A method and system for monitoring the sealing condition of a slip-on loader bucket

Abstract

The application discloses a kind of slip loader bucket sealing state monitoring method and system, belong to engineering machinery hydraulic system state monitoring field, to solve bucket and hydraulic connection position in silt, dust, impact load and frequent action condition early leakage difficult identification, fault source difficult positioning and life state difficult stable updating problem, the present application is by collecting turnover oil cylinder multi-source operation signal, according to natural circulation phase constructs transmission time difference impact window, establishes two cavity continuity residual outside window and distributes leakage conductance according to hydraulic topology, then with particle difference, acoustic emission and maintenance event constraint distinguishes pollution invasion and wear debris, realizes the technical effect of separating impact transient, locating sealing failure and outputting maintenance window under non-stop condition.

Description

Technical Field

[0001] This invention relates to the field of hydraulic system condition monitoring in construction machinery, and in particular to a method and system for monitoring the sealing condition of a skid steer loader bucket. Background Technology

[0002] During the loading, lifting, tipping, unloading, and recovery processes of skid steer loaders, the bucket tipping cylinder, its seals, hoses, hydraulic joints, and control valves are subjected to long-term exposure to mud and dust, sudden load changes, structural impacts, and frequent reversals. Existing maintenance typically relies on visual oil stains, pressure drops, sluggish movements, or manual experience to judge the sealing condition. Some methods also use individual monitoring of cylinder pressure, stroke, oil temperature, or particulate contamination levels. These signals change under natural operating conditions with load, bucket posture, valve opening, and oil temperature, and are superimposed with the propagation transients caused by bucket collisions, making early leakage characteristics easily masked by operating condition fluctuations and impact responses.

[0003] Existing methods struggle to isolate and locate internal piston seal leaks, external rod and dust scraper seal leaks or contamination intrusion, leaks from hoses and detachable hydraulic joints, and control valve mid-position leaks. Furthermore, relying solely on particle quantity makes it difficult to distinguish between external mud and sand intrusion and internal seal wear debris. On the other hand, cross-cycle condition estimations, without constraints on comparability of operating conditions and maintenance events, are prone to problems such as unsubstantiated wear condition improvements, baseline reuse of failure data, or maintenance window drift.

[0004] Therefore, there is a need for a method and system for monitoring the sealing status of skid steer loader buckets that can overcome the shortcomings of the existing technologies. Summary of the Invention

[0005] One objective of this invention is to propose a method and system for monitoring the sealing status of a skid steer loader bucket. Addressing the problems of existing technologies, such as difficulty in separating transient impact from actual leakage during natural operation, difficulty in locating multiple types of seals and hydraulic leakage channels, difficulty in distinguishing contamination sources, and the tendency for wear status data to erroneously regress before and after maintenance, this invention proposes a technical solution that simultaneously acquires pressure, motion, valve control, oil temperature, and impact reference signals from both chambers and divides the operation phase. An impact influence window is formed using the impact initiation point and propagation time difference, allowing data outside the window to be included in the calculation of the continuity residual between the two chambers, while data inside the window is accumulated only for impact damage. Under the constraints of flow conservation, non-negative conductivity, and observability, the equivalent leakage conductivity of each channel is allocated. Then, particle differential, rod-end acoustic emission, and conductivity are performed for time-series consistency calculations, and the wear status is updated in conjunction with comparable operating cycles and maintenance event constraints. This invention achieves the technical effects of suppressing false impact alarms, locating fault channels, distinguishing contamination sources, and stabilizing the output confidence level and maintenance window without stopping the machine.

[0006] This invention provides a method for monitoring the sealing status of a skid steer loader bucket, comprising the following steps: S1. Synchronously collect the pressure, motion parameters, directional valve control parameters, oil temperature and impact reference signal of the two chambers of the bucket tilting cylinder. Based on the directional valve control direction, piston movement direction, bucket posture and pressure changes, divide the natural operation cycle and generate operation samples with operation phase identifiers. S2. Determine the impact starting point based on the impact reference signal, and determine the impact influence window based on the propagation time difference from the bucket impact source to each measuring point. Determine the data outside the window of the work sample as the estimated data outside the window, and accumulate the data inside the window to obtain the impact damage amount. Do not use the data inside the window as direct evidence of leakage triggering. S3. Using out-of-window estimation data, construct the two-chamber continuity residual based on valve orifice flow rate, effective volume of the two chambers, oil volumetric elastic modulus, differential pressure of the two chambers, and piston speed. S4. Input the two-chamber continuity residual into the hydraulic topology model. The hydraulic topology model includes the internal leakage channel of the piston seal, the external leakage or intrusion channel of the rod seal and dust scraper seal, the external leakage channel of the hose and detachable hydraulic joint, and the leakage channel in the control valve center position. Under the constraints of flow conservation, non-negative leakage conductance, and observable relationship of the measuring point, the two-chamber continuity residual is allocated to obtain the equivalent leakage conductance of each channel. S5. Extract the particle differential characteristics and rod end acoustic emission characteristics between the bucket branch and the oil supply branch. Perform time-series consistency calculations on the equivalent leakage conductivity of each channel with the particle differential characteristics and rod end acoustic emission characteristics to obtain the probability of external contamination intrusion and the probability of internal wear debris, and form the topology observation factor of the corresponding sealed channel. S6. Based on load, oil temperature, bucket posture and working phase, select comparable working cycles, update the wear status of each channel across cycles according to equivalent leakage conductance, topology observation factor, impact damage amount and maintenance events, and output fault location, wear or leakage level, confidence level and maintenance window.

[0007] Optionally, S1 includes: the motion parameter is at least one of the cylinder stroke and the bucket angle, and the directional valve control parameter is at least one of the directional valve command and the valve core displacement; The phase discrimination conditions are composed of the directional state of the reversing valve control parameters, the sign of the piston speed, the attitude range of the bucket angle, and the direction of pressure change in the two chambers. According to the phase discrimination conditions, the natural operation cycle is marked as the loading contact phase, lifting phase, steady-state holding phase, overturning unloading phase, recovery phase and reversing phase, and the start and end times of the phase, load characterization quantity, oil temperature and bucket posture are written into the operation sample.

[0008] Optionally, S2 includes: obtaining the impact arrival time using an impact reference sensor installed in the bucket, cylinder body, sealing seat, and pressure measuring point, and using the impact arrival time at the bucket as the impact starting point; The propagation time difference is determined based on the propagation path and corresponding propagation speed between each measuring point and the impact source of the bucket. The impact influence window of each measuring point is generated by the impact starting point, the propagation time difference, and the window width determined by the response duration of the non-leakage impact sample. The pressure and acoustic emission signals within the impact influence window are used only to accumulate the impact damage amount and are not used as evidence of leakage triggering. The data outside the impact influence window are used as the window estimation data in step S3.

[0009] Optionally, S3 includes: collecting the oil supply pressure and return pressure respectively by pressure sensors installed in the main oil supply line and the main oil return line; selecting the oil supply pressure or return pressure as the pressure on one side of the valve port according to the valve control direction and using the corresponding cylinder chamber pressure as the pressure on the other side of the valve port to obtain the pressure difference before and after the valve port. The flow rate at the valve orifice of the two chambers is determined based on the valve core displacement, the pressure difference across the valve orifice, and the flow coefficient corresponding to the oil temperature. The effective volume of the two chambers is determined based on the initial chamber volume of the cylinder, the effective area of ​​the piston, and the cylinder stroke, and the bulk modulus of the oil is corrected based on the oil temperature. The pressure signal is processed by bandwidth-limited differentiation to obtain the two-chamber pressure differential, and the cylinder stroke is processed by time difference to obtain the piston speed. For each chamber, the difference between the valve orifice flow rate and the volumetric flow rate caused by piston movement and the compressive flow rate caused by pressure change is used as the continuity residual of the corresponding chamber.

[0010] Optionally, S4 includes: determining the flow direction of the piston seal internal leakage channel by the pressure difference between the two chambers; determining the flow direction of the rod seal and dust scraper seal external leakage or intrusion channel by the pressure difference between the rod chamber pressure and the external pressure; determining the flow direction of the hose and detachable hydraulic joint external leakage channel by the pressure difference between the internal pressure and the external pressure of the pipe section where the hose or detachable hydraulic joint is located; wherein the internal pressure is the cylinder chamber pressure, supply pressure or return pressure connected to the pipe section; and determining the conduction state of the control valve mid-position leakage channel by the directional valve mid-position command, supply pressure, return pressure and corresponding cylinder chamber pressure. Establish a mapping between the equivalent leakage flow rate of each channel and the corresponding pressure difference and equivalent leakage conductivity, and form a flow conservation equation by combining the continuity residuals of the two chambers; Among the two-chamber pressure measuring points, the supply pressure measuring point, and the return pressure measuring point, the measuring points connected to the pipeline in the conducting state under the current operating phase are determined as available pressure measuring points. Based on the available pressure measuring points and the valve control state, an observable relationship is generated. For indistinguishable channels, the combined conductance is output. For distinguishable channels, the equivalent leakage conductance is solved under the constraint of non-negative equivalent leakage conductance.

