High-end equipment pump motor high-fidelity state evaluation method

Abstract

The application discloses a high-end equipment pump motor high-fidelity state evaluation method, comprising the following steps: sequentially constructing an ideal mathematical model, an internal leakage amount mechanism model and an actual output torque calculation model, obtaining theoretical performance parameters, dynamic volumetric efficiency and torque benchmarks; based on health state test data, adopting a multi-objective optimization algorithm to inverse model undetermined parameters, and establishing an individualized health benchmark model; inputting real-time performance signals into the model, calculating the residual error of measured data and model response, extracting a dimensionless health feature vector through normalization processing, obtaining a comprehensive health index through weighted fusion, and realizing precise quantitative evaluation of the health state of the pump motor.

Description

Technical Field

[0001] This invention belongs to the field of condition monitoring of fluid transmission equipment, and in particular relates to a high-fidelity condition assessment method for pump motors in high-end equipment. Background Technology

[0002] As the core power unit of hydraulic static transmission systems, pump motors, with their advantages of high power density, compact structure, and significant efficiency, are widely used in national strategic fields such as aerospace, polar scientific research equipment, high-end intelligent manufacturing, and military special equipment. Existing technologies for assessing the health status of pump motors mainly fall into three categories: first, monitoring methods based on vibration spectrum analysis or fixed efficiency thresholds, which judge anomalies by setting empirical thresholds; second, pure data-driven methods based on deep learning, which use a large amount of historical data to train models to achieve state recognition; and third, evaluation methods based on fixed-parameter physical mechanism models, which calculate performance benchmarks through theoretical formulas. These methods can achieve basic monitoring functions under normal operating conditions, but their applicability is significantly limited under the highly coupled, time-varying, and non-stationary operating conditions faced by high-end equipment, such as high and low temperature cycles, alternating variable speed and load, and long-term low-temperature heavy loads.

[0003] However, existing technologies suffer from the following prominent problems: Traditional monitoring methods based on vibration spectra or fixed thresholds cannot distinguish between normal parameter drift induced by operating conditions and actual physical degradation under varying conditions. They misjudge fluid pulsation noise and normal leakage fluctuations as fault signals, leading to frequent false alarms and failing to meet the stringent "zero false alarm" requirements of scenarios such as aerospace on-orbit operation and polar scientific research. Pure data-driven methods based on deep learning rely on massive amounts of full lifecycle sample data, while pumps and motors in high-end equipment are mostly customized products with scarce full lifecycle data and incomplete operating condition coverage. The generalization ability of the models is severely insufficient, and the black-box models lack physical interpretability, making it impossible to locate degradation sources and support accurate operation and maintenance decisions. Traditional physical mechanism models use fixed empirical parameters and cannot dynamically follow the actual operating condition evolution, such as viscosity changes caused by oil temperature changes and clearance changes caused by wear. This results in a significant deviation between theoretical health benchmarks and actual physical processes, rendering them unreliable as a reliable assessment reference. These problems mean that the accuracy, reliability, and engineering practicality of existing methods under complex operating conditions cannot meet the requirements of high-end equipment for pump and motor condition assessment. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a high-fidelity condition assessment method for pump motors in high-end equipment, comprising:

[0005] Based on the structural parameters and operating condition data of the pump motor, an ideal mathematical model of the pump motor is constructed to obtain the theoretical performance parameters of the pump motor.

[0006] Based on the theoretical performance parameters, a mechanism model for internal leakage of the pump motor is constructed to obtain the dynamic volumetric efficiency.

[0007] Based on the dynamic volumetric efficiency and friction loss mechanism, a calculation model for the actual output torque of the pump motor is constructed to obtain the theoretical output torque benchmark value.

[0008] Based on the test data of the pump motor under healthy conditions, a multi-objective optimization algorithm is used to invert the undetermined parameters in the internal leakage mechanism model and the actual output torque calculation model of the pump motor to obtain an individualized health benchmark model of the pump motor.

[0009] The real-time collected pump motor performance signals are input into the pump motor individualized health benchmark model to obtain the model response output;

[0010] The residual between the real-time collected pump motor operation data and the model response output is normalized to obtain a dimensionless health feature vector.

[0011] The dimensionless health feature vector is weighted and fused to obtain a comprehensive health index. The health status of the pump motor is evaluated based on the comprehensive health index to obtain the evaluation result of the pump motor health status.

