Hydraulic motor working state detection method and device and storage medium

Abstract

The invention provides a hydraulic motor working state detection method and device and a storage medium, and relates to the technical field of fault detection.The method comprises the steps that a sensor cluster is arranged at key nodes of a hydraulic motor, and a multi-dimensional monitoring data set is constructed; calculating displacement data of a hydraulic motor and kinematic viscosity data of hydraulic oil according to the multi-dimensional monitoring data set to obtain a global monitoring data set; the actual leakage amount of the hydraulic motor is calculated based on the global monitoring data set, and an operation mode switching instruction is generated; receiving the load monitoring instruction or the shutdown maintenance instruction, starting multi-mode fault diagnosis, and obtaining a fault cause set of the hydraulic motor by using a Bayesian network model; and determining a maintenance scheme according to the fault cause set of the hydraulic motor. The technical problems that in the prior art, hysteresis exists, effective preventive detection measures are lacked, potential problems possibly existing in a hydraulic motor cannot be found in time, the maintenance cost is high, and the equipment recovery difficulty is large are solved.

Description

Technical Field

[0001] The present application relates to the field of fault detection technology, and in particular to a method, device and storage medium for detecting the working status of a hydraulic motor. Background Art

[0002] In the marine industry, hydraulic motors, thanks to their exceptional performance in power output and torque transmission, are an indispensable component for the proper operation of numerous marine equipment. For example, in key equipment such as a ship's steering gear, cargo winch, and anchor winch systems, the stable operation of hydraulic motors is directly related to the ship's maneuverability, cargo loading and unloading efficiency, and safe mooring. Given the critical role of hydraulic motors in ship operations, ensuring their healthy and stable operation is essential. To achieve this, regular inspections of hydraulic motors are essential. However, current inspection technologies for marine hydraulic motors have numerous shortcomings. During operation, there is a lack of effective means to monitor operating parameters and status. Consequently, fault detection is often only initiated when unusual noises or malfunctions occur. This post-inspection approach presents significant risks, as by the time obvious abnormalities are detected, the hydraulic motor's internal components may already be damaged or even destroyed. Once this occurs, even if the cause of the fault can be identified, repairs can be challenging. On the one hand, the cost of repairs is usually very high, which undoubtedly increases the operating costs of the ship. On the other hand, in some cases, the damage to the hydraulic motor may be beyond repair and can only be replaced. This not only leads to an increase in equipment replacement costs, but may also affect the normal operation of the ship. From the perspective of equipment management and maintenance costs, it costs a high price to restore the normal function of the equipment.

[0003] In summary, existing technologies for hydraulic motor detection exhibit significant lags and lack effective preventative detection measures, making it difficult to promptly identify potential problems with hydraulic motors. This can easily lead to a series of issues, including high repair costs and difficulty restoring equipment. Therefore, a method is needed to address these issues. Summary of the Invention

[0004] The present disclosure provides a method, device and storage medium for detecting the working status of a hydraulic motor, which are used to solve the technical problems existing in the prior art, such as hysteresis, lack of effective preventive detection measures, inability to timely discover potential problems that may exist in the hydraulic motor, and easy result in high maintenance costs and difficulty in equipment recovery.

[0005] According to a first aspect of the present disclosure, a method for detecting a working state of a hydraulic motor is provided, comprising: Deploy sensor clusters at key nodes of the hydraulic motor to obtain multi-dimensional state parameters of the hydraulic motor in real time, pre-process the multi-dimensional state parameters of the hydraulic motor, and construct a multi-dimensional monitoring data set, which includes flow data, speed data, pressure data, and temperature data; Calculating the displacement data of the hydraulic motor and the kinematic viscosity data of the hydraulic oil according to the multi-dimensional monitoring data set to obtain a global monitoring data set; Calculating the actual leakage of the hydraulic motor based on the global monitoring data set, assessing the state of the hydraulic motor according to the actual leakage of the hydraulic motor, and generating an operation mode switching instruction, wherein the operation mode switching instruction includes a continuous operation instruction, a load monitoring instruction, and a shutdown maintenance instruction; receiving the load monitoring instruction or the shutdown maintenance instruction, starting multimodal fault diagnosis, and obtaining a set of hydraulic motor fault causes using a Bayesian network model; A maintenance plan is determined according to the hydraulic motor fault cause set, wherein the maintenance plan includes fault location information, a spare parts list, a process route map, and an estimated maintenance time.

[0006] Furthermore, a multi-dimensional monitoring dataset is constructed, including: The hydraulic motor inlet and outlet pipes, shaft ends, and housing oil drain ports are selected as key nodes for deploying sensor clusters, including flow sensors, speed sensors, pressure sensors, and temperature sensors. The sensor cluster continuously collects flow data, speed data, pressure data, and temperature data of the hydraulic motor during operation to obtain multi-dimensional state parameters of the hydraulic motor, and adopts the IEEE 1588 precision time protocol to ensure synchronous capture of the multi-dimensional state parameters of the hydraulic motor; The isolation forest algorithm is used to remove outliers in the multidimensional state parameters of the hydraulic motor, and the cubic spline interpolation method is used to fill the missing data segments, reconstruct the continuous signal, and construct a multidimensional monitoring data set.

[0007] Furthermore, a global monitoring data set is obtained, including: Acquiring displacement data of the hydraulic motor based on the flow data and speed data in the multi-dimensional monitoring data set; The hydraulic oil viscosity-temperature relationship calculation model is set according to the type of hydraulic oil. The specific formula is: ISO VG32: η=144-3.2T; ISO VG48: η=107.5-1.25T; ISO VG68: η=159-2.4T; Among them, ISO VG32, ISO VG48, and ISO VG68 are three types of hydraulic oil commonly used in hydraulic motors. η represents the kinematic viscosity of the hydraulic oil, and T represents the temperature data. Input the temperature data in the multi-dimensional monitoring data set to obtain the corresponding kinematic viscosity data of the hydraulic oil; The displacement data, kinematic viscosity data and multi-dimensional monitoring data set are integrated to construct a global monitoring data set.