[0011] Optionally, S5 includes: establishing a healthy baseline using data from multiple natural work cycles that have been confirmed to be in good working order, have not triggered leakage due to continuous residuals, and whose load, oil temperature, bucket attitude, and work phase meet the same baseline acquisition conditions; After the corresponding health baseline becomes invalid due to seal replacement, joint tightening, oil purification, or sensor configuration changes, the corresponding health baseline shall be discontinued and reconstructed with newly acquired data when the baseline acquisition conditions are verified to be met after maintenance. A first particle counting unit is installed in the return oil bypass of the bucket branch, and a second particle counting unit is installed in the oil supply bypass downstream of the main filter. After a dust or sediment exposure event is recorded, the period during which the reversing valve control parameters continuously indicate piston rod retraction, the piston speed is within the admission range determined by the healthy baseline retraction speed distribution, and the rod chamber pressure change rate does not exceed the admission threshold determined by the healthy baseline pressure change rate distribution is defined as the rod retraction admission period. During the rod retraction access period, the first particle counting unit and the second particle counting unit are controlled to obtain equal sampling volumes, and the number of particles is counted according to the same particle size range. The particle differential feature is obtained by subtracting the corresponding number of particles of the second particle counting unit from the number of particles of the first particle counting unit by particle size. Furthermore, the acoustic emission characteristics at the rod end are obtained from the envelope of the output signal of the acoustic emission sensor disposed at the piston rod end seal. The degree of matching of dust or sediment exposure events, changes in particle differential characteristics, changes in rod end acoustic emission characteristics, and changes in the equivalent leakage conductivity of rod seal and dust scraper seal leakage or intrusion channels within the consistency time limit determined by the synchronous sampling error of the health baseline is defined as the first degree of matching. The degree of matching between the particle differential characteristic changes that do not match the dust or sediment exposure event within the consistency time limit and the equivalent leakage conductance changes of the piston seal internal leakage channel and the rod seal and dust scraper seal external leakage or intrusion channel is determined as the second degree of matching. Normalize the first and second matching degrees respectively to obtain the probability of external contamination intrusion and the probability of internal wear debris. Based on the observable relationship generated in step S4, the external contamination intrusion probability is used to weight the leakage or intrusion channels of the rod seal and dust scraper seal, and the internal wear debris probability is used to weight the internal leakage channel of the piston seal and the external leakage or intrusion channels of the rod seal and dust scraper seal, thus forming the topological observation factor.

[0012] Optionally, S6 includes: determining the load characterization quantity based on the pressure difference between the two chambers and the effective area of ​​the piston, and determining the phase duration based on the difference between the start and end times of the working phase; The corresponding comparable deviation thresholds are determined by the distribution of load characteristics, oil temperature, bucket posture and duration of each operation phase in the health baseline. Operation cycles with the same operation phase and whose differences in load characteristics, oil temperature, bucket posture and duration of each operation phase do not exceed the corresponding comparable deviation thresholds are grouped into the same comparable cycle group. Within each comparable cycle group, the wear status of each channel is updated based on the direction of change of equivalent leakage conductance, topology observation factor, and cumulative impact damage over multiple consecutive operating cycles. The confidence level is determined based on the fitting error of the continuity residual, the number of observable channels in the current working phase, and the consistency of conduction across multiple working cycles. The wear or leakage level is determined based on the quantile boundaries of the equivalent leakage conduction of each channel in the health baseline. Furthermore, when no maintenance event is received, the wear status update constraint for each channel is set to remain unchanged or change in the direction of increasing wear; Upon receiving a seal replacement event and passing post-maintenance verification, the wear state of the channel where the replaced seal is located is reset and the corresponding health baseline is invalidated. Upon receiving a joint tightening event and passing post-maintenance verification, the wear state of the leakage channel of the hose and the detachable hydraulic joint is reset and the corresponding health baseline is invalidated. Upon receiving an oil purification event, only the health baseline corresponding to the particle differential feature is invalidated without resetting the wear state of any channel, and the invalidated health baseline is reconstructed according to the health baseline reconstruction method. Each channel is assigned a maintenance threshold, which is determined by the smaller of the lower bound of the wear or leakage level that the corresponding channel needs to maintain and the allowable leakage limit given by the equipment manufacturer, which are converted into the wear state boundary. The number of operation cycles required to reach the corresponding maintenance threshold is predicted based on the cross-cycle change rate of wear status of each channel, and the number of operation cycles is converted into a maintenance window.

[0013] On the other hand, the present invention also provides a skid steer loader bucket sealing status monitoring system, comprising: The data acquisition and phase division module includes a two-chamber pressure sensor, an oil supply pressure sensor, an oil return pressure sensor, a motion parameter sensor, a directional valve control parameter acquisition unit, an oil temperature sensor, and an impact reference sensor, used to generate operation samples with operation phase identifiers; the impact window module is used to determine the impact influence window and generate estimated data outside the window and impact damage amount; the residual generation module is used to generate two-chamber continuity residuals based on the estimated data outside the window; the topology allocation module is used to allocate the two-chamber continuity residuals in the hydraulic topology model and obtain the equivalent leakage conductivity of each channel; the pollution source identification module includes a first particle counting unit set in the return oil bypass of the bucket branch, a second particle counting unit set in the oil supply bypass downstream of the main filter, and an acoustic emission sensor set in the piston rod end seal, used to generate the probability of external pollution intrusion, the probability of internal wear debris, and the topology observation factor based on the equivalent leakage conductivity of each channel, particle differential characteristics, and rod end acoustic emission characteristics; the status update output module is used to filter comparable operation cycles, update the wear status of each channel across cycles, and output the fault location, wear or leakage level, confidence level, and maintenance window.

[0014] The beneficial effects of this invention are: 1. By establishing impact influence windows for each measuring point according to the impact initiation point and propagation time difference, the signal inside the window only forms the impact damage amount and does not directly trigger leakage. Furthermore, the two-cavity continuity residual is constructed only with the data outside the window, which can reduce the interference of shovel collision and overturning impact on early leakage judgment.

[0015] 2. By allocating residuals in a hydraulic topology model that includes piston seals, rod seals and dust scraper seals, hoses and detachable hydraulic joints and control valve leakage channels, based on flow conservation, non-negative conduction, and observable relationship of measurement points, single-channel or combined conduction can be output when observable conditions change, thereby forming a traceable fault location judgment.

[0016] 3. By comparing the temporal consistency of branch and oil supply particle differential, rod end acoustic emission and leakage conductivity, external contamination intrusion and internal wear debris can be distinguished. Maintenance events limit the unfounded improvement of wear status and trigger the corresponding health baseline reconstruction, which can improve the stability of cross-cycle status trend and maintenance window output. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the skid steer loader bucket sealing status monitoring method of the present invention.

[0018] Figure 2 This is a flowchart of the S4 hydraulic topology allocation process of the present invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figures 1-2 A method for monitoring the sealing status of a skid steer loader bucket includes the following steps: S1. Synchronously collect the pressure, motion parameters, directional valve control parameters, oil temperature and impact reference signal of the two chambers of the bucket tilting cylinder. Based on the directional valve control direction, piston movement direction, bucket posture and pressure changes, divide the natural operation cycle and generate operation samples with operation phase identifiers. S2. Determine the impact starting point based on the impact reference signal, and determine the impact influence window based on the propagation time difference from the bucket impact source to each measuring point. Determine the data outside the window of the work sample as the estimated data outside the window, and accumulate the data inside the window to obtain the impact damage amount. Do not use the data inside the window as direct evidence of leakage triggering. S3. Using out-of-window estimation data, construct the two-chamber continuity residual based on valve orifice flow rate, effective volume of the two chambers, oil volumetric elastic modulus, differential pressure of the two chambers, and piston speed. S4. Input the two-chamber continuity residual into the hydraulic topology model. The hydraulic topology model includes the internal leakage channel of the piston seal, the external leakage or intrusion channel of the rod seal and dust scraper seal, the external leakage channel of the hose and detachable hydraulic joint, and the leakage channel in the control valve center position. Under the constraints of flow conservation, non-negative leakage conductance, and observable relationship of the measuring point, the two-chamber continuity residual is allocated to obtain the equivalent leakage conductance of each channel. S5. Extract the particle differential characteristics and rod end acoustic emission characteristics between the bucket branch and the oil supply branch. Perform time-series consistency calculations on the equivalent leakage conductivity of each channel with the particle differential characteristics and rod end acoustic emission characteristics to obtain the probability of external contamination intrusion and the probability of internal wear debris, and form the topology observation factor of the corresponding sealed channel. S6. Based on load, oil temperature, bucket posture and working phase, select comparable working cycles, update the wear status of each channel across cycles according to equivalent leakage conductance, topology observation factor, impact damage amount and maintenance events, and output fault location, wear or leakage level, confidence level and maintenance window.