[0012] Preferably, the process of constructing an ideal pump motor mathematical model and obtaining the theoretical performance parameters of the pump motor includes:

[0013] Based on the maximum displacement of the variable pump, the swashplate angle, and the input speed, construct the equations for the pump's input and output flow rates;

[0014] Based on the motor displacement and input speed, construct the equations for the motor's input flow rate and output flow rate;

[0015] Based on the pump's volumetric efficiency and speed transmission ratio, an equation for calculating the motor speed is constructed.

[0016] Based on the pressure difference across the motor and the motor displacement, an equation for calculating the motor torque is constructed.

[0017] Based on the flow-pressure gradient of the check valve and the replenishment pressure, a replenishment flow equation is constructed.

[0018] Based on the principle of flow continuity on the high-pressure side and the low-pressure side, flow balance differential equations are constructed respectively;

[0019] By combining the input and output flow equations of the pump, the input and output flow equations of the motor, the motor speed calculation equation, the motor torque calculation equation, the oil replenishment flow equation, and the flow balance differential equation, an ideal pump motor mathematical model is constructed, and the theoretical performance parameters of the pump motor are obtained.

[0020] The theoretical performance parameters of the pump motor include system pressure distribution, dynamic flow characteristics, output speed and torque.

[0021] Preferably, the process of constructing a mechanism model for internal leakage of the pump motor and obtaining dynamic volumetric efficiency includes:

[0022] Based on oil temperature, oil pressure, viscosity-temperature characteristic parameters and pressure viscosity coefficient, a dynamic equivalent viscosity calculation model for oil is constructed to obtain real-time dynamic viscosity.

[0023] Based on the geometric conductance, shear conductance, effective clearance, inlet pressure and outlet pressure of the mating pair, a leakage calculation model for the mating pair is constructed based on the gap flow theory.

[0024] Based on the sum of the leakage amounts of the plunger and cylinder bore mating pairs, the slipper and swashplate mating pairs, and the distributor plate and cylinder block end face mating pairs, a model of the total leakage amount inside the pump motor is constructed, and the dynamic volumetric efficiency is calculated based on the total leakage amount and the theoretical flow rate.

[0025] Preferably, the process of constructing a dynamic equivalent viscosity calculation model for oil includes:

[0026] Determine the viscosity-temperature characteristic parameters and viscosity-temperature coefficient based on the type of oil;

[0027] The pressure viscosity coefficient is determined based on the oil pressure to correct for the viscosity increase effect under high load;

[0028] The real-time dynamic viscosity is calculated based on the real-time collected oil temperature, high-pressure side oil pressure, viscosity-temperature characteristic parameters, pressure viscosity coefficient, and viscosity-temperature coefficient.

[0029] Preferably, the process of constructing a calculation model for the actual output torque of the pump motor and obtaining the theoretical output torque reference value includes:

[0030] Based on the pressure difference across the motor, the motor displacement, and the dynamic volumetric efficiency, an ideal output torque equation is constructed.

[0031] Based on real-time dynamic viscosity, motor speed, Coulomb friction coefficient and static friction parameters, a hybrid friction torque model including viscous friction, Coulomb friction and static friction is constructed.

[0032] Based on the equivalent moment of inertia and angular acceleration of the rotating parts of the pump motor, an inertial torque compensation model is constructed.

[0033] Based on the ideal output torque, the mixed friction torque, and the inertial torque, a calculation equation for the theoretical output torque benchmark value is constructed to obtain the theoretical output torque benchmark value.

[0034] Preferably, the process of obtaining an individualized health benchmark model for the pump motor includes:

[0035] Under the healthy condition of the pump motor, multi-condition operating data covering different speeds, load pressures, and oil temperatures are collected to construct a healthy sample dataset.

[0036] The oil reference viscosity, compressive viscosity coefficient, equivalent effective clearance reference value and geometric flow channel are used to construct a parameter vector to be identified;

[0037] The simulation model containing the parameter vector to be identified is driven by the working condition data in the health sample dataset, and a multi-objective optimization function is constructed based on the degree of flow fitting error, torque fitting error and parameter deviation from prior knowledge.

[0038] The NSGA-III framework is used for multi-objective optimization. The Gaussian process regression model is used to predict the optimization objective value, and the expected hypervolume improvement is used as the acquisition function to select candidate solutions.