[0008] Furthermore, an operation mode switching instruction is generated, including: The actual leakage of the hydraulic motor is calculated based on the global monitoring data set. The specific formula is: ; Among them, Q L represents the actual leakage of the hydraulic motor, X represents the leakage coefficient, the specific value is 0.36, Q represents the displacement data of the hydraulic motor, P represents the pressure data of the hydraulic motor, and η represents the kinematic viscosity data of the hydraulic oil; The actual leakage of the hydraulic motor is compared with the leakage baseline to establish a three-level evaluation system. The leakage baseline includes a mild leakage baseline M and a severe leakage baseline N, wherein the mild leakage baseline M is 3%Q / 100bar and the severe leakage baseline N is 5%Q / 100bar; When 0 L ≤N, the hydraulic motor is judged to be in the first-level health state, and a continuous operation instruction is generated. The current leakage volume is within the allowable fluctuation range, the hydraulic motor is in good condition, and the regular inspection cycle is continued; When N L ≤M, the hydraulic motor is judged to be in the second-level warning state and a load monitoring instruction is generated. At this time, the hydraulic motor leakage exceeds the safety margin and needs to be derated to 85% of the rated power. Multi-modal fault diagnosis is also started simultaneously; When Q L >M, the hydraulic motor is judged to be in the third-level fault state and a shutdown and maintenance instruction is generated. At this time, the hydraulic motor is leaking seriously and needs to be stopped immediately for multi-modal fault diagnosis.

[0009] Furthermore, the load monitoring instruction or shutdown maintenance instruction is received, multimodal fault diagnosis is started, and a set of hydraulic motor fault causes is obtained using a Bayesian network model, including: Obtain the global monitoring data set, divide the global monitoring data set, regard the data collected at the last 10 time points as real-time monitoring data, and regard the remaining data as historical monitoring data; Constructing a Bayesian network model based on the historical monitoring data; Acquire an instantaneous state feature group based on the instantaneous monitoring data, wherein the instantaneous state feature group includes an instantaneous actual leakage amount, an instantaneous leakage velocity, an instantaneous pressure decay gradient, an instantaneous kinematic viscosity change residual, and an instantaneous temperature anomaly index; ​​The instantaneous state feature group is input into the Bayesian network model to obtain a hydraulic motor fault cause set.

[0010] Furthermore, a Bayesian network model is constructed, including: Performing multimodal data feature extraction on the historical monitoring data to obtain a historical leakage state feature group and a historical oil state feature group, wherein the historical leakage state feature group includes a historical actual leakage amount, a historical leakage velocity, and a historical pressure decay gradient, and the historical oil state feature group includes a historical kinematic viscosity change residual and a historical temperature anomaly index; By analyzing the historical leakage state feature group, the internal and external leakage of the hydraulic motor is diagnosed to obtain the internal and external leakage inducement set; By analyzing the historical oil state feature group, performing oil deterioration diagnosis on the hydraulic oil, and obtaining an oil deterioration inducement set; A Bayesian network model is constructed with the historical leakage state feature group and the historical oil state feature group as observation variables, and the internal and external leakage inducement set and the oil degradation inducement set as implicit variables.

[0011] According to a second aspect of the present disclosure, a device for detecting the working state of a hydraulic motor is provided, comprising: A multi-dimensional monitoring data set acquisition module is used to deploy sensor clusters at key nodes of the hydraulic motor, obtain multi-dimensional state parameters of the hydraulic motor in real time, pre-process the multi-dimensional state parameters of the hydraulic motor, and construct a multi-dimensional monitoring data set. The multi-dimensional monitoring data set includes flow data, speed data, pressure data, and temperature data; a global monitoring data set construction module, the global monitoring data set construction module being used to calculate the displacement data of the hydraulic motor and the kinematic viscosity data of the hydraulic oil according to the multi-dimensional monitoring data set to obtain the global monitoring data set; a state assessment and instruction generation module, the state assessment and instruction generation module being used to calculate the actual leakage of the hydraulic motor based on the global monitoring data set, assess the state of the hydraulic motor according to the actual leakage of the hydraulic motor, and generate an operation mode switching instruction, the operation mode switching instruction including a continuous operation instruction, a load monitoring instruction, and a shutdown maintenance instruction; a fault diagnosis module, the fault diagnosis module being configured to receive the load monitoring instruction or the shutdown maintenance instruction, initiate multimodal fault diagnosis, and obtain a set of hydraulic motor fault causes using a Bayesian network model; A maintenance plan determination module is used to determine a maintenance plan based on the hydraulic motor fault cause set, the maintenance plan including fault location information, spare parts list, process roadmap and estimated maintenance time.

[0012] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above-mentioned method for detecting the working state of a hydraulic motor when executed by a processor.

[0013] One or more technical solutions provided in this disclosure have at least the following technical effects or advantages: A sensor cluster is deployed at key nodes of the hydraulic motor to acquire multidimensional state parameters of the hydraulic motor in real time. These parameters are preprocessed to construct a multidimensional monitoring data set, including flow rate, speed, pressure, and temperature data. Based on this multidimensional monitoring data set, the hydraulic motor's displacement data and the kinematic viscosity of the hydraulic oil are calculated to obtain a global monitoring data set. The actual leakage of the hydraulic motor is calculated based on this global monitoring data set. The hydraulic motor's state is assessed based on this actual leakage, and an operating mode switching instruction is generated, including a continuous operation instruction, a load monitoring instruction, and a shutdown and maintenance instruction. Upon receiving the load monitoring instruction or shutdown and maintenance instruction, multimodal fault diagnosis is initiated, and a Bayesian network model is used to obtain a set of hydraulic motor fault causes. Based on this set of hydraulic motor fault causes, a maintenance plan is determined, including fault location information, a spare parts list, a process roadmap, and an estimated maintenance time. This solves the technical problems of existing technologies, such as lags, a lack of effective preventive detection measures, and an inability to promptly detect potential hydraulic motor problems, which can lead to high maintenance costs and difficulty in equipment recovery. The technical effect of effectively monitoring the hydraulic motor, discovering problems in time and reducing equipment maintenance costs is achieved.

[0014] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the present disclosure or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and a person of ordinary skill in the art can obtain other drawings based on the provided drawings without any creative work.

[0016] Figure 1 A flow chart of a method for detecting the working state of a hydraulic motor provided in an embodiment of the present application; Figure 2 A schematic structural diagram of a hydraulic motor working status detection device provided in an embodiment of the present application.