[0021] In this specific embodiment, S1 includes: In this specific embodiment, a skid steer loader with a rated load of 750kg, a tilting cylinder diameter of 80mm, and a piston rod diameter of 45mm is used as the monitoring object. During natural loading operations without dedicated shutdown and pressure maintenance, the on-board monitoring controller samples the rodless chamber pressure, rod chamber pressure, oil supply pressure, return oil pressure, cylinder stroke, bucket angle, directional valve command, valve core displacement, and oil temperature at 200Hz. The impact reference channels arranged on the bucket ear plate, cylinder body, and piston rod end seal are sampled by an independent high-speed front end at a rate of not less than 20kS / s, and the original acoustic emission channel at the rod end is sampled at a rate of not less than 250kS / s. The low-frequency channel and the two types of high-speed channels are also included. All components are locked to the same hardware clock and receive the same trigger pulse. After completing anti-aliasing filtering, bandpass filtering, envelope filtering, threshold arrival time determination, and feature extraction locally, the high-speed front end outputs a 200Hz feature frame to the vehicle monitoring controller. The propagation time difference is estimated using a high-speed timestamp with a resolution of no more than 50μs instead of a 200Hz feature frame. Each sampling record carries the device number, natural cycle number, sensor number, controller monotonic clock timestamp, and maintenance baseline version number. Pressure is stored in Pa, stroke in m, angle in rad, and oil temperature in degrees Celsius. The maximum timestamp error across the front end after common clock correction is limited to within 0.5ms. The data acquisition and phase division module performs zero-point compensation, range check, and three-point median deburring for each channel. The pressure channel uses a fourth-order low-pass filter with a cutoff frequency of 25Hz, and the stroke and angle channels use a fourth-order low-pass filter with a cutoff frequency of 15Hz. When a channel is continuously missing for more than 25ms, the pressure value exceeds the calibrated range of 0 to 35MPa, or the stroke value exceeds the mechanical limit of 0 to 0.42m, the natural cycle is marked as invalid and the leakage calculation is stopped. Single-point missing values ​​not exceeding 25ms are obtained by linear interpolation of the effective values ​​before and after. The processed data with the same timestamp is written to the original work record cache. The controller follows the formula Calculate the piston speed and follow the formula. and Calculate the direction of pressure change in both chambers, where the sign... Indicates the first The piston velocity at each sampling time point, in m / s, with positive values ​​indicating piston rod extension; symbol and symbols They represent the first The and the first The cylinder stroke at each sampling moment, in meters, is derived from a stroke sensor; symbol This represents the time interval between adjacent sampling moments, which is 0.005s in this specific embodiment; symbol and symbols They represent the first Discrete pressure change rates in the rodless and rod chambers at each sampling time, in Pa / s; symbol ,symbol ,symbol and symbols This represents the filter pressure of the corresponding cavity at the corresponding sampling time, in Pa; symbol This indicates the sampling sequence number that increments over time within the natural cycle; The phase division module uses the directional valve command and valve core displacement direction, piston speed sign, bucket angle attitude range, and pressure change direction of the two chambers to form phase discrimination conditions. When the absolute value of valve core displacement is not less than 8% of the rated stroke and the direction is maintained for not less than 50ms, the valve control direction is confirmed. When the absolute value of piston speed is less than 0.003m / s and lasts for not less than 100ms, the near-stationary state is confirmed. The bucket angle is divided into ground attitude boundary, transport attitude boundary, and unloading attitude boundary obtained by no-load calibration. The pressure change direction is determined by the positive or negative of the median slope of five consecutive sampled values. The threshold is stored in the phase discrimination parameter table and calibrated by no-load reciprocating action and rated load test. The controller executes phase transitions according to the state machine: when the bucket is in the grounded position, the valve command points to retraction, and the pressure in the rodless chamber rises, it enters the loading contact phase; when the bucket leaves the grounded position, the piston rod extends, and the load indicator is stable and positive, it enters the lifting phase; when the valve core returns to the neutral position, the piston is approximately stationary, and the pressure in both chambers is maintained, it enters the steady-state holding phase; when the bucket enters the unloading position and the valve command points to extension, it enters the tipping unloading phase; when the valve command points to retraction and the bucket angle changes towards the transport position, it enters the recovery phase; the 100ms transition period after the valve control direction reverses enters the reversing phase. If multiple conditions are met simultaneously, a unique label is assigned according to the priority order of the reversing phase, loading contact phase, tipping unloading phase, lifting phase, recovery phase, and steady-state holding phase. Each time the recovery phase re-enters the shovel contact phase, a new natural cycle number is generated. Any phase is only written to the phase start point after the discrimination condition is met for five consecutive sampling points, and is written to the phase end point when the next phase is confirmed. For isolated tag segments with a duration of less than 50ms, the phase with consistent tags before and after is merged. If the tags before and after are inconsistent, the previous phase in the state machine's allowed transition relationship is merged to suppress phase fragmentation caused by valve core jitter. After phase division is completed, the controller generates operation samples with operation phase identifiers. The sample fields include natural cycle number, phase label, phase start and end time, each original and filtered signal sequence, load characterization quantity obtained from the pressure difference between the two chambers and the effective area of ​​the piston, median oil temperature in the phase, median bucket angle, valve control direction, acquisition quality mark and maintenance baseline version number. The samples are written into the operation sample library indexed by natural cycle number for step S2 to establish the impact influence window and for step S6 to perform condition comparability screening.

[0022] In this specific embodiment, S2 includes: The impact window module reads the operation sample generated in step S1 and the original impact reference signal on the same hardware clock time axis obtained by the high-speed front end with a speed of not less than 20kS / s. It calculates the Hilbert envelope of the bandpass signal of the bucket ear plate sensor in the range of 1kHz to 8kHz, and uses the median of the leak-free loading sample envelope plus six times the median absolute deviation as the impact starting threshold. When the high-speed envelope exceeds the threshold for three consecutive sampling points and a local peak appears within 20ms thereafter, the high-speed timestamp of the first threshold crossing is determined as the impact starting point at the bucket. The high-speed front end merges the envelope peak, the starting timestamp, and the quality flag into the 200Hz feature frame, but the propagation time difference and the window boundary are always calculated based on the high-speed timestamp. The natural cycle that does not detect the threshold crossing event does not generate an impact influence window and uses all valid data as candidate data outside the window. For the cylinder body, piston rod end seal, and various pressure measuring points, the module follows the formula. Determine the time difference of shock propagation, where the sign... This indicates that the impact source of the bucket has spread to the first... The propagation time difference at each measuring point, in seconds; symbol This indicates the distance from the impact source along the bucket connecting rod, pin, cylinder, and hydraulic lines to the [missing information]. The equivalent propagation path length for each measuring point, in meters, is derived from equipment assembly dimensions and impact calibration; symbol Represents the equivalent propagation speed along the corresponding propagation path, in m / s, calibrated by the median of path length divided by time difference in ten instrumented impact tests conducted on equipment confirmed to be in good maintenance condition and leak-free; symbol Indicates the serial number of the impact reference sensor or pressure measuring point; The impact influence window for each measuring point is calculated according to the formula. Generate, where, symbol Indicates the first The impact time interval of each measuring point; symbol This indicates the impact threshold at which the impact envelope of the bucket ear plate first crosses the threshold, measured in seconds (s); symbol Indicates the first The propagation time difference at each measuring point, in seconds; symbol and symbols These represent the pre- and post-protection margins for coverage synchronization error and propagation time difference dispersion, respectively, in seconds (s). In this specific embodiment, they are taken as the sum of three times the standard deviation of the arrival time difference at the measurement point and 0.5 ms; (symbol) Indicates that the leak-free impact sample is in the first The response duration of each measurement point is measured in seconds. The 95th percentile of the duration when the envelope falls back to the baseline median plus three times the median absolute deviation and lasts for 5 ms is taken. When the interval between adjacent impact windows within the same natural cycle is less than 10ms, the impact window module merges the adjacent intervals; when the merged window length exceeds 35% of the duration of the corresponding working phase, the phase is marked as having excessive impact contamination, and only the impact damage amount is retained without generating leakage estimation input for that phase; for pressure, valve control, stroke, and oil temperature sampling points not covered by the impact window, the module retains the original timestamp and phase label and writes them into the estimated data record outside the window, and uses local cubic splines to generate only the pressure estimation value required for continuous calculation for the gaps 20ms before and after the window edge, while writing the interpolation flag, and the interpolation value does not participate in the health baseline update; The module measures the in-window impact envelope at each measuring point according to the formula. The total impact damage from this natural cycle is calculated, where the symbol is... Indicates as of the date Dimensionless cumulative impact damage over one natural cycle; symbol Represents the dimensionless cumulative impact damage amount of the previous natural cycle; symbol Indicates the natural cycle number sorted by completion time; symbol Indicates the first The dimensionless damage weight for each measuring point is derived from the regression coefficient of the response energy of that measuring point to the impact load of the pin and the sealing seat in the leak-free impact test, and normalized so that the sum of all weights is 1; (symbol) Indicates the sampling sequence number that falls into the impact window; symbol Indicates the first The impact influence window for each measuring point follows the definition defined above in this step; symbol Indicates the first The measuring point at the ... The envelope amplitude at each sampling time point, in V; symbol This represents the median of the non-impact baseline envelope at the measurement point, in units of V; symbol This represents the calibration scale for the envelope amplitude of the leakage-free impact sample at this measuring point, in units of V with a lower limit of 1 mV; symbol Indicates the adjacent sampling interval of the high-speed envelope, in seconds; symbol This represents the reference time used to make the time integral dimensionless, in seconds (s). In this specific implementation, it is taken as 1 second and saved with the impact damage calibration version; function symbol. This indicates that the normalized shock amplitude will be limited to 0 to 10; This step does not use the trained recognition model, but instead uses a deterministic impact window model consisting of an envelope threshold detector, a propagation time difference calibration table, response duration quantiles, and an interval merger. The model input fields are the measurement point number, impact envelope, timestamp, propagation path length, propagation speed, and leakage-free response duration. According to the aforementioned threshold crossing rules and window formula, the start and end times of the window for each measurement point, the window merging relationship, and the excessive impact contamination indicator are directly output. When the calibration table version changes or the impact sensor position changes, the knock calibration is re-executed and the model version is updated. The impact window module writes an in-window flag, an out-of-window flag, an impact event number, a measurement point number, and an interpolation flag to each sampling point. The evidence admission rule for the leakage trigger is fixed to only accept continuous residuals where the out-of-window flag is true and the interpolation flag is false. In-window data is not used as direct evidence of leakage triggering. Any pressure peak, acoustic emission peak, and instantaneous residual formed therefrom within the window must not directly change the state of the leakage channel. The signal within the window only updates the cumulative impact damage amount. After processing, the module writes the out-of-window estimated data object containing the natural cycle number, phase label, effective out-of-window sampling sequence, window boundary of each measuring point, excessive impact contamination indicator, and cumulative impact damage into the impact window cache. Step S3 reads only the effective out-of-window sequence from this cache, and step S6 reads the cumulative impact damage and uses it as the impact load item in the wear state update.