[0039] After obtaining the Pareto optimal solution set, the parameter confidence interval is calculated, the globally optimal parameter vector is output and fixed into the internal leakage mechanism model of the pump motor and the actual output torque calculation model of the pump motor, and an individualized health benchmark model of the pump motor is constructed.

[0040] Preferably, the process of multi-objective optimization using the NSGA-III framework includes:

[0041] The population is initialized by uniform sampling in the parameter space, and the multi-objective values ​​of individuals in the initial population are evaluated using a simulation model to construct the initial Pareto front.

[0042] Offspring are generated based on the selection, crossover, and mutation mechanisms of NSGA-III;

[0043] Using evaluated individuals to train a Gaussian process regression model, we predict the flow fitting accuracy target, torque fitting accuracy target, and model simplicity target, respectively.

[0044] The desired hypervolume improvement is used as the acquisition function, and individuals that improve the current Pareto front are selected from the candidate solutions for real simulation evaluation.

[0045] Repeat the above process until the algorithm converges and outputs a uniformly distributed Pareto optimal solution set.

[0046] Preferably, the process of assessing the health status of the pump motor based on the comprehensive health index and obtaining the assessment result of the pump motor health status includes:

[0047] Based on the real-time operating data of the pump motor and the output data of the pump motor's individualized health benchmark model, the residual of the motor output speed and the residual of the motor output torque are calculated.

[0048] The torque residual is dynamically standardized using the Z-Score method to obtain dimensionless speed anomaly multiples and torque anomaly multiples;

[0049] The abnormal multiples are processed by root mean square or moving average using a sliding time window to extract robust trend features and construct a multidimensional health feature vector.

[0050] The weighted sum is calculated based on the multidimensional health feature vector and the preset weight coefficient vector, and the weighted sum result is mapped to the [0,1] interval using a mapping function;

[0051] A shortcoming correction term is introduced, and the mapping result is multiplied by the shortcoming correction term to obtain a comprehensive health index. The health level is then classified according to the value of the comprehensive health index.

[0052] Preferably, the process of introducing a weakness correction term and multiplying the mapping result by the weakness correction term to obtain the comprehensive health index includes:

[0053] A multidimensional health feature vector is constructed based on the abnormal speed and abnormal torque ratios;

[0054] Weighting coefficients are assigned to the sensitivity of fault modes based on the characteristics of each dimension. Among them, the weighting coefficient corresponding to the abnormal torque multiple used to characterize the increase in leakage is greater than the weighting coefficient corresponding to the abnormal speed multiple.

[0055] Calculate the health of each subsystem and the average health of all subsystems, and construct a shortcoming correction term based on the ratio of the health of each subsystem to the average health, wherein the correction strength coefficient is greater than or equal to 1;

[0056] The comprehensive health index is obtained by multiplying the weighted sum of the multidimensional health feature vectors with the shortcoming correction term.

[0057] Preferably, the process of classifying health levels based on the value of the comprehensive health index includes:

[0058] When the comprehensive health index is greater than 0.8 and less than or equal to 1, it is determined to be a health level;

[0059] When the comprehensive health index is greater than 0.6 and less than or equal to 0.8, it is judged to be at a good level;

[0060] When the comprehensive health index is greater than 0.4 and less than or equal to 0.6, it is judged as a deterioration level;

[0061] When the comprehensive health index is greater than or equal to 0 and less than or equal to 0.4, it is determined to be a fault level.

[0062] Compared with the prior art, the present invention has the following advantages and technical effects:

[0063] This invention generates a dynamic health benchmark that changes synchronously with operating conditions by constructing a refined fluid dynamics model that integrates dynamic compressive viscosity correction and physical gap-driven leakage, as well as a mechanical dynamics model that includes hybrid friction and inertia compensation. This achieves adaptive stripping away of operating condition influences and significantly reduces the false alarm rate. Through model inversion technology, the health index is transformed into geometric and mechanical parameters with clear physical meaning, realizing transparency of degradation mechanisms and accurate fault location. The physical model built based on first principles requires only a small amount of health state data for parameter calibration. It has self-learning capabilities through online parameter identification, effectively reducing dependence on historical fault data samples. It has natural adaptability and reasoning ability for new operating condition combinations not appearing in the training set, significantly improving the model's engineering applicability and generalization ability. Attached Figure Description

[0064] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0065] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention;

[0066] Figure 2 This is a flowchart illustrating the construction of an ideal pump motor mathematical model according to an embodiment of the present invention.