[0017] Explanation of reference numerals: multi-dimensional monitoring data set acquisition module 11 , global monitoring data set construction module 12 , status assessment and instruction generation module 13 , fault diagnosis module 14 , maintenance plan determination module 15 . DETAILED DESCRIPTION

[0018] The following description of exemplary embodiments of the present disclosure is provided in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0019] Embodiment 1: A method for detecting the working state of a hydraulic motor provided by the present disclosure is referred to Figure 1 For illustration, the methods include: S1: Deploy sensor clusters at key nodes of the hydraulic motor to obtain multi-dimensional state parameters of the hydraulic motor in real time, pre-process the multi-dimensional state parameters of the hydraulic motor, and construct a multi-dimensional monitoring data set, which includes flow data, speed data, pressure data, and temperature data; Specifically, based on the operating principle and structural characteristics of hydraulic motors, the inlet and outlet pipelines, shaft ends, and housing drain ports were identified as key monitoring nodes. A high-precision sensor cluster consisting of flow, speed, pressure, and temperature sensors was deployed to build a distributed data acquisition network. A time synchronization mechanism was established for the sensor cluster, using the IEEE 1588 precision time protocol to achieve sub-microsecond clock synchronization, addressing data time misalignment caused by internal sensor clock discrepancies. The sensor cluster acquisition frequency was set to 1 Hz. Each sensor collected raw data at this frequency, generating multiple sets of initial data sequences. An isolation forest algorithm was run on each initial data sequence to identify and remove outliers. After anomaly detection and interpolation repair, the initial data sequences were classified and integrated. Initial data sequences of the same category were integrated, aligned with the time axis, and the average value was taken to obtain the final data. This constructs a multidimensional monitoring dataset for condition monitoring and fault diagnosis.

[0020] S2: Calculating the displacement data of the hydraulic motor and the kinematic viscosity data of the hydraulic oil according to the multi-dimensional monitoring data set to obtain a global monitoring data set; Specifically, flow and speed data are obtained from the multidimensional monitoring data set, and the hydraulic motor displacement data is calculated according to the formula. The hydraulic oil kinematic viscosity data is obtained through temperature data, and a hydraulic oil viscosity-temperature relationship calculation model is constructed to determine the viscosity-temperature relationship calculation formula corresponding to each of the three types of hydraulic oil. Since the hydraulic oil viscosity has a hysteresis in its response to temperature, a first-order inertia link formula with a lag time constant is introduced to correct the kinematic viscosity data during the hydraulic motor startup and temperature rise phase. Initially, the hydraulic oil kinematic viscosity does not respond to temperature changes, but approaches the ideal value over time. Finally, the multidimensional monitoring data set is expanded based on the obtained displacement and kinematic viscosity data, and the timestamp is used as the index to ensure the synchronization of all data sampling times. Two columns are added to the original data set to store the displacement and kinematic viscosity data respectively, and a global monitoring data set is constructed to lay the data foundation for subsequent leakage calculations.

[0021] S3: Calculating the actual leakage of the hydraulic motor based on the global monitoring data set, assessing the state of the hydraulic motor according to the actual leakage of the hydraulic motor, and generating an operation mode switching instruction, wherein the operation mode switching instruction includes a continuous operation instruction, a load monitoring instruction, and a shutdown maintenance instruction; Specifically, the actual leakage of the hydraulic motor is calculated by combining displacement, pressure, and kinematic viscosity data from the global monitoring dataset. This data is then compared with the leakage baseline to determine the motor's current condition. Leakage baselines are primarily divided into a mild leakage baseline and a severe leakage baseline, representing the short-term fluctuation range and the threshold for irreversible damage, respectively. When the actual leakage is less than or equal to the mild leakage baseline, the hydraulic motor is operating healthily, indicating good equipment condition. At this point, the leakage is normal, with no material impact on equipment performance or safety, and a continuous operation instruction is generated. If the leakage is close to the mild leakage baseline but within the specified limits, the sampling frequency is increased to prevent breaches of the baseline and missed faults. If the actual leakage exceeds the mild leakage baseline but does not reach the severe leakage baseline, the hydraulic motor is considered to have an early-stage abnormality and a load monitoring instruction is generated. Although irreversible damage has not yet occurred, the safety margin has been exceeded, requiring derating and initiating multimodal fault diagnosis to identify the source of the fault. When the actual leakage exceeds the severe leakage baseline, the hydraulic motor has failed severely, and continued operation will damage the equipment. At this time, the leakage volume reaches a dangerous value. The power source must be cut off to avoid serious failure. A shutdown and maintenance instruction must be generated, the power supply must be cut off, the locking device must be activated, and multimodal diagnosis must be used to locate the fault point. Maintenance-related information will be pushed based on the results.

[0022] S4: receiving the load monitoring instruction or the shutdown maintenance instruction, starting multimodal fault diagnosis, and obtaining a hydraulic motor fault cause set using a Bayesian network model; Specifically, upon receiving a load monitoring command or a shutdown maintenance command, multimodal fault diagnosis is initiated, and the cause of the hydraulic motor fault is determined based on the existing monitoring data. A global monitoring dataset is obtained and partitioned, with the data collected at the last 10 time points considered as real-time monitoring data and the remaining data as historical monitoring data. Historical monitoring data for the hydraulic motor is obtained and a historical monitoring dataset is constructed. Multimodal feature extraction is performed on the historical monitoring dataset to generate a historical leakage state feature set and a historical oil state feature set. By analyzing and processing these historical leakage state feature sets and historical oil state feature sets, internal and external leakage and oil deterioration diagnosis of the hydraulic motor is completed, and a set of internal and external leakage and oil deterioration cause sets are obtained. A Bayesian network model for hydraulic motor fault diagnosis is constructed using these historical leakage state feature sets, historical oil state feature sets, internal and external leakage cause sets, and oil deterioration cause sets. The real-time state feature set is extracted from the real-time monitoring data and input into the Bayesian network model. The probability distribution of the fault cause is calculated, and the top three fault cause combinations are output as the hydraulic motor fault cause set.

[0023] S5: Determine a maintenance plan based on the hydraulic motor fault cause set, where the maintenance plan includes fault location information, a spare parts list, a process roadmap, and an estimated maintenance time.

[0024] Specifically, a set of hydraulic motor fault causes is obtained, and a maintenance plan is determined based on the detailed fault causes in the set. A maintenance plan is a comprehensive and systematic plan designed to provide precise and effective solutions for various possible hydraulic motor faults. It covers several key aspects, such as fault location information, spare parts lists, process roadmaps, and estimated repair times. The key to generating a maintenance plan lies in the fact that once the fault location is determined, it goes beyond superficial location analysis and requires detailed inspection of each specific location. This inspection requires the use of specialized testing equipment and methods to comprehensively and in-depth analyze the specific conditions at the fault location, such as the degree of wear, corrosion, component deformation, and other possible fault manifestations. Based on the results of this detailed inspection, and drawing on extensive experience and expertise, a targeted maintenance plan is developed for the faulty areas. The design of the process roadmap is a key component of the maintenance plan. This requires planning a scientific, rational, efficient, and feasible maintenance process based on multiple factors, including the equipment's structural characteristics, the type of fault, and the difficulty of the repair. Furthermore, an estimated repair time is an integral part of the maintenance plan. This requires comprehensive consideration of multiple factors, including the complexity of the fault, the difficulty of the repair operation, and the skill level of the maintenance personnel. Finally, all of this information is integrated into a report, including precise fault location information, a complete spare parts list, a detailed process roadmap, and a reasonable estimated repair time. This allows personnel to fully understand the overall repair plan, providing accurate guidance for subsequent maintenance work and strong data support for subsequent fault detection.