[0023] In this specific embodiment, S3 includes: The residual generation module reads the estimated data outside the window from the impact window cache in step S2, and checks the impact arrival time and measurement point number obtained by the impact reference sensor set in the bucket, cylinder body, sealing seat and pressure measuring point. It confirms that the impact arrival time at the bucket is used as the impact starting point in the cache and that the current pressure sampling point is not in any impact influence window. If the impact metadata is missing, the window version is inconsistent, or the pressure sampling point falls into the impact influence window, the point is rejected from entering the residual calculation. The module assembles the supply pressure, return pressure, and corresponding cylinder chamber pressure into a pressure candidate set on both sides of the valve port. It selects the upstream and downstream pressures of the valve port according to the valve control direction. When the valve command points to the piston rod extension, the supply pressure and rodless chamber pressure form the rodless chamber valve port pressure difference, and the rod chamber pressure and return pressure form the rod chamber valve port pressure difference. When the valve command points to the retraction, the supply and return oil side relationships are exchanged. In the mid-position phase, the valve port flow is set to zero, and the supply, return, and two chamber pressures required for the mid-position leakage channel are retained. When the reversing valve command direction and the valve core displacement direction are inconsistent for 50ms, the window is marked as a valve control state conflict and the valve port flow is not output. When the two are consistent, the actual valve core displacement direction is used as the primary selection criterion. When the direction state and absolute pressure difference of the two pressure candidates are the same, the pressure record with the timestamp closer to the current valve core displacement sampling value is used as the parallel decision rule. If a unique selection still cannot be made, the window is set to uncertain pressure difference source and the residual output is stopped. The module is based on the formula Determine the first The flow rate at each chamber-side valve port, where the symbol is... Indicates the first The cylinder chamber is in the first... The valve orifice flow rate at each sampling time, in m³ / s. 3 / s, the flow into this cavity is positive; sign Indicates oil temperature Next Dimensionless flow coefficient of each valve port; symbol Indicates the first Oil temperature at each sampling time, in degrees Celsius; symbol Indicates the first The area gradient of each valve port, in meters; symbol This represents the valve core displacement after zero-position compensation, in meters (m); symbol This represents the difference between the upstream and downstream pressures at the valve orifice, measured in Pa, according to the valve control direction. This indicates the oil density at the corresponding oil temperature, in kg / m³. 3 ;symbol Pick or And respectively represent rodless cavity or rod cavity; symbol Indicates the valid sampling sequence number outside the window; function symbol This represents a function indicating the direction of the pressure difference. The flow coefficient, oil density, and bulk modulus are given by the oil temperature correction parameter table. This table uses oil grade, temperature compartment, and valve port direction as the joint key. Fields include flow coefficient, density, bulk modulus, effective temperature range, calibration date, and version number. One implementation record in the table is: oil grade 46 high viscosity index anti-wear hydraulic oil, temperature 40 degrees Celsius, rodless chamber inlet direction, flow coefficient 0.68, and density 860 kg / m³.3 The bulk elastic modulus is 1.45 GPa. The parameters are fitted by flow test bench and constant volume compression test. When the temperature exceeds the effective range of 20 to 80 degrees Celsius, the residual calculation is stopped and a temperature over-limit flag is output. The effective volume of the two cavities is calculated according to the formula. and Determined, where the symbol and symbols They represent the first The effective volume of the rodless cavity and the rod cavity at each sampling time is in cubic meters. 3 ;symbol and symbols These represent the initial volumes of the corresponding cavities and connecting short tubes when the stroke is at zero position, in cubic meters (m³). 3 Derived from manufacturing dimensions and fluid injection calibration; symbol This represents the effective area of ​​the rodless chamber piston, in meters (m²). 2 ;symbol This represents the effective annular area of ​​the rod cavity, in meters. 2 ;symbol Indicates the first The cylinder stroke at each sampling time, in meters; symbol Indicates the total effective stroke of the hydraulic cylinder, in meters (m); symbol Indicates the valid sampling sequence number outside the window; To avoid amplifying high-frequency noise at the edge of the impact window through differentiation, the module performs bandwidth-limited differentiation on the pressure signal to obtain the pressure differential for both chambers. A first-order bandwidth-limited differentiator with a cutoff frequency of 12Hz is used, according to the formula... Generate the pressure differential, where the sign is... and symbols Indicates the first The rate of pressure change of each chamber at the current and previous effective sampling times outside the window, in Pa / s; symbol This represents a dimensionless smoothing coefficient determined by the sampling period and a 12Hz cutoff frequency, limited to between 0 and 1; symbol and symbols This represents the cavity pressure at the corresponding sampling time, in Pa; symbol Represents the time difference between adjacent valid sampling times, in seconds; symbol Pick or ;symbol This indicates the valid sampling sequence number outside the window; when the interval between adjacent valid points exceeds 25ms, the differentiator is reinitialized and the residual of that point is not output. The piston speed is resampled at the effective point outside the window by the stroke difference value from step S1, and the residual generation module generates the residual according to the formula. and Construct a two-cavity continuity residual, where the sign... and symbols These represent the rodless cavity and the rod cavity at the 1st... The flow rate not explained by normal valve orifice flow, piston sweep flow, and oil compression flow at each effective sampling time outside the window, in m³ / s. 3 / s; symbol and symbols These represent the flow rates at the two valve orifices, in units of m³. 3 / s; symbol and symbols These represent the effective areas of the two cavities, in meters (m²). 2 ;symbol Indicates piston speed, unit is m / s, with extension being positive; symbol and symbols These represent the effective volumes of the two cavities, in cubic meters (m³). 3 ;symbol This represents the effective bulk modulus of elasticity of the oil, obtained from the oil temperature correction parameter table, in Pa; symbol and symbols These represent the pressure change rates of the two chambers, in Pa / s; symbols Indicates the valid sampling sequence number outside the window; The module calculates the median and median absolute deviation of the continuous residuals of the two chambers using a 50ms sliding window. If there are fewer than six valid samples in the same window, the valve core displacement is at the edge of the zero dead zone, or the sensor quality mark is abnormal, the window is marked as low confidence and is not sent to the single channel solution. The normal window forms a residual observation record including the natural cycle number, phase label, window time range, median of the residuals of the two chambers, residual dispersion, valve control direction, pressure of the two chambers, supply and return oil pressure, and number of valid samples. The residual observation records, along with the reference keys for the impact window boundary and the cumulative impact damage, are written into the continuous residual queue. Step S4 reads the queue in chronological order to perform hydraulic topology assignment, and only records with the outside window flag being true, the interpolation flag being false, and the residual window confidence flag being true can be used as inputs for solving the leakage conductivity.

