmation is required to resolve the risk and reset the flag. Equipment anomalies are linked to geospatial information, guiding operations and maintenance personnel to conduct targeted inspections for potential hazards (such as misaligned track grade sensor installation and mechanical brake jamming), shifting from reactive maintenance to proactive prevention and reducing the probability of unplanned downtime.

[0097] Through multi-dimensional status verification of acceleration, temperature, and pressure, it is ensured that the termination operation is only executed after the system has fully recovered and stabilized; stepped brake release and linear torque retraction avoid overload damage to gears and bearings caused by sudden changes in the force system, extending the life of key components by more than 30%; automatic identification and threshold adjustment of high-risk areas enable the monitoring system to continuously evolve based on historical operating data, improving its adaptability to complex scenarios by 40%, forming a closed-loop management system of "monitoring-control-optimization".

[0098] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. The driving station status monitoring system of the gravity energy storage transport track is characterized by: include: The acquisition module is used to collect the temperature parameters and operating vibration spectrum of the main drive station in real time. When it is detected that the temperature parameter exceeds the first threshold for 5 consecutive minutes and the similarity between the vibration characteristic vector and the reference spectrum is lower than the set deviation value, a primary warning signal is generated; A calculation module is used to respond to the primary warning signal, collect the real-time oil pressure data of the gearbox of the target standby drive station, the bearing temperature rise rate curve and the dynamic torque feedback value, and calculate the availability index; A correction module is used to determine, on the transmission link of the target standby drive station, the coupling between the gearbox input end flange and the drive motor output end as the first monitoring point and the connection between the bearing seat and the output shaft as the second monitoring point, respectively calculate the axial temperature gradient difference and radial vibration energy ratio between the two monitoring points, generate a dynamic correction value using a preset coupling coefficient, and compensate the availability index based on the dynamic correction value to obtain a corrected index; a judgment module, configured to automatically cut off the control link of the current backup drive station and activate the cascade switching protocol when the corrected index is lower than a second threshold, and select a suboptimal backup drive station according to device status parameters pre-stored in the drive station topology network; The control module is used to monitor the real-time acceleration change rate of the slope section of the transport track after completing the drive station switching operation. When the speed decay gradient exceeds 120% of the safety threshold within 3 seconds, the emergency braking program of the hydraulic wedge brake device is triggered and the torque compensation mechanism of the two adjacent drive stations is activated until the speed parameters return to the safe range.

2. The driving station state monitoring system for the gravity energy storage transport track according to claim 1 is characterized in that: The temperature parameters and operating vibration spectrum of the main drive station are collected in real time. When it is detected that the temperature parameter exceeds the first threshold for 5 consecutive minutes and the similarity between the vibration characteristic vector and the reference spectrum is lower than the set deviation value, a primary warning signal is generated, including: The temperature data and vibration signals of key components of the main drive station are collected synchronously using a temperature sensor group and a vibration spectrum analyzer. The temperature parameters are filtered using a sliding window mean to eliminate transient interference noise. The eigenvector of the vibration signal is extracted, and the dynamic time warping algorithm is used to calculate the similarity of its frequency domain energy distribution with a pre-stored reference spectrum. When the temperature filtered data exceeds the first threshold for five consecutive sampling periods and the similarity calculation result of the vibration characteristic vector is lower than the set deviation value, it is determined that the main drive station has a potential risk of compound failure, and a primary warning signal is generated, including the temperature exceeding the standard range, the abnormal frequency band of the vibration spectrum, and the recommended switching priority.

3. The driving station status monitoring system for the gravity energy storage transport track according to claim 2 is characterized in that: Extract the eigenvector of the vibration signal and use the dynamic time warping algorithm to calculate the similarity of its frequency domain energy distribution with the pre-stored reference spectrum, including: The vibration signals collected from the main drive station are preprocessed. A bandpass filter is used to isolate the preset fault-sensitive frequency band, which is then divided into several equally spaced sub-bands. For each sub-band, a short-time Fourier transform is used to extract the time-frequency energy distribution. The energy integral of each sub-band per unit time is calculated, and normalization is performed to eliminate the influence of amplitude fluctuations, forming a frequency-domain energy distribution feature vector that represents the current vibration state. Based on the dynamic time warping algorithm, the current feature vector is aligned with the pre-stored standard feature vector of the benchmark spectrum in a non-rigid time series. The minimum cumulative path distance between the two sequences is calculated through dynamic programming, and the inverse of this distance value is mapped as the similarity score. The standard feature vector of the reference spectrum is generated in the following way: during the historical normal operation phase, multi-period vibration signals under different load conditions are collected, the energy integral value of each sub-band is extracted according to the same frequency band segmentation rule, and the mean and standard deviation are calculated to form a reference energy distribution template; the spectrum fluctuation characteristics of different load intervals are clustered and analyzed to generate a dynamic matching interval.