[0025] For example, if the hydraulic motor failure is primarily caused by internal leakage due to piston and cylinder wear, the fault is likely located in the piston and cylinder section of the hydraulic motor. A thorough inspection of the cylinder's inner wall wear is necessary. Using an endoscope or other inspection equipment, the wear location and extent can be precisely determined. Once the extent of wear is confirmed, a spare parts list can be compiled. If the cylinder is a monolithic structure, the entire cylinder should be prepared as a spare. If the cylinder is detachable, only the worn cylinder or piston components should be prepared, depending on the situation. A process roadmap should be developed, including: removing the hydraulic motor from the equipment according to operating procedures; disassembling it step by step down to the piston and cylinder components according to established procedures; using specialized tools to remove the old piston and cylinder and clean the installation area; strictly controlling the seal and installation accuracy when installing the new piston and cylinder; performing necessary post-installation commissioning operations such as pressure and flow tests; and reinstalling the hydraulic motor and performing a complete commissioning. Given the complexity of disassembly and installation of the plunger cylinder and the need for precision debugging, the estimated maintenance time is 8 to 12 hours. The actual duration is affected by the complexity of the equipment structure and the proficiency of the maintenance personnel.

[0026] Furthermore, step S1 of this application also includes: The hydraulic motor inlet and outlet pipes, shaft ends, and housing oil drain ports are selected as key nodes for deploying sensor clusters, including flow sensors, speed sensors, pressure sensors, and temperature sensors. The sensor cluster continuously collects flow data, speed data, pressure data, and temperature data of the hydraulic motor during operation to obtain multi-dimensional state parameters of the hydraulic motor, and adopts the IEEE 1588 precision time protocol to ensure synchronous capture of the multi-dimensional state parameters of the hydraulic motor; The isolation forest algorithm is used to remove outliers in the multidimensional state parameters of the hydraulic motor, and the cubic spline interpolation method is used to fill the missing data segments, reconstruct the continuous signal, and construct a multidimensional monitoring data set.

[0027] Specifically, based on the operating principle and structural characteristics of the hydraulic motor, the inlet and outlet pipelines, shaft ends, and case drain ports are identified as key monitoring nodes. A high-precision sensor cluster is deployed to build a distributed data acquisition network. The sensor cluster includes flow sensors, speed sensors, pressure sensors, and temperature sensors. The flow sensor is a pipeline-inductive electromagnetic flowmeter installed in the inlet and outlet pipelines of the hydraulic motor. The flow sensor incorporates an annular clamp-type mounting base and an embedded non-contact Hall-effect sensor, which captures flow waveforms in real time and enables precise monitoring of hydraulic motor flow. The speed sensor is a photoelectric speed sensor installed at the shaft end of the hydraulic motor. This speed sensor detects changes in optical signals at the shaft end to determine the motor's speed. The pressure sensor is a silicon piezoresistive pressure sensor, mounted on the pressure measurement manifold of the inlet and outlet pipelines with an O-ring seal. The temperature sensor is a platinum resistance temperature sensor, embedded in the heat conduction groove on the outer wall of the case drain port pipeline.

[0028] A sensor cluster time synchronization mechanism was established, using the IEEE 1588 Precision Time Protocol to achieve sub-microsecond clock synchronization across the sensor cluster, resolving data time misalignment caused by internal clock discrepancies between sensors. The IEEE 1588 Precision Time Protocol is a protocol for achieving high-precision time synchronization in Ethernet networks. It operates by establishing a master-slave hierarchy, with master and slave clocks located in a central controller and sensors. The master clock sends a synchronization signal to all sensors every second, and the sensors immediately calibrate their own clocks upon receiving the signal. This method allows each sensor to calibrate its local clock offset, achieving time and frequency synchronization and ensuring time alignment errors of less than 10μs for flow, speed, pressure, and temperature data.

[0029] The sensor cluster acquisition frequency is set to 1Hz based on the hydraulic motor's operating characteristics and fault detection requirements. Although flow and temperature change slowly during hydraulic motor operation, in this application, all data must be aligned on the time axis to provide a consistent data foundation for subsequent calculations. Therefore, to avoid data gaps, all data must be collected at the same frequency as the most frequently collected pressure and flow data.

[0030] Each sensor continuously acquires raw data at a specific acquisition frequency. This data is arranged chronologically to generate an initial data sequence corresponding to each sensor. Because the sensor cluster includes not only four sensor types, but also a variable number of sensors in each type, the initial data sequences specifically include multiple data sets, such as initial flow data sequence 1, initial flow data sequence 2, initial speed data sequence 1, initial pressure data sequence 1, initial pressure data sequence 2, initial temperature data sequence 1, and initial temperature data sequence 2. The isolation forest algorithm is run separately on each sequence to identify and remove outliers. The specific operation involves first setting the time window length and sliding step size based on the data characteristics. For example, for pressure data, the window length is 10 seconds, the window sliding step is 5 seconds, and 50% overlap is achieved. This operation avoids memory overflow caused by processing large amounts of data at once. The initial data sequences collected by each sensor are individually normalized and converted into a standardized format to eliminate the impact of dimensional differences on the algorithm and enhance its robustness. Next, an isolation forest model is independently constructed for each initial data series. During model construction, 200 isolation trees are trained in parallel, with 256 randomly selected samples from each tree to construct a decision path. Overfitting is prevented by limiting the maximum tree depth. This process randomly partitions the feature space, resulting in longer paths for normal data points to be isolated, while outliers, due to their significant deviation from the population distribution, can be quickly identified at shallower nodes. Third, based on the constructed isolation forest model, the system quantifies the probability of anomaly for each data point. By traversing all isolation trees, the average path length required to isolate the target data point is calculated. Normal data, due to their dense distribution, typically requires multiple node partitions to isolate, resulting in longer paths. However, outliers, due to their distance from the main data population, require only a few partitions to isolate, resulting in significantly shorter paths. The average path length is calculated and used to identify outliers. Finally, data cleaning is performed. Data points identified as outliers are directly marked as invalid values ​​(NaN), while their position information is retained in the time series. Simultaneously, a cubic spline interpolation algorithm is initiated to reconstruct missing segments using preceding and following normal data. For example, when signal interference causes pressure data acquisition errors, the system automatically calls back to normal pressure data after marking all five abnormal data points, generating a smooth transition curve to ensure time-series continuity for subsequent analysis. A detailed log is generated for all processing steps, including the number of abnormal data points and the corrective action, providing a traceable data foundation for equipment health management.