[0024] In this specific embodiment, S4 includes: The topology allocation module takes the residual observation records from step S3 as input and establishes a directed hydraulic topology map that includes the internal leakage channel of the piston seal, the external leakage or intrusion channels of the rod seal and dust scraper seal, the external leakage channels of the hose and detachable hydraulic joint, and the leakage channel in the control valve center position. The hoses and joints directly monitored by the two-chamber continuity residuals are limited to the monitored pipe sections between the pressure measuring point at the working port of the reversing valve and the corresponding pressure measuring point in the cylinder chamber. Flow measuring points or segmented pressure measuring points are also set upstream and downstream of the target pipe sections in the main supply line and the main return line, and the main line continuity residuals are established. Supply and return oil pipe sections without the configured independent main road observation are only registered as unobservable candidate pipe sections and are not included in the conduction volume numerical estimation; the node records the pressure, flow rate and valid marker of the measuring point in the rodless cavity, rod cavity, main supply oil road, main return oil road and external environment, and the edge records the channel type, connection node, allowed flow direction, conduction status, conduction volume to be estimated, monitoring boundary, equipment part number and maintenance baseline version. The topology version is identified by the hydraulic schematic diagram version and sensor configuration version. The node edge table is regenerated when any connection relationship or monitoring boundary changes. The directional pressure difference in each channel is calculated according to the formula. Generate, where, symbol Indicates the first The first channel in the The pressure difference between the windows along a preset direction, in Pa; symbol Indicates the first The upstream node of the channel is in the first Pressure per window, unit: Pa; symbol Indicates the first The downstream node of the channel is in the first Pressure per window, unit: Pa; function symbol and function symbols These represent the first node in the topology node edge list. Upstream and downstream node mappings of an edge; symbol Indicates the leakage channel number; symbol Indicates the residual window number; the external node pressure is obtained from the environmental pressure sensor. If the sensor is invalid, the atmospheric pressure from the most recent maintenance record at the same altitude is used and the observability level is reduced. The module determines the direction of the piston seal internal leakage channel by the pressure difference between the two chambers, determines the leakage or intrusion direction of the rod seal and dust scraper seal channel by subtracting the external pressure from the rod chamber pressure, and determines the corresponding external leakage direction by subtracting the external pressure from the internal pressure of the pipe section where the hose or connector is located. The internal pressure is selected from the connected cylinder chamber pressure, supply pressure or return pressure according to the current valve control state. When the control valve neutral position command lasts for no less than 100ms and the difference between the supply pressure and the corresponding chamber pressure is greater than 0.5MPa, the control valve neutral position leakage side is set to conduction. The side with an absolute pressure difference of less than 0.1MPa is closed in this window because the direction cannot be stably determined. The equivalent leakage flow rate of each channel is calculated according to the formula. Mapping, where the symbol Indicates the first The leakage channel is in the first The equivalent leakage flow rate of each residual window, in m³ / s. 3 / s; symbol Indicates the first The equivalent leakage flux of each channel, in m 3 / (s·Pascal's first-half power) and the constraint is non-negative; sign Indicates the first The pressure difference calculated at both ends of each channel in a preset positive direction, in Pa; symbol Indicates the leakage channel number in the hydraulic topology diagram; symbol Indicates the residual window number that increases over time; function symbol This represents a function indicating the direction of the pressure difference. The flow conservation relationship can be written as a formula. , where the symbol The observation vector consists of the rodless cavity continuity residual, the rod cavity continuity residual, and the supply and return main pipeline continuity residuals formed by the upstream and downstream flow difference or segmented pressure-flow mapping of the target pipe section when configuring independent main pipeline observations. The element unit is m. 3 / s; symbol This represents a dimensionless node-edge correlation matrix generated based on the current valve control status, operating phase, monitoring boundary, and effective measuring points. Inflow nodes are positive, outflow nodes are negative, and the corresponding column for candidate pipe segments without independent main road observation is fixed as a zero-sensitivity column. This represents a matrix with the square root of the directional pressure difference of each channel as its diagonal element, and the unit of each element is a 1 / 2 Pascal; symbol This represents the non-negative equivalent leakage flux vector for each channel, with elements in units of m. 3 / (s·Pascal to the power of one half); symbol This represents the fitting error vector formed by sensor noise and unmodeled flow, in units of m. 3 / s; symbol Indicates the residual window number; The module identifies usable measurement points as those connected to the current conductive pipeline and with valid quality flags among the pressure measurement points of the two chambers, supply and return oil, and independent main flow measurement points, and deletes columns in the correlation matrix that correspond to closed channels or have no effective pressure difference. For sensitivity columns with a norm of zero, it directly outputs the unobservable state, candidate pipe segment number, and missing observation reason code, without replacing missing observations with rank tests. For the remaining effective matrices, the rank is calculated through singular value decomposition. When the matrix column is full rank and the ratio of the minimum singular value to the maximum singular value is not less than 0.02, the corresponding channel is identified as a distinguishable channel. When multiple non-zero columns are linearly correlated, these channels are grouped into the same indistinguishable channel group and the combined conductivity is output. The combined record retains all candidate part numbers without forcibly giving a single fault location. For indistinguishable channel groups, the module follows the formula. Output combined conduction, where, sign Indicates the first The indistinguishable channel group in the first The combined equivalent leakage flux of the windows is expressed in m. 3 / (s·Pascal to the power of one half); symbol Indicates belonging to the first A set of channel numbers for an indistinguishable channel group; symbols Represents the set of elements. The first channel in the The non-negative equivalent leakage flux of each window, in m³. 3 / (s·Pascal to the power of one half); symbol Indicates the indistinguishable channel group number; symbol Indicates the channel number; symbol This indicates the residual window number; the combined value is only used for group-level state updates and is not split in any proportion. For distinguishable channels, the module follows the formula Solve for the nonnegative conduction flux, where the sign is... Indicates the first The equivalent leakage flux estimation vector for each window, with elements in units of m. 3 / (s·Pascal to the power of one half); symbol This represents the candidate flux vector during the optimization process, with elements having the same unit as before and all elements being non-negative; [symbol] This represents the residual scaling diagonal matrix, where the diagonal elements are the median absolute deviation of the residuals corresponding to step S3 at the healthy baseline, with a lower bound of 1. m 3 / s, element unit is m 3 / s; symbol This represents a diagonal matrix representing the conduction scale, where the diagonal elements are the 95th percentile of the healthy conduction capacity of the corresponding channel, with a lower limit of 1. m3 / (s·Pascal's first power of 2), with elements having the same units as the conductance; the two inverse scaling matrices make the residual term and the conductance variation term dimensionless, respectively; symbols Represents the current association matrix; symbol Represents the current directed pressure difference square root matrix; symbol This represents the current continuous residual vector, with elements in units of m. 3 / s; symbol This represents the dimensionless regularization weight that suppresses unwarranted jumps between adjacent windows. In this specific implementation, it is set to 0.05 and determined by leave-one-out verification of a healthy, leak-free cycle; (symbol) This represents the flux estimation vector of the previous comparable window, with elements in the same units as before; symbol Represents the 2-norm; symbol Indicates the residual window number; The solution employs non-negative least squares iteration with a maximum iteration count of 100. When the fitting error exceeds the 99th quantile of the healthy residual fitting error, the matrix condition number exceeds 50, or the iteration fails to converge, the current window is deemed infeasible. Single-channel flux is not updated; the previous comparable window state is frozen, and only the currently observable relationships and combined fluxes directly determined by the conservation equations are output. When all channels are unobservable, an empty flux vector and an unobservable cause code are output, and replacing unknown values ​​with zero values ​​is prohibited. For normally converged results, estimates below the 95th quantile of the healthy baseline flux are retained but marked as the baseline range, and leakage levels are not directly triggered. The module is based on the formula Calculate the dimensionless relative fitting error, where the sign... Indicates the first Dimensionless relative fitting error for each window; symbol Represents the edge incidence matrix of the current node; symbol Represents the current directed pressure difference square root matrix; symbol Represents the current nonnegative conduction flux estimation vector; symbol This represents the current continuous residual vector, in units of m. 3 / s; symbol This represents the lower 5th percentile of the norm of the baseline residuals for healthy individuals, expressed in m. 3 / s and not less than m 3 / s; symbol Represents the residual window number; function symbol Representation of the 2-norm; function symbol This indicates that the option with the higher value is selected; if the relative fitting error is greater than 0.3, the aforementioned infeasible branch is executed. To provide auxiliary evidence on the same time axis for step S5, the topology observation table also reserves a reference field for rod end acoustic emission characteristics. The rod end acoustic emission characteristics are obtained by the envelope of the output signal of the acoustic emission sensor set on the piston rod end seal. This step only saves the sensor number, time range and data reference key, and does not use the acoustic emission characteristics to change the flow conservation solution result of this step. The topology allocation module writes the observable relationships of each window, single-channel or combined-channel identifiers, equivalent leakage flux, differential pressure direction, fitting error, matrix rank, condition number, solution status, infeasibility cause code, and rod end acoustic emission feature reference key into the topology observation table, and establishes an index with natural cycle number and phase label. Step S5 reads the flux changes of each channel, the basic observable relationships, and the acoustic emission references in the same time range to generate pollution source observation factors. Step S6 reads the same records to perform cross-cycle wear state update.