4. The driving station status monitoring system for the gravity energy storage transport track according to claim 3 is characterized in that: In response to the primary warning signal, the system collects real-time gearbox oil pressure data, bearing temperature rise rate curve, and dynamic torque feedback value of the target standby drive station and calculates the availability index, including: The gearbox oil pressure sensor acquires real-time oil pressure data of the target standby drive station at a sampling frequency of 10 times per second, and a sliding window is used to calculate the standard deviation of oil pressure fluctuations. The upper and lower limits are set based on the historical normal oil pressure fluctuation range, and the currently calculated oil pressure fluctuation standard deviation is mapped to the range of 0-1 to obtain a preliminary oil pressure status score; The bearing temperature sensor is used to continuously collect temperature data. The temperature rise rate is calculated every 5 seconds. This is the temperature change at adjacent time points divided by the time difference to generate a temperature rise rate curve. The maximum slope value within the last 3 minutes is extracted and normalized with the rated temperature rise threshold of the bearing model to obtain a normalized score for the initial oil pressure status score. Obtain dynamic torque feedback values ​​through the torque sensor, record the percentage deviation from the preset theoretical torque value, and calculate the difference between the maximum and minimum torque fluctuation amplitudes in a 10-second cycle; calculate the geometric mean of the deviation percentage and the fluctuation amplitude difference; Generate an initial usability index based on the normalized scores and geometric mean; The exponentially weighted moving average algorithm is used to integrate the historical index values ​​of the first five calculation cycles to correct the initial availability index to obtain the final availability index.

5. The driving station state monitoring system for the gravity energy storage transport track according to claim 4, characterized in that: The process for determining the revised index is as follows: High-precision temperature sensors are deployed at the first and second monitoring points to collect real-time temperature data at a frequency of once per second. The temperature data from the two monitoring points is time-aligned to eliminate phase errors caused by signal transmission delays. The instantaneous temperature difference between the two monitoring points along the drive shaft is calculated, and the axial temperature gradient difference is generated based on the temperature change rate within the last 10 minutes to indicate heat transfer anomalies in the drive chain. Three-axis vibration acceleration sensors are installed at the first and second monitoring points to simultaneously collect radial vibration signals, that is, vibration signals perpendicular to the transmission shaft. Wavelet packet decomposition is performed on the vibration signals to extract the preset fault-sensitive frequency bands, and the energy integral values ​​of each frequency band at the two monitoring points are calculated. The total energy integral value of the first monitoring point and the total energy integral value of the second monitoring point are weighted and summed according to the frequency band to generate the radial vibration energy ratio. A ratio greater than 1 indicates that the vibration energy is concentrated toward the input end, and a ratio less than 1 indicates that it is diffused toward the output end. A coupling coefficient matrix of the axial temperature gradient difference and radial vibration energy ratio is preset, and the matrix weight is obtained based on the correlation training of the two parameters in historical failure cases; According to the absolute value and change direction of the current axial temperature gradient difference, a positive gradient indicates that the heat source migrates to the output end, and a negative gradient indicates accumulation to the input end. Combined with the distribution characteristics of the vibration energy ratio, that is, aggregation or diffusion, the corresponding corrected index is matched from the coupling coefficient matrix.

6. The driving station status monitoring system for the gravity energy storage transport track according to claim 5, characterized in that: Perform wavelet packet decomposition on the vibration signal, extract the preset fault-sensitive frequency band, and calculate the energy integral value of the two monitoring points in each frequency band, including: Determine the number of wavelet packet decomposition layers based on the target transmission chain's fault characteristic frequency range, including gear meshing frequency, bearing defect characteristic frequency, and vibration signal sampling rate; The original vibration signal of each monitoring point is decomposed into wavelet packets using a certain number of layers to obtain the wavelet packet coefficients of each sub-band. The wavelet packet coefficients represent the vibration information in different frequency bands. For the preset fault-sensitive sub-band, the corresponding time domain signal is reconstructed using its wavelet packet coefficients, and the square of the reconstructed signal amplitude is integrated within the selected time window to calculate the energy integral value of the frequency band; For the energy integral values ​​that do not belong to the fault-sensitive frequency band, a baseline subtraction operation is first performed to eliminate the influence of background noise, and the energy integral values ​​of all sensitive sub-frequency bands are normalized to the range of 0-1 according to their maximum and minimum values; A weight is assigned to each sensitive sub-band according to the correlation between the fault type and the preset frequency band, and a weighted sub-band energy value is generated by multiplying the normalized energy integral value by its corresponding weight.