[0031] After anomaly detection and interpolation repair, the initial data sequences are classified and integrated. Initial data sequences of the same category are integrated, with the time axis as the alignment standard. The final data is the average of all sensor data recorded at that time. For example, at 10:00, the pressure data collected by initial pressure data sequence 1 and initial pressure data sequence 2 are 19.8 MPa and 20.0 MPa, respectively, so the final pressure data is 19.9 MPa. This method obtains integrated flow data, speed data, pressure data, and temperature data, constructing a multidimensional monitoring data set for condition monitoring and fault diagnosis.

[0032] Furthermore, step S2 of this application also includes: Acquiring displacement data of the hydraulic motor based on the flow data and speed data in the multi-dimensional monitoring data set; The hydraulic oil viscosity-temperature relationship calculation model is set according to the type of hydraulic oil. The specific formula is: ISO VG32: η=144-3.2T; ISO VG48: η=107.5-1.25T; ISO VG68: η=159-2.4T; Among them, ISO VG32, ISO VG48, and ISO VG68 are three types of hydraulic oil commonly used in hydraulic motors. η represents the kinematic viscosity of the hydraulic oil, and T represents the temperature data. Input the temperature data in the multi-dimensional monitoring data set to obtain the corresponding kinematic viscosity data of the hydraulic oil; The displacement data, kinematic viscosity data and multi-dimensional monitoring data set are integrated to construct a global monitoring data set.

[0033] Specifically, a multi-dimensional monitoring data set is obtained, and the flow data and speed data in the multi-dimensional monitoring data set are extracted to calculate the displacement data of the hydraulic motor. The specific formula is: , where D is the displacement data, Q is the flow data, N is the speed data, and ω is the volumetric efficiency. The volumetric efficiency is determined by the hydraulic motor model. Due to the mechanical friction and hysteresis effect in actual working conditions, which will slightly reduce the effective speed, the displacement data needs to be corrected in combination with the volumetric efficiency. The kinematic viscosity data of the hydraulic oil is obtained based on the temperature data in the multi-dimensional monitoring data set. A hydraulic oil viscosity-temperature relationship calculation model is constructed, and the viscosity-temperature relationship calculation formulas corresponding to the three types of hydraulic oil are determined. It should be noted that in the hydraulic system, the response of hydraulic oil viscosity to temperature is not instantaneous. When the temperature changes, it takes a certain amount of time for the molecular structure within the oil to adjust, causing the viscosity change to lag behind the temperature change. Therefore, during the startup of the hydraulic system, the temperature gradually rises, and the calculated kinematic viscosity data needs to be corrected. However, during the stable operation of the hydraulic motor, since the temperature change is no longer significant, the kinematic viscosity can be corrected. In view of the temperature rise process during the startup of the hydraulic motor, a first-order inertia link is introduced to describe this dynamic process with hysteresis characteristics, so as to achieve the correction of the hydraulic oil viscosity. The specific formula is: , where η actual is the corrected kinematic viscosity data, η is the kinematic viscosity calculated based on the viscosity-temperature relationship calculation model, e is the natural constant, t is the time, and μ is the lag time constant, which characterizes the lag degree of the oil viscosity response to temperature changes. When t=0, =1, then η actual =0, which means that at the initial moment, the kinematic viscosity of the hydraulic oil has not yet responded to the temperature change. As time t increases, Gradually decreases, η actual gradually approaches η. When t is large enough, is approximately equal to 0, at this time η actual Approximately equal to η, which means that after a long enough time, the oil viscosity will eventually reach the ideal model-calculated value.

[0034] Based on the displacement and kinematic viscosity data calculated using the above formulas, the multidimensional monitoring dataset was expanded. Using timestamps as indexes to ensure strict synchronization of sampling times for all data sequences, two new columns were added to the original multidimensional dataset to store displacement and kinematic viscosity data, respectively. This global monitoring dataset was constructed, providing the data foundation for subsequent leakage calculations.

[0035] Furthermore, step S3 of this application also includes: The actual leakage of the hydraulic motor is calculated based on the global monitoring data set. The specific formula is: ; Among them, Q Lrepresents the actual leakage of the hydraulic motor, X represents the leakage coefficient, the specific value is 0.36, Q represents the displacement data of the hydraulic motor, P represents the pressure data of the hydraulic motor, and η represents the kinematic viscosity data of the hydraulic oil; The actual leakage of the hydraulic motor is compared with the leakage baseline to establish a three-level evaluation system. The leakage baseline includes a mild leakage baseline M and a severe leakage baseline N, wherein the mild leakage baseline M is 3%Q / 100bar and the severe leakage baseline N is 5%Q / 100bar; When 0 L ≤N, the hydraulic motor is judged to be in the first-level health state, and a continuous operation instruction is generated. The current leakage volume is within the allowable fluctuation range, the hydraulic motor is in good condition, and the regular inspection cycle is continued; When N L ≤M, the hydraulic motor is judged to be in the second-level warning state and a load monitoring instruction is generated. At this time, the hydraulic motor leakage exceeds the safety margin and needs to be derated to 85% of the rated power. Multi-modal fault diagnosis is also started simultaneously; When Q L >M, the hydraulic motor is judged to be in the third-level fault state and a shutdown and maintenance instruction is generated. At this time, the hydraulic motor is leaking seriously and needs to be stopped immediately for multi-modal fault diagnosis.

[0036] Specifically, displacement data, pressure data, and kinematic viscosity data from the global monitoring dataset are collected and combined with a predefined formula derived from extensive experimental data to calculate the actual leakage of the hydraulic motor. A leakage baseline is established, a dynamic reference value that quantifies the safety threshold for hydraulic motor leakage. The leakage baseline is divided into a mild leakage baseline (M) and a severe leakage baseline (N), with specific values ​​of M = 3%Q / 100bar and N = 5%Q / 100bar. The mild leakage baseline (M) is defined as the allowable short-term fluctuation range, while the severe leakage baseline (N) is defined as the potential for irreversible damage to the hydraulic motor.