[0067] Figure 3 This is a flowchart of the mechanism model for internal leakage of the pump motor according to an embodiment of the present invention;

[0068] Figure 4 This is a flowchart of the pump motor actual output torque calculation model according to an embodiment of the present invention;

[0069] Figure 5 This is a flowchart illustrating the inversion of key internal parameters of the pump motor based on experimental data, according to an embodiment of the present invention.

[0070] Figure 6 This is a flowchart illustrating the pump motor condition assessment based on health benchmarks and experimental data residuals, according to an embodiment of the present invention. Detailed Implementation

[0071] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0072] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0073] In this embodiment, to address the technical challenge that ideal mechanism models and traditional data-driven methods cannot distinguish between normal operating condition fluctuations and actual physical degradation under conditions of drastic fluctuations in pump motor speed, load, and temperature, thus making it difficult to obtain accurate theoretical health benchmark values ​​in real time and reflect the true health status of the equipment, a pump motor condition assessment method based on a high-fidelity mechanism model is provided. This method uses a physical model constructed based on first principles as its core framework. By modeling each key operating state during pump motor operation and combining parameter calibration and health feature extraction, it achieves a quantitative assessment of the pump motor condition under complex operating conditions.

[0074] like Figure 1 As shown, this embodiment provides a high-fidelity condition assessment method for high-end equipment pump motors, specifically including the following steps:

[0075] Based on the structural parameters and operating condition data of the pump motor, an ideal mathematical model of the pump motor is constructed to obtain the theoretical performance parameters of the pump motor.

[0076] Based on theoretical performance parameters, a mechanism model of internal leakage of pump motor is constructed to obtain dynamic volumetric efficiency.

[0077] Based on dynamic volumetric efficiency and friction loss mechanism, a calculation model for the actual output torque of pump motor is constructed to obtain the theoretical output torque benchmark value;

[0078] Based on the test data of the pump motor under healthy conditions, a multi-objective optimization algorithm is used to invert the undetermined parameters in the internal leakage mechanism model and the actual output torque calculation model of the pump motor to obtain an individualized health benchmark model of the pump motor.

[0079] The real-time collected pump motor performance signals are input into the pump motor individualized health benchmark model to obtain the model response output;

[0080] Based on the residual between the real-time collected pump motor operation data and the model response output, and after normalization processing, a dimensionless health feature vector is obtained.

[0081] The dimensionless health feature vectors are weighted and fused to obtain a comprehensive health index. The health status of the pump motor is then assessed based on the comprehensive health index to obtain the assessment result of the pump motor's health status.

[0082] Furthermore, the process of constructing a mathematical model of an ideal pump motor and obtaining its theoretical performance parameters includes:

[0083] Based on the maximum displacement of the variable pump, the swashplate angle, and the input speed, construct the equations for the pump's input and output flow rates;

[0084] Based on the motor displacement and input speed, construct the equations for the motor's input flow rate and output flow rate;

[0085] Based on the pump's volumetric efficiency and speed transmission ratio, an equation for calculating the motor speed is constructed.

[0086] Based on the pressure difference across the motor and the motor displacement, an equation for calculating the motor torque is constructed.

[0087] Based on the flow-pressure gradient of the check valve and the replenishment pressure, a replenishment flow equation is constructed.

[0088] Based on the principle of flow continuity on the high-pressure side and the low-pressure side, flow balance differential equations are constructed respectively;

[0089] The mathematical model of an ideal pump motor is constructed by using equations for the input and output flow rates of the combined pump, the input and output flow rates of the motor, the motor speed calculation equation, the motor torque calculation equation, the oil replenishment flow equation, and the flow balance differential equation, and the theoretical performance parameters of the pump motor are obtained.

[0090] The theoretical performance parameters of a pump motor include system pressure distribution, dynamic flow characteristics, output speed, and torque.

[0091] like Figure 2 As shown, further, the process of constructing the mathematical model of the ideal pump motor in this embodiment includes:

[0092] (1) During the operation of the pump motor, the hydraulic oil serves as the core working medium for power transmission, and its dynamic flow