7. The driving station state monitoring system for the gravity energy storage transport track according to claim 6, characterized in that: The total energy integral value of the first monitoring point and the total energy integral value of the second monitoring point are weighted and summed according to the frequency band to generate a radial vibration energy ratio. A ratio greater than 1 indicates that the vibration energy is concentrated toward the input end, and a ratio less than 1 indicates that it is diffused toward the output end, including: Based on the historical fault database, the correlation between different frequency bands and specific fault types is determined, namely: the high frequency band of 2-4kHz corresponds to gear tooth surface wear, the medium frequency band of 500Hz-2kHz corresponds to bearing raceway defects, and the low frequency band below 500Hz corresponds to coupling misalignment; Based on the current operating conditions of the drive station and the potential fault types indicated in the primary warning signal, dynamic weights are assigned to each frequency band; The energy integral values ​​of each sensitive frequency band of the first monitoring point are weighted and summed according to the assigned dynamic weights. The energy of each frequency band is multiplied by the corresponding weight and then accumulated to obtain the total energy value on the input side; For the energy integral value of the same frequency band at the second monitoring point, the same weight distribution rule as that on the input side is used for weighted summation to calculate the total energy value on the output side; Deduct the preset baseline energy value from the total energy value of the input and output sides. The baseline value is the weighted sum of the energy of each frequency band when the equipment is in a no-load and fault-free state. The corrected total energy value on the input side is divided by the corrected total energy value on the output side to generate a radial vibration energy ratio. If the energy on the input side is significantly higher than that on the output side, the ratio is greater than 1; otherwise, it is less than 1.

8. The driving station status monitoring system for the gravity energy storage transport track according to claim 7, characterized in that: When the corrected index is lower than the second threshold, the control link of the current backup drive station is automatically cut off and the cascade switching protocol is activated. The next best backup drive station is selected based on the device status parameters pre-stored in the drive station topology network, including: Based on the real-time load distribution of the driver station topology network, the second threshold is dynamically adjusted. That is, if the overall network load rate is greater than 80%, the second threshold is adjusted from 0.6 to 0.7 to improve the switching sensitivity. The second threshold is compensated based on temperature and humidity. If the temperature and humidity exceed the upper limit of the device's rated operating range, the threshold is increased by 10%; Before sending a shutdown command to the current standby drive station, its operating parameters are synchronously backed up to the central database; Sending a shutdown command simultaneously through the primary and backup communication links to ensure that at least one link is successfully executed; After cutting off the control link, the current decay curve of the drive motor is monitored in real time. The mechanical brake lock is released only after confirming that the motor has completely stopped. The pre-stored drive station topology map is read to identify the geographical location and connection status of all backup drive stations. The priority scores of the remaining candidate backup stations are calculated, an activation command is sent to the suboptimal backup drive station, and the operating parameters of the faulty station are synchronized to linearly increase the output torque of the suboptimal station from 0 to the target value within 3 seconds.

9. The driving station status monitoring system for the gravity energy storage transport track according to claim 8, characterized in that: After the drive station switching operation is completed, the real-time acceleration change rate of the transport track slope section is monitored. When the speed decay gradient exceeds 120% of the safety threshold within 3 seconds, the emergency braking program of the hydraulic wedge brake device is triggered and the torque compensation mechanism of the two adjacent drive stations is activated until the speed parameters return to the safe range, including: Real-time collection of acceleration data from the transport track, sliding window mean filtering of the raw acceleration data, and extraction of the smoothed acceleration change rate, i.e., the derivative of acceleration per unit time; The safety threshold is dynamically calculated based on the slope angle and current load mass: For every 1° increase in inclination, the safety threshold increases by 5%; For every 10% increase in load mass over the rated value, the safety threshold increases by 3%; Based on historical operating data, a 10% redundancy margin is applied to the safety threshold; In a 3-second time window, the moving average of the acceleration change rate is calculated at intervals of 0.5 seconds to generate a velocity decay gradient curve; Perform linear fitting on the gradient curve and calculate the absolute value of its slope. If the slope of three consecutive data points exceeds 120% of the dynamic safety threshold, it is determined to be an abnormal attenuation event. Multi-stage braking strategy that triggers the hydraulic wedge brake: Level 1 braking: applies braking force at 50% of the rated pressure for 1 second to reduce the risk of mechanical shock; Secondary braking: If the gradient does not fall within the threshold after primary braking, the pressure is increased to 80% of the rated value within 2 seconds, and the brake disc temperature is monitored in real time; Level 3 braking: If the level 2 braking is still ineffective, full-pressure braking is triggered and the brake pad spray cooling system is started.

10. The driving station state monitoring system for the gravity energy storage transport track according to claim 9, characterized in that: When the speed value returns to the safe range, braking and compensation are terminated, including: When the speed value returns to the safe range, braking and compensation can be terminated only when the following conditions are met at the same time: The sliding window standard deviation of the acceleration rate of change is less than 1.2 times the baseline value; The temperature gradient difference between adjacent drive stations is less than 5℃ / min; Hydraulic brake pressure fluctuation <10% of rated value; Gradual Exit Strategy: Brake release: Gradually release the hydraulic brake at a rate of 20% pressure reduction every 0.5 seconds, while monitoring the speed rebound gradient. If the rebound exceeds the threshold of 50%, the release is suspended; Torque withdrawal: The compensation torque is reduced at a rate of 10% / second until it returns to the initial setting value after switching. If emergency braking is triggered twice or more on the same slope within 24 hours, the calibration coefficient of the corresponding safety threshold will be automatically increased by 15%, and the area will be marked as a high-risk area, triggering manual inspection.

Images