[0025] In this specific embodiment, S5 includes: The pollution source identification module first establishes a healthy baseline bound to the maintenance baseline version. The baseline sample consists of at least 30 natural operating cycles where the maintenance status is confirmed, the continuous residual in step S3 does not trigger leakage, and the load characterization, oil temperature, bucket posture, and operating phase meet the same acquisition conditions. Each baseline record saves the phase label, retraction speed distribution, rod cavity pressure change rate distribution, particle concentration by size, acoustic emission envelope energy, conductivity distribution of each channel, and synchronous sampling error. When the seal is replaced, the joint is tightened, the oil is purified, or the sensor configuration is changed, the affected baseline version is immediately set to invalid. After maintenance, a new version is generated after obtaining 10 consecutive natural cycles that meet the acquisition conditions and whose residuals are within the healthy range. A first particle counting unit is installed in the return oil bypass of the bucket branch, and a second particle counting unit is installed in the supply oil bypass downstream of the main filter. Both units count particles in the same particle size range: 4 to 6 μm, 6 to 14 μm, and greater than 14 μm. A directional sampling valve and a 10 mL small-volume sampler, driven by a valve-controlled state, are installed upstream of the first particle counting unit. After recording dust or sediment exposure events, the module records candidate rod-end intrusion events during the period when the reversing valve control parameters continuously indicate piston rod retraction, the piston speed is within the 5% to 95th percentile of the healthy baseline retraction speed, and the absolute value of the rod chamber pressure change rate does not exceed the 95th percentile threshold of the healthy baseline. However, this event is not recorded as a candidate event. The oil discharged from the rodless chamber during the period is used as the rod end particle sample; in the same natural circulation, when the piston rod extends and the rod chamber turns to discharge oil, the directional sampling valve introduces the oil discharged from the rod chamber into the accumulator and obtains a 10mL sample after flushing with oil of no less than twice the total volume of the accumulator and connecting pipe. The second particle counting unit simultaneously obtains a 10mL supply reference sample; the transport delay from the start of the rod chamber discharge to the first counting unit is calibrated by the tracer particle test after maintenance and matched with the natural circulation number to the corresponding retraction event. If the valve position is uncertain, the flushing volume is insufficient, the transport delay error exceeds the 95th percentile of the health calibration, or cannot be uniquely matched, the particle record is discarded. The particle difference characteristics of each particle size range are calculated according to the formula. Calculation, where the symbol Indicates the first The first natural cycle in the 1st The particle concentration difference across different particle size ranges, expressed in particles / mL. A positive value indicates that the particle concentration on the return oil side of the bucket branch is higher than that on the supply oil side downstream of the main filter; [symbol missing] and symbols The first and second particle counting units respectively represent the first and second particle counting units in the first... The first natural cycle, the first The number of particles in each particle size range, expressed in units of particles; symbol and symbols These represent the actual sampling volumes of the two units in this natural cycle, in mL, and the difference between them must not exceed 0.1 mL; (symbol) Indicates the particle size range number; symbol Indicates the natural cycle number; if the sampling volume does not meet the consistency requirements, the particle difference record for that sampling step is discarded. The acoustic emission characteristics at the piston rod end are obtained by processing the output signal of a high-speed acoustic emission front-end located at the piston rod end seal with an original sampling rate of not less than 250 kS / s, followed by anti-aliasing bandpass filtering from 20 kHz to 100 kHz, full-wave rectification, and low-pass processing with a 1 ms time constant. The envelope is then processed according to the formula... Calculate the root mean square of the envelope during the admission period, where the sign... Indicates the first The root mean square envelope of the acoustic emission at the rod end of a natural cycle, in V; symbol Indicates the first The number of effective high-speed envelope samples within each natural cycle admission period is a positive integer; sign Indicates the number of the period One effective high-speed envelope sample value, in V; symbol Indicates the high-speed envelope sampling sequence number; symbol The natural cycle number is indicated; the high-speed front end incorporates the feature value, start and end high-speed timestamps, and quality flags into the 200Hz feature frame, and marks the acoustic emission feature as invalid when the effective sampling duration is less than 0.2s; The module defines the consistency time limit as the sum of the 95th percentile of the health baseline synchronization error, the 95th percentile of the transport delay error, and the upper limit of the timestamp error of each sensor. In this specific embodiment, the time is 2 seconds, and cross-evidence hysteresis matching is performed on the dust or sediment exposure event indication sequence, the particle differential feature increment after refitting, the rod end acoustic emission feature increment, and the conduction increment of the rod seal and dust scraper seal channels on the real high-speed timestamp axis; the first matching degree is calculated according to the formula Calculation, where the symbol Indicates the first The dimensionless first matching degree of a natural cycle is restricted to 0 to 1; symbol Indicates the first The exposure event indicator increments in a natural cycle, set to 1 for a confirmed exposure event and 0 otherwise; (symbol) Indicates the first Class of evidence in time The median and median absolute deviation relative to the baseline version are normalized and truncated to positive increments between 0 and 1, where , and The symbols represent, in order, the evidence for particle difference, rod-end acoustic emission, and rod-seal conductivity; Indicates the first A high-speed timestamp of each natural cycle exposure event, in seconds; symbol This represents the allowable time delay of the corresponding evidence relative to the event of exposure, in seconds; symbol Indicates the consistency time limit, in seconds; symbol Indicates the natural cycle number; function symbol This means that the product of the three types of non-event evidence is taken, and the fourth root keeps the matching degree dimensionless. If any evidence is missing within the allowed time limit or the baseline version is inconsistent, the first matching degree is marked as invalid instead of being replaced by zero. When there are multiple peaks with the same value, the one with the smallest sum of the three absolute time delays is selected, and the negative increment is truncated to 0. For particle differential increments that do not match dust or sediment exposure events within the consistency timeframe, the module follows the formula... Determine the second matching degree, and apply the formulas respectively. and Normalization, where the symbol Indicates the first The dimensionless second matching degree of a natural cycle; symbol Indicates the candidate channel category for internal wear. This indicates the piston seal internal leakage channel. Indicates the rod seal and dust scraper seal channel; symbol Represents a non-negative hysteresis quantity in units of comparable natural cycles; symbol This represents the maximum number of lag cycles, which is taken as 3 in this specific implementation; symbol Indicates as of the date A relevant window consisting of 20 comparable cycles from the same baseline version preceding each natural cycle; symbol Indicates the comparable cycle number within the relevant window; symbol Indicates the first The arithmetic mean of the positive particle concentration increments across three particle size ranges in one comparable cycle, expressed in particles / mL; symbol Indicates category The positive increment of conduction after lag, in meters. 3 / (s·Pascal's first-half power); function symbol This represents the Pearson correlation coefficient calculated for paired valid samples within the relevant window; function symbol. This indicates summing the relevant peaks of two candidate channels; a coefficient of 1 / 2 indicates taking the mean. If there are fewer than 10 paired valid samples, zero variance for any sequence, missing values ​​that cannot be matched, or inconsistent baseline versions, the second match is marked as invalid and internal wear probability updates are stopped. Indicates the first The probability of external pollution intrusion in a natural cycle; symbol Indicates the first The probability of internal wear debris in a natural cycle; symbol Indicates the first dimensionless matching degree; symbol and symbols These represent the 50th and 95th percentiles of the first match between healthy baselines, respectively; (symbols omitted) and symbols These represent the 50th and 95th percentiles of the second match between healthy baselines, respectively; (symbols omitted) Indicates the natural cycle number; function symbol This means that the result is restricted to 0 to 1, and the higher the probability, the more consistent the corresponding source evidence; when the upper and lower bounds of the quantiles are equal, the probability update stops and a flag indicating insufficient baseline dispersion is output. The module uses the basic observable relationship from step S4 as its basis, and follows the formula... Form channel topology observation factors, where, symbol Indicates the first The first channel in the A dimensionless topological observation factor of a natural cycle; symbol This represents the basic observable state given in step S4, with distinguishable channels assigned a value of 1, channels that can only be combined for observation assigned a value of 0.5, and unobservable channels assigned a value of 0; (symbol) This indicates that evidence of foreign contamination intrusion is relevant to the first... The dimensionless weights for each channel are 1 only for channels with rod seals and dust scraper seals that leak or intrude, and 0 for the remaining channels; (symbol) Evidence of internal wear and debris is relevant to the first The dimensionless weights for each channel are: 1 for the piston seal internal leakage channel and the rod seal and dust scraper seal channels; and 0 for the remaining channels. and symbols These represent the probability of external contamination intrusion and the probability of internal wear and debris, respectively; (symbols omitted) Indicates the leakage channel number; symbol Indicates the natural cycle sequence number; When particle differential or acoustic emission characteristics are invalid, the module does not replace the zero value with a missing value; it only retains the previous valid result of the corresponding probability when the baseline version, sensor configuration, and operating phase have not changed, and the retention time is at most two comparable natural cycles. The evidence integrity coefficient is generated according to the rule of multiplying by 0.5 after each cycle. After more than two cycles, maintenance, or baseline version changes, the probability is marked as invalid. External pollution and internal wear retain independent evidence integrity coefficients. When any probability is invalid, its corresponding weight in step S6 is reset to zero. When all auxiliary evidence is invalid, the source probability-driven state component is frozen and the status of continued acquisition or manual review is output. Finally, the particle differential vector, rod end acoustic emission characteristics, two types of matching degree, two types of probability, topology observation factor, basic observable relation reference key, baseline version number, probability validity period, and evidence integrity coefficient are written into the pollution source observation record for step S6 to update the channel wear status together with the equivalent leakage conductivity and impact damage.