[0037] When the actual leakage volume is less than or equal to the mild leakage baseline, the hydraulic motor is in a healthy operating state. In this state, the leakage volume is within the fluctuation range allowed by the equipment design, indicating that the sealing system is intact, the oil viscosity is stable, and there are no significant signs of wear or fatigue. At this point, it is determined that the current leakage volume has no substantial impact on equipment performance and safety. No operating parameter adjustment is required. A continuous operation instruction is generated to maintain rated power output, the original inspection cycle is maintained, and a periodic health report is generated to record trend data. However, if the leakage volume is extremely close to the mild leakage baseline and does not exceed the limit, the data sampling frequency must be increased to prevent the mild leakage baseline from being breached and a fault from being missed.

[0038] ​​When the actual leakage exceeds the mild leakage baseline but falls short of the severe leakage baseline, it indicates an early-stage hydraulic motor anomaly, potentially caused by minor wear of internal components, oil contamination, or localized overheating. Immediate intervention is required to prevent further deterioration. At this point, the leakage exceeds the safety margin but has not yet reached the irreversible damage threshold. Risk reduction requires derating. A load monitoring command is generated, limiting output power to 85% of the rated value and reducing system pressure to slow the leakage. Multimodal fault diagnosis is simultaneously initiated to locate the potential source of the fault. The data sampling frequency is increased to 10 times per second to track leakage trends in real time.

[0039] When the actual leakage exceeds the severe leakage baseline, it indicates a serious failure of the hydraulic motor, such as excessive wear of the plunger-cylinder clearance or abnormal scratching of the oil distribution plate. Continued operation will lead to the spread of oil contamination, a sudden drop in efficiency, and even equipment damage. At this point, the leakage is considered to have reached the dangerous threshold, and the power source must be immediately shut down to avoid catastrophic failure. A shutdown and maintenance instruction is generated, shutting off the power to the hydraulic pump and activating the safety lockout device to prevent inertial slip. Multimodal fault diagnosis is then used to locate the fault point, and based on the diagnostic results, a spare parts list, process roadmap, and estimated repair time are provided.

[0040] Furthermore, step S4 of this application also includes: Obtain the global monitoring data set, divide the global monitoring data set, regard the data collected at the last 10 time points as real-time monitoring data, and regard the remaining data as historical monitoring data; Constructing a Bayesian network model based on the historical monitoring data; Acquire an instantaneous state feature group based on the instantaneous monitoring data, wherein the instantaneous state feature group includes an instantaneous actual leakage amount, an instantaneous leakage velocity, an instantaneous pressure decay gradient, an instantaneous kinematic viscosity change residual, and an instantaneous temperature anomaly index; The instantaneous state feature group is input into the Bayesian network model to obtain a hydraulic motor fault cause set.

[0041] Specifically, a Bayesian network model was constructed based on historical monitoring data and fault detection records of the hydraulic motor. Real-time monitoring data was obtained from the global monitoring dataset, and key features of this real-time monitoring data were extracted to construct a real-time state feature set, which served as the input data for the Bayesian network. The data in the real-time state feature set was obtained in the same manner as the corresponding data in the historical leakage state feature set and the historical oil state feature set.

[0042] The data from the immediate state feature set is input into the Bayesian network model to calculate the probability distribution of the fault causes. Each data point in the immediate state feature set is mapped to a leaf node of the Bayesian network, activating the associated fault hypothesis. A junction tree algorithm is used to calculate the posterior probability distribution, outputting the top three fault causes by probability. These causes are then combined into a set of hydraulic motor fault causes.

[0043] Furthermore, step S4 of this application also includes: Performing multimodal data feature extraction on the historical monitoring data to obtain a historical leakage state feature group and a historical oil state feature group, wherein the historical leakage state feature group includes a historical actual leakage amount, a historical leakage velocity, and a historical pressure decay gradient, and the historical oil state feature group includes a historical kinematic viscosity change residual and a historical temperature anomaly index; By analyzing the historical leakage state feature group, the internal and external leakage of the hydraulic motor is diagnosed to obtain the internal and external leakage inducement set; By analyzing the historical oil state feature group, performing oil deterioration diagnosis on the hydraulic oil, and obtaining an oil deterioration inducement set; A Bayesian network model is constructed with the historical leakage state feature group and the historical oil state feature group as observation variables, and the internal and external leakage inducement set and the oil degradation inducement set as implicit variables.

[0044] Specifically, historical monitoring data for hydraulic motors was obtained from the global monitoring dataset to construct a historical monitoring dataset. This dataset includes historical flow, speed, pressure, temperature, displacement, and kinematic viscosity data. Multimodal feature extraction was performed on this historical monitoring dataset to obtain historical actual leakage, leakage velocity, pressure decay gradient, kinematic viscosity change residual, and temperature anomaly index.

[0045] Specifically, the historical actual leakage Q leak The calculation method is consistent with the actual leakage of the hydraulic motor mentioned above. The remaining data are obtained as follows: The formula for calculating the historical leakage rate is: , where V represents the historical leakage rate, ΔQ Leak represents the difference in actual leakage, indicating the actual leakage value during the two acquisition times, and ΔT represents the time difference; The formula for calculating the historical pressure decay gradient is: , where B represents the historical pressure decay gradient, ΔP represents the pressure difference, and ΔT represents the time difference; The equation of the line is constructed by fitting the long-term trend line of the historical kinematic viscosity data using the least squares method Calculate the residual of the historical kinematic viscosity change, where y is the fitted historical kinematic viscosity data, a is the slope, x is the index value corresponding to the historical kinematic viscosity data sorted by timestamp, and b is the intercept, which represents the initial fitted historical kinematic viscosity data. Based on all the historical kinematic viscosity data currently obtained, the mean of x and y is obtained. and , and calculate the slope a of the equation through two mean data. The specific formula is , where xi represents the i-th index value, yi represents the i-th historical kinematic viscosity data, Represents the mean of all index values, Represents the mean of all historical kinematic viscosity data. 、 , the intercept b of the straight line equation is derived from the data a, specifically According to the constructed linear equation, the fitted historical kinematic viscosity data can be calculated, and the residual C of the historical kinematic viscosity change can be calculated based on this, which is specifically expressed as C=η-y, where η is the original historical kinematic viscosity data and y is the fitted historical kinematic viscosity data; The formula for calculating the historical temperature anomaly index is: , where T anomaly represents the total number of temperature anomalies, i is the index value corresponding to the historical temperature data sorted by timestamp, n is the total number of historical temperature data, T i Represents the temperature value corresponding to each acquisition time, T threshold Represents the preset safety temperature threshold, such as 90°C, I (T i >T threshold ) is an exponential function used to determine whether the current temperature exceeds the threshold. If it exceeds, it is 1, otherwise it is 0.

[0046] The historical actual leakage volume, historical leakage velocity, and historical pressure decay gradient data are integrated into the historical leakage state feature group, while the historical kinematic viscosity change residual and historical temperature anomaly index are integrated into the historical oil state feature group.

[0047] By analyzing the historical leakage status feature group, the internal and external leakage of the hydraulic motor is diagnosed and the leakage fault causes are statistically analyzed. leak If the flow rate is continuously greater than 5L / min, while V fluctuates slightly and B decreases slowly, it is judged to be an internal leakage. The specific reasons include wear of the plunger cylinder, scratches on the distribution plate, fatigue deformation of the sliding shoe, etc. leak A sudden increase in V, with an instantaneous peak greater than 0.1 L / s, and a rapid drop in B indicates external leakage. Specific causes include loose pipe joints, aging or damage to O-rings, etc. All internal and external leakage fault causes are combined into an internal and external leakage cause set.

[0048] By analyzing the historical oil state feature group, the hydraulic oil is diagnosed for oil degradation and all oil degradation causes are counted. The historical kinematic viscosity change residual and the historical temperature anomaly index are analyzed to obtain the type of hydraulic oil fault. When the residual is a persistent positive residual and T anomaly When it is greater than 10, it is judged that the hydraulic oil has undergone oxidation and deterioration. At this time, the oxidation products increase the oil viscosity, causing the actual viscosity to increase. anomaly If the residual error is less than 3, the hydraulic oil is suspected of water contamination. This causes localized temperature measurement distortion and instability in theoretically calculated values. If the residual error exhibits periodic negative spikes, the hydraulic oil is suspected of particulate contamination. Particles accumulate, clogging the sensor and causing low temperature measurements. If the residual error trend shifts from positive to negative, the additive is depleted or ineffective, altering the viscosity-temperature characteristics and gradually invalidating the theoretical model. All causes of oil degradation are integrated into the oil degradation cause set.

[0049] A Bayesian network model for hydraulic motor fault diagnosis is constructed based on the historical leakage state feature group, the historical oil state feature group, the internal and external leakage inducement set, and the oil deterioration inducement set.

[0050] The historical actual leakage volume, historical leakage velocity, historical pressure decay gradient, historical kinematic viscosity change residual, and historical temperature anomaly index are set as observed variables (leaf nodes), internal leakage, external leakage, and oil degradation are set as intermediate variables, and piston cylinder wear, distribution plate scratches, sliding shoe fatigue deformation, pipe joint loosening, O-ring aging or damage, oxidation deterioration, water contamination, and additive depletion or failure are set as implicit variables (root nodes).

[0051] Construct a directed acyclic graph based on the previously determined causal logical relationship between each observed variable and the intermediate variable and implicit variable, such as Q leak Continuously greater than 5L / min → internal leakage → piston cylinder wear, Q leak Continuously exceeding 5L / min → internal leakage → valve plate scratches. When constructing a directed acyclic graph, each node's parent node should include all its direct causes, while ensuring there are no loops in the graph. A conditional probability table is set based on historical maintenance and inspection data, defining the conditional probability of each node under its parent node state. For combinations of parent nodes not found in the data, Laplace smoothing is used to avoid zero probability, assuming each conditional probability occurs at least once. A Bayesian network model is constructed using the directed acyclic graph and conditional probability table.

[0052] In the second embodiment, based on the same inventive concept as the method for detecting the working state of a hydraulic motor in the above embodiment, the present application also provides a device for detecting the working state of a hydraulic motor, as shown in the attached Figure 2 , the device comprises: A multi-dimensional monitoring data set acquisition module 11 is used to deploy sensor clusters at key nodes of the hydraulic motor, obtain multi-dimensional state parameters of the hydraulic motor in real time, pre-process the multi-dimensional state parameters of the hydraulic motor, and construct a multi-dimensional monitoring data set, wherein the multi-dimensional monitoring data set includes flow data, speed data, pressure data, and temperature data; A global monitoring data set construction module 12, configured to calculate the displacement data of the hydraulic motor and the kinematic viscosity data of the hydraulic oil according to the multi-dimensional monitoring data set to obtain a global monitoring data set; a state assessment and instruction generation module 13, configured to calculate an actual leakage of the hydraulic motor based on the global monitoring data set, assess the state of the hydraulic motor according to the actual leakage, and generate an operation mode switching instruction, wherein the operation mode switching instruction includes a continuous operation instruction, a load monitoring instruction, and a shutdown maintenance instruction; A fault diagnosis module 14 is configured to receive the load monitoring instruction or the shutdown maintenance instruction, initiate multimodal fault diagnosis, and obtain a set of hydraulic motor fault causes using a Bayesian network model; The maintenance plan determination module 15 is used to determine a maintenance plan according to the hydraulic motor fault cause set, and the maintenance plan includes fault location information, a spare parts list, a process roadmap, and an estimated maintenance time.

[0053] Based on the same inventive concept, an embodiment further provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of a method for detecting the working state of a hydraulic motor as described in the above embodiment are implemented.

[0054] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0055] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting the working state of a hydraulic motor, characterized in that: The method comprises: Deploy sensor clusters at key nodes of the hydraulic motor to obtain multi-dimensional state parameters of the hydraulic motor in real time, pre-process the multi-dimensional state parameters of the hydraulic motor, and construct a multi-dimensional monitoring data set, which includes flow data, speed data, pressure data, and temperature data; Calculating the displacement data of the hydraulic motor and the kinematic viscosity data of the hydraulic oil according to the multi-dimensional monitoring data set to obtain a global monitoring data set; Calculating the actual leakage of the hydraulic motor based on the global monitoring data set, assessing the state of the hydraulic motor according to the actual leakage of the hydraulic motor, and generating an operation mode switching instruction, wherein the operation mode switching instruction includes a continuous operation instruction, a load monitoring instruction, and a shutdown maintenance instruction; receiving the load monitoring instruction or the shutdown maintenance instruction, starting multimodal fault diagnosis, and obtaining a set of hydraulic motor fault causes using a Bayesian network model; A maintenance plan is determined according to the hydraulic motor fault cause set, wherein the maintenance plan includes fault location information, a spare parts list, a process route map, and an estimated maintenance time.