[0026] In this specific embodiment, S6 includes: The status update output module reads the work sample from step S1, the cumulative impact damage from step S2, the equivalent leakage conductivity from step S4, and the topology observation factor from step S5. First, it follows the formula... Determine the first The load characterization of a natural cycle, where the symbol Indicates the first The median of the resultant hydraulic pressure in the two chambers during the corresponding working phase of each natural cycle, in N; symbol and symbols These represent the median pressures in the rodless and rod-shaped chambers within this phase, respectively, in Pa; symbol and symbols These represent the effective areas of the rodless cavity and the rod cavity, respectively, in m². 2 ;symbol Indicates the natural cycle number; the module obtains the phase duration by subtracting the phase start point from the phase end point, in seconds; For load characteristics, oil temperature, bucket attitude and phase duration, the module obtains the median and median absolute deviation from the current effective health baseline, and sets the comparable deviation threshold to three times the corresponding median absolute deviation and not less than three times the sensor repeatability error. Only when the working phase is the same and the difference between the four working conditions and the baseline center or the candidate group center does not exceed the threshold of their respective units, are they classified into the same comparable cycle group. Cycles in which any working condition is missing or exceeds the range of 1% to 99% of the health baseline do not participate in the status trend update but retain the diagnostic record. Within each comparable cycle group, the module will... The conduction excess of each channel, the contamination source probability corrected for evidence completeness, the topological observation factor, and the incremental impact damage of this cycle are combined to form the observations, which are then calculated according to the following formula: , where the symbol Indicates the first The first channel in the Dimensionless wear observations of a natural cycle; symbol This indicates the normalized excess percentage of the channel throughput exceeding the 95th percentile of the healthy baseline, limited to 0 to 1, with a higher value indicating a higher risk of leakage; symbol This represents the dimensionless observation support gating coefficient obtained by dividing the topological observation factor in step S5 by its theoretical maximum value. It is limited to 0 to 1 and is used only to gate flux and contamination source anomalies, not as independent risks; (symbol) and symbols These represent the probability of external contamination intrusion and the probability of internal wear and debris, respectively; (symbols omitted) and symbols These represent the completeness coefficients of external contamination evidence and internal wear evidence output in step S5, respectively, ranging from 0 to 1, with 0 representing invalid or expired evidence; (symbols omitted). and symbols Using the corresponding source from step S5 respectively, the first The weight of each channel; symbol This represents the 99th quantile normalized value of the incremental impact damage from the current cycle relative to the baseline impact damage, limited to 0 to 1; symbol This indicates that the impact is transmitted from the bucket structure to the first... The dimensionless correlation weight of each channel, ranging from 0 to 1, is determined by the proportion of the response energy of the adjacent measuring points in the leak-free instrumented impact test, and is set to 0 for channels with no structure transmission path; The symbol is... ,symbol ,symbol and symbols The dimensionless weights representing conductivity, external contamination, internal wear, and impact damage are respectively taken as 0.5, 0.15, 0.15, and 0.2 in this specific embodiment, and the sum of the four is 1; (symbol) Indicates the leakage channel number; symbol Indicates the natural cycle number; function symbol This means that the results are limited to 0 to 1; when the flux is excessive, the two types of effective contamination probabilities and the related shock anomalies are all zero, the wear observation must be zero. When no maintenance event is received, the module follows the formula. Update wear status, where the symbol Indicates the first The first channel in the The dimensionless wear state after one natural cycle, ranging from 0 to 1, with a larger value indicating more severe wear or leakage; symbol Indicates the wear state after the previous comparable natural cycle; symbol Indicates the first The dimensionless update coefficient for each channel is set to 0.15 in this specific implementation and determined by minimizing the prediction error through a tag-based cycle before and after maintenance; (symbol) Indicates current wear observations; symbol Indicates the leakage channel number; symbol Indicates the natural cycle number; function symbol Used to ensure that the state remains unchanged or changes only in the direction of increasing wear when there are no maintenance events; For the indistinguishable channel group output in step S4, the module combines the groups. For independent state objects, utilize combined flux. The excess proportion relative to the 95th percentile of the healthy baseline of this group was used to replace the single-channel excess proportion, and the excess proportion was calculated according to the aforementioned observation structure using the maximum value of the effective source weight within the group, the group-level observation support gating coefficient, and the group-level impact correlation weight. Then follow the formula Update group-level wear status, where the symbol Indicates the first The combination group in the first The combined conductance of a natural cycle, in m 3 / (s·Pascal to the power of one half); symbol and symbols They represent the first The dimensionless wear state of each combination group compared to the previous comparable cycle; symbol Indicates the first The dimensionless update coefficients of each group are calibrated by group-level healthy samples and maintenance tags; symbol Indicates the first Dimensionless wear observations for each combination group; symbol Indicates the combined channel group number; symbol Indicates the natural cycle sequence number; group-level status, confidence level, level, maintenance threshold, and maintenance window are only compared with the group-level baseline and group-level manufacturer limit, and the combined throughput is not split or backfilled as member channel history; When a seal replacement event is received and the no-load operation, pressure holding check, and health residual verification after maintenance are all passed, the module resets the wear state of the distinguishable channel where the replaced seal is located to 0 and invalidates its health baseline. When a joint tightening event is received and the same verification is passed, only the state of the corresponding distinguishable hose and detachable hydraulic joint leakage channel is reset to 0 and the baseline of that channel is invalidated. If the maintained member belongs to an indistinguishable combination group, the state of other members is not reset or split. Instead, the state and health baseline of the combination group are marked as pending reconstruction and the group-level trend update is paused. When the subsequent operation phase makes the member distinguishable again, the single-channel state is established from the new observation without backfilling the previous combination value. When an oil purification event is received, only the particle differential characteristic baseline is invalidated without resetting any channel wear state. When the sensor configuration changes, the observation baseline of the single channel or combination group involving the sensor is invalidated. All invalidated baselines are re-established according to the health sample conditions specified in step S5 before the state update is restored. When a module has at least five comparable cycles of the same baseline version for the same object, according to the formula... Determine the confidence level, where the symbol is... Indicates the first The first channel in the The dimensionless confidence level for each natural cycle ranges from 0 to 1, with a higher value indicating a more reliable diagnosis; symbol This represents the root mean square error of the flow rate fitting in step S4 within this loop, in meters. 3 / s; symbol This represents the 95th percentile of the fit error to the healthy baseline, in meters. 3 / s and the lower limit is set to m 3 / s; symbol The channel observability coefficient represents the current operational phase; it is 1 for distinguishable observability, 0.5 for combined observability, and 0 for unobservable observability; (symbol) The dimensionless consistency is obtained by dividing the number of cycles in which the channel flux changes in the same direction among the last five comparable cycles by 5; symbol Indicates the leakage channel number; symbol Indicates the natural cycle number; function symbol This represents the natural exponential function; if there are fewer than five comparable cycles, the division by 5 is not performed, the confidence field is set to initial state, the cause code is set to insufficient history, and the determined fault location is not output; after five cycles, the confidence calculation is resumed, and when the confidence is below 0.4, only the pending review status is output; combined groups The same formula is used, and the corresponding group-level fitting error, group-level observable coefficient, and group-level directional consistency are substituted into it; The wear or leakage level is determined by the conductivity quantile boundary of the current effective health baseline and the equipment manufacturer's allowable leakage limit. When the conductivity does not exceed the 95th percentile of health, it is classified as normal; when it exceeds the 95th percentile but does not exceed the 99th percentile, it is classified as concern; when it exceeds the 99th percentile but is lower than the manufacturer's allowable leakage limit converted to conductivity, it is classified as warning; when it reaches or exceeds the converted conductivity, it is classified as maintenance. For only combined observable channel groups, the module compares the combined conductivity with the combined group's own 95th percentile of health, 99th percentile of health, and group-level allowable leakage converted limit, and outputs the combined fault location and group-level level. It does not express the highest level among unknown members, nor does it split the combined result into single-channel conclusions. The maintenance threshold for each distinguishable channel is calculated according to the formula. Determined, where the symbol Indicates the first Dimensionless maintenance state threshold for each channel; symbol This indicates the wear condition boundary when the channel enters the maintenance stage; symbol This represents the dimensionless wear state boundary obtained by converting the manufacturer-allowed leakage flow rate into conductivity under the corresponding phase pressure difference and then mapping it; symbol Indicates the leakage channel number; function symbol This indicates that the lower of the two boundary values ​​is taken; the group is determined by the lower limit of the group level and the group-level allowable leakage limit according to the same rule. ,symbol Indicates the first The dimensionless maintenance state threshold of each combination group, symbol Indicates the combined channel group number; manufacturer limit conversion and status mapping parameters are stored in the maintenance threshold table and use equipment model, channel or combined group type and oil temperature range as the joint key; The module uses a robust linear regression slope as the rate of change of wear status across cycles only when there are at least 20 most recent comparable cycles of the same baseline version in the same channel or combination group; when the growth rate is greater than the significant growth lower limit, it follows the formula... Predict the number of remaining job cycles, where the symbol Indicates the first The estimated remaining natural cycles before each channel reaches the maintenance threshold, a non-negative integer; sign Indicates the maintenance threshold for this channel; symbol Indicates the current wear condition; symbol This represents the wear rate per comparable natural cycle of the channel, which is dimensionless; the lower limit for significant growth is taken as... Each cycle is checked against the 99th quantile of the health status slope; symbol Indicates the leakage channel number; symbol Indicates the current natural loop index; function symbol Indicates rounding up; when there are fewer than 20 comparable cycles, the trend is not calculated and the remaining cycle number segment is marked as insufficient historical data; when the growth rate is not greater than the significant growth lower limit, division is prohibited and the remaining cycle number segment is marked as no growth trend detected; when the current state has reached the maintenance threshold, the remaining cycle number is set to 0; combined group The group-level maintenance threshold, current group-level status, and group-level growth rate are calculated using the same branch. ; The status update output module converts the remaining cycle count into a date range only when a limited estimated remaining cycle count has been obtained, the most recent effective working days are no less than 30, and the average daily natural cycle count of the stated working days is greater than 0. When there are 20 comparable cycles but fewer than 30 effective working days, the remaining cycle count is retained, and the start and end dates of the maintenance window are marked as insufficient utilization history. When the average daily cycle count is 0, division is disabled, and the date field is marked as downtime or no recent utilization. When there are fewer than five comparable cycles, the initialization status and a "continue collection" prompt are output, but the fault location is uncertain. When there are five but fewer than 20 cycles, the current level, the location to be reviewed, and a "continue collection" prompt are output, but the fault location is uncertain. The module outputs trends and maintenance dates, and only outputs trends and maintenance windows according to the aforementioned conditions after reaching 20. The module writes object type, fault location or combination location, wear or leakage level, confidence level, current wear status, maintenance threshold, estimated remaining cycle number and its status code, maintenance window start and end dates and their status codes, baseline version, evidence completeness, and most recent maintenance event into the sealed status report. When the maintenance level is reached and the confidence level is not lower than 0.7, a maintenance work order is generated. When the confidence level is lower than 0.4, all key evidence is invalid, or any historical insufficient branch is established, a prompt to continue collection or manual review is output. The report and work order are the final outputs of this method.

[0027] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0028] This invention sequentially connects the natural operating phase, propagation time difference impact window, two-chamber continuity residual, and hydraulic topology allocation, so that pressure, motion, and valve control signals are first converted into equivalent leakage flux of each leakage channel under operating conditions and evidence access conditions, and then combined with particle and acoustic emission evidence to form a sealed channel observation factor, thereby distinguishing different fault sources without performing a special shutdown pressure holding test.

[0029] Compared to processing structures that judge the sealing status solely based on pressure decay or total particle count, this invention adds the generation of propagation time difference of the impact window and the relationship between the prohibition of evidence within the window, the mapping of contamination sources of particle difference and acoustic emission, the screening of comparable operating conditions, and the maintenance event-driven status and health baseline update constraints. This allows impact damage, actual leakage, external contamination, and internal wear to enter their respective processing channels, thereby better achieving sealing status positioning and life trend prediction.

Claims

1. A method for monitoring the sealing status of a skid steer loader bucket, characterized in that, include: S1. Synchronously collect the pressure, motion parameters, directional valve control parameters, oil temperature and impact reference signal of the two chambers of the bucket tilting cylinder. Based on the directional valve control direction, piston movement direction, bucket posture and pressure changes, divide the natural operation cycle and generate operation samples with operation phase identifiers. S2. Determine the impact starting point based on the impact reference signal, and determine the impact influence window based on the propagation time difference from the bucket impact source to each measuring point. Determine the data outside the window of the work sample as the estimated data outside the window, and accumulate the data inside the window to obtain the impact damage amount. Do not use the data inside the window as direct evidence of leakage triggering. S3. Using out-of-window estimation data, construct the two-chamber continuity residual based on valve orifice flow, effective volume of the two chambers, oil volumetric elastic modulus, differential pressure of the two chambers, and piston speed. S4. Input the two-chamber continuity residual into the hydraulic topology model. The hydraulic topology model includes the internal leakage channel of the piston seal, the external leakage or intrusion channel of the rod seal and dust scraper seal, the external leakage channel of the hose and detachable hydraulic joint, and the leakage channel in the control valve center position. Under the constraints of flow conservation, non-negative leakage conductance, and observable relationship of the measuring point, the two-chamber continuity residual is allocated to obtain the equivalent leakage conductance of each channel. S5. Extract the particle differential characteristics and rod end acoustic emission characteristics between the bucket branch and the oil supply branch. Perform time-series consistency calculations on the equivalent leakage conductivity of each channel with the particle differential characteristics and rod end acoustic emission characteristics to obtain the probability of external contamination intrusion and the probability of internal wear debris, and form the topology observation factor of the corresponding sealed channel. S6. Based on load, oil temperature, bucket posture and working phase, select comparable working cycles, update the wear status of each channel across cycles according to equivalent leakage conductance, topology observation factor, impact damage amount and maintenance events, and output fault location, wear or leakage level, confidence level and maintenance window.