2. A method for detecting the working state of a hydraulic motor according to claim 1, characterized in that: Construct a multidimensional monitoring dataset, including: The hydraulic motor inlet and outlet pipes, shaft ends, and housing oil drain ports are selected as key nodes for deploying sensor clusters, including flow sensors, speed sensors, pressure sensors, and temperature sensors. The sensor cluster continuously collects flow data, speed data, pressure data, and temperature data of the hydraulic motor during operation to obtain multi-dimensional state parameters of the hydraulic motor, and adopts the IEEE 1588 precision time protocol to ensure synchronous capture of the multi-dimensional state parameters of the hydraulic motor; The isolation forest algorithm is used to remove outliers in the multidimensional state parameters of the hydraulic motor, and the cubic spline interpolation method is used to fill the missing data segments, reconstruct the continuous signal, and construct a multidimensional monitoring data set.

3. A method for detecting the working state of a hydraulic motor according to claim 1, characterized in that: Obtain global monitoring data sets, including: Acquiring displacement data of the hydraulic motor based on the flow data and speed data in the multi-dimensional monitoring data set; The hydraulic oil viscosity-temperature relationship calculation model is set according to the type of hydraulic oil. The specific formula is: ISO VG32: η=144-3.2T; ISO VG48: η=107.5-1.25T; ISO VG68: η=159-2.4T; Among them, ISO VG32, ISO VG48, and ISO VG68 are three types of hydraulic oil commonly used in hydraulic motors. η represents the kinematic viscosity of the hydraulic oil, and T represents the temperature data. Input the temperature data in the multi-dimensional monitoring data set to obtain the corresponding kinematic viscosity data of the hydraulic oil; The displacement data, kinematic viscosity data and multi-dimensional monitoring data set are integrated to construct a global monitoring data set.

4. A method for detecting the working state of a hydraulic motor according to claim 1, characterized in that: Generate operating mode switching instructions, including: The actual leakage of the hydraulic motor is calculated based on the global monitoring data set. The specific formula is: ； Among them, Q L represents the actual leakage of the hydraulic motor, X represents the leakage coefficient, the specific value is 0.36, Q represents the displacement data of the hydraulic motor, P represents the pressure data of the hydraulic motor, and η represents the kinematic viscosity data of the hydraulic oil; The actual leakage of the hydraulic motor is compared with the leakage baseline to establish a three-level evaluation system. The leakage baseline includes a mild leakage baseline M and a severe leakage baseline N, wherein the mild leakage baseline M is 3%Q / 100bar and the severe leakage baseline N is 5%Q / 100bar; When 0 L ≤N, the hydraulic motor is judged to be in the first-level health state, and a continuous operation instruction is generated. The current leakage volume is within the allowable fluctuation range, the hydraulic motor is in good condition, and the regular inspection cycle is continued;​ When N L ≤M, the hydraulic motor is judged to be in the second-level warning state and a load monitoring instruction is generated. At this time, the hydraulic motor leakage exceeds the safety margin and needs to be derated to 85% of the rated power. Multi-modal fault diagnosis is also started simultaneously;​ When Q L >M, the hydraulic motor is judged to be in the third-level fault state and a shutdown and maintenance instruction is generated. At this time, the hydraulic motor is leaking seriously and needs to be stopped immediately for multi-modal fault diagnosis.

5. A method for detecting the working state of a hydraulic motor according to claim 1, characterized in that: Receive the load monitoring instruction or shutdown maintenance instruction, start multimodal fault diagnosis, and use the Bayesian network model to obtain the hydraulic motor fault cause set, including: Obtain the global monitoring data set, divide the global monitoring data set, regard the data collected at the last 10 time points as real-time monitoring data, and regard the remaining data as historical monitoring data; Constructing a Bayesian network model based on the historical monitoring data; Acquire an instantaneous state feature group based on the instantaneous monitoring data, wherein the instantaneous state feature group includes an instantaneous actual leakage amount, an instantaneous leakage velocity, an instantaneous pressure decay gradient, an instantaneous kinematic viscosity change residual, and an instantaneous temperature anomaly index; The instantaneous state feature group is input into the Bayesian network model to obtain a hydraulic motor fault cause set.

6. A method for detecting the working state of a hydraulic motor according to claim 5, characterized in that: Build a Bayesian network model, including: Performing multimodal data feature extraction on the historical monitoring data to obtain a historical leakage state feature group and a historical oil state feature group, wherein the historical leakage state feature group includes a historical actual leakage amount, a historical leakage velocity, and a historical pressure decay gradient, and the historical oil state feature group includes a historical kinematic viscosity change residual and a historical temperature anomaly index; By analyzing the historical leakage state feature group, the internal and external leakage of the hydraulic motor is diagnosed to obtain the internal and external leakage inducement set; By analyzing the historical oil state feature group, performing oil deterioration diagnosis on the hydraulic oil, and obtaining an oil deterioration inducement set; A Bayesian network model is constructed with the historical leakage state feature group and the historical oil state feature group as observation variables, and the internal and external leakage inducement set and the oil degradation inducement set as implicit variables.

7. A hydraulic motor working state detection device, characterized in that: The device is used to implement a method for detecting the working state of a hydraulic motor according to any one of claims 1 to 6, and the device comprises: A multi-dimensional monitoring data set acquisition module is used to deploy sensor clusters at key nodes of the hydraulic motor, obtain multi-dimensional state parameters of the hydraulic motor in real time, pre-process the multi-dimensional state parameters of the hydraulic motor, and construct a multi-dimensional monitoring data set. The multi-dimensional monitoring data set includes flow data, speed data, pressure data, and temperature data; a global monitoring data set construction module, the global monitoring data set construction module being used to calculate the displacement data of the hydraulic motor and the kinematic viscosity data of the hydraulic oil according to the multi-dimensional monitoring data set to obtain the global monitoring data set; a state assessment and instruction generation module, the state assessment and instruction generation module being used to calculate the actual leakage of the hydraulic motor based on the global monitoring data set, assess the state of the hydraulic motor according to the actual leakage of the hydraulic motor, and generate an operation mode switching instruction, the operation mode switching instruction including a continuous operation instruction, a load monitoring instruction, and a shutdown maintenance instruction; a fault diagnosis module, the fault diagnosis module being configured to receive the load monitoring instruction or the shutdown maintenance instruction, initiate multimodal fault diagnosis, and obtain a set of hydraulic motor fault causes using a Bayesian network model; A maintenance plan determination module is used to determine a maintenance plan based on the hydraulic motor fault cause set, the maintenance plan including fault location information, spare parts list, process roadmap and estimated maintenance time.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, a method for detecting the working state of a hydraulic motor according to any one of claims 1 to 6 is implemented.

Images