2. The method for monitoring the sealing status of a skid steer loader bucket according to claim 1, characterized in that, S1 includes: The motion parameters are at least one of the cylinder stroke and the bucket angle, and the directional valve control parameters are at least one of the directional valve command and the valve core displacement. The directional valve control parameters, the sign of the piston speed, the attitude range of the bucket angle, and the direction of pressure change in the two chambers are used to form the phase discrimination conditions. According to the phase discrimination conditions, the natural operation cycle is marked as the loading contact phase, lifting phase, steady-state holding phase, tipping unloading phase, recovery phase, and directional phase. The start and end times of the phases, the load characteristics, the oil temperature, and the bucket attitude are written into the operation sample.

3. The method for monitoring the sealing status of a skid steer loader bucket according to claim 1, characterized in that, S2 includes: obtaining the impact arrival time using impact reference sensors installed in the bucket, cylinder body, sealing seat, and pressure measuring point, and taking the impact arrival time at the bucket as the impact starting point; determining the propagation time difference based on the propagation path and corresponding propagation speed between each measuring point and the impact source in the bucket, and generating an impact influence window for each measuring point using the impact starting point, the propagation time difference, and the window width determined by the response duration of the non-leakage impact sample; the pressure and acoustic emission signals within the impact influence window are only used to accumulate the impact damage amount and are not used as evidence of leakage triggering, and the data outside the impact influence window are used as the window-out estimation data in step S3.

4. The method for monitoring the sealing status of a skid steer loader bucket according to claim 1, characterized in that, S3 includes: collecting the supply pressure and return pressure respectively by pressure sensors installed in the main supply line and the main return line; selecting the supply pressure or return pressure as the pressure on one side of the valve port according to the valve control direction and using the corresponding cylinder chamber pressure as the pressure on the other side of the valve port to obtain the pressure difference across the valve port; determining the valve port flow rate of the two chambers according to the valve core displacement, the pressure difference across the valve port, and the flow coefficient corresponding to the oil temperature; determining the effective volume of the two chambers according to the initial cylinder chamber volume, the effective piston area, and the cylinder stroke, and correcting the bulk modulus of the oil according to the oil temperature; performing bandwidth-limited differential processing on the pressure signal to obtain the differential pressure of the two chambers, and performing time differential processing on the cylinder stroke to obtain the piston speed; for each chamber, using the difference between the valve port flow rate and the volumetric flow rate caused by the piston movement and the compression flow rate caused by the pressure change as the continuity residual of the corresponding chamber.

5. The method for monitoring the sealing status of a skid steer loader bucket according to claim 4, characterized in that, S4 includes: determining the flow direction of the piston seal internal leakage channel based on the pressure difference between the two chambers; determining the flow direction of the rod seal and dust scraper seal external leakage or intrusion channels based on the pressure difference between the rod chamber pressure and the external pressure; determining the flow direction of the hose and detachable hydraulic joint external leakage channels based on the pressure difference between the internal pressure and the external pressure of the pipe section where the hose or detachable hydraulic joint is located, wherein the internal pressure is the cylinder chamber pressure, supply pressure, or return pressure connected to the pipe section; and determining the conduction status of the control valve mid-position leakage channel based on the directional valve neutral position command, supply pressure, return pressure, and corresponding cylinder chamber pressure. The process involves establishing a mapping between the equivalent leakage flow rate of each channel and the corresponding pressure difference and equivalent leakage conductance, and forming a flow conservation equation from the continuity residuals of the two chambers. The available pressure measurement points among the pressure measurement points of the two chambers, the supply pressure measurement point, and the return pressure measurement point, which are connected to the pipeline in the conducting state under the current operating phase, are identified as available pressure measurement points. An observable relationship is generated based on the available pressure measurement points and the valve control state. For indistinguishable channels, a combined conductance is output; for distinguishable channels, their respective equivalent leakage conductance is solved under the constraint of non-negativity of the equivalent leakage conductance.

6. The method for monitoring the sealing status of a skid steer loader bucket according to claim 5, characterized in that, S5 includes: establishing a healthy baseline using multiple natural work cycle data points that have been maintained and confirmed, where continuous residuals have not triggered leakage, and where load, oil temperature, bucket posture, and working phase meet the same baseline acquisition conditions; discontinuing the use of the corresponding healthy baseline after seal replacement, joint tightening, oil purification, or sensor configuration changes cause it to fail, and reconstructing the corresponding healthy baseline with newly acquired data after maintenance and verification that the baseline acquisition conditions are met; setting a first particle counting unit in the bucket branch return oil bypass and a second particle counting unit in the supply oil bypass downstream of the main filter; and recording the particle count... Following a dust or sediment exposure event, the period during which the reversing valve control parameters continuously indicate piston rod retraction, the piston speed is within the admission range determined by the healthy baseline retraction speed distribution, and the rod chamber pressure change rate does not exceed the admission threshold determined by the healthy baseline pressure change rate distribution is defined as the rod retraction admission period. During the rod retraction admission period, the first particle counting unit and the second particle counting unit are controlled to obtain equal sampling volumes, and the particle count is counted according to the same particle size range. The particle differential feature is obtained by subtracting the corresponding particle size count of the second particle counting unit from the particle size count of the first particle counting unit.

7. The method for monitoring the sealing status of a skid steer loader bucket according to claim 6, characterized in that, The acoustic emission characteristics at the rod end are obtained by the envelope of the output signal of the acoustic emission sensor located at the piston rod end seal seat; the degree of matching of dust or mud exposure events, particle differential characteristic changes, rod end acoustic emission characteristic changes, and equivalent leakage conductivity changes of the rod seal and dust scraper seal leakage or intrusion channels within a consistency time limit determined by the synchronous sampling error of the health baseline is determined as the first matching degree; The degree of matching between the particle differential characteristic changes that do not match the dust or sediment exposure event within the consistency time limit and the equivalent leakage conductivity changes of the piston seal internal leakage channel and the rod seal and dust scraper seal external leakage or intrusion channel is determined as the second matching degree; the first matching degree and the second matching degree are normalized respectively to obtain the external contamination intrusion probability and the internal wear debris probability; based on the observable relationship generated in step S4, the external contamination intrusion probability is used to weight the rod seal and dust scraper seal external leakage or intrusion channel, and the internal wear debris probability is used to weight the piston seal internal leakage channel and the rod seal and dust scraper seal external leakage or intrusion channel, forming the topological observation factor.

8. The method for monitoring the sealing status of a skid steer loader bucket according to claim 7, characterized in that, S6 include: The load characterization quantity is determined based on the pressure difference between the two chambers and the effective area of ​​the piston, and the phase duration is determined based on the difference between the start and end times of the working phase. The corresponding comparable deviation thresholds are determined by the distribution of load characteristics, oil temperature, bucket posture and duration of each operation phase in the health baseline. Operation cycles with the same operation phase and whose differences in load characteristics, oil temperature, bucket posture and duration of each operation phase do not exceed the corresponding comparable deviation thresholds are grouped into the same comparable cycle group. Within each comparable cycle group, the wear status of each channel is updated based on the direction of change of equivalent leakage conductance, topology observation factor, and cumulative impact damage over multiple consecutive operating cycles. The confidence level is determined based on the fitting error of the continuity residual, the number of observable channels in the current working phase, and the consistency of conduction across multiple working cycles. The wear or leakage level is determined based on the quantile boundaries of the equivalent leakage conduction of each channel in the health baseline.

9. The method for monitoring the sealing status of a skid steer loader bucket according to claim 8, characterized in that, When no maintenance event is received, the wear status of each channel is updated and constrained to remain unchanged or change in the direction of increasing wear. When a seal replacement event is received and verified after maintenance, the wear status of the channel where the replaced seal is located is reset and the corresponding health baseline is invalidated. When a joint tightening event is received and verified after maintenance, the wear status of the leakage channels of the hose and detachable hydraulic joint is reset and the corresponding health baseline is invalidated. When an oil purification event is received, only the health baseline corresponding to the particle differential feature is invalidated without resetting the wear status of any channel, and the invalidated health baseline is reconstructed according to the health baseline reconstruction method. A maintenance threshold is set for each channel, which is determined by the smaller of the lower bound of the wear or leakage level that the corresponding channel needs to be maintained and the wear status boundary obtained by converting the allowable leakage limit given by the equipment manufacturer. The number of operation cycles required to reach the corresponding maintenance threshold is predicted based on the cross-cycle change rate of the wear status of each channel, and the number of operation cycles is converted into a maintenance window.

10. A skid steer loader bucket seal condition monitoring system, used to execute the skid steer loader bucket seal condition monitoring method according to any one of claims 1 to 9, characterized in that, include: The data acquisition and phase division module includes a two-chamber pressure sensor, an oil supply pressure sensor, an oil return pressure sensor, a motion parameter sensor, a directional valve control parameter acquisition unit, an oil temperature sensor, and an impact reference sensor, used to generate operation samples with operation phase identifiers; the impact window module is used to determine the impact influence window and generate estimated data outside the window and the impact damage amount; the residual generation module is used to generate two-chamber continuity residuals based on the estimated data outside the window. The topology allocation module is used to allocate the continuity residuals of the two chambers in the hydraulic topology model and obtain the equivalent leakage conductivity of each channel; the pollution source identification module includes a first particle counting unit set in the return oil bypass of the bucket branch, a second particle counting unit set in the oil supply bypass downstream of the main filter, and an acoustic emission sensor set in the piston rod end seal seat, which is used to generate the probability of external pollution intrusion, the probability of internal wear debris, and the topology observation factor based on the equivalent leakage conductivity of each channel, particle differential characteristics, and rod end acoustic emission characteristics; The status update output module is used to filter comparable work cycles, update the wear status of each channel across cycles, and output the fault location, wear or leakage level, confidence level, and maintenance window.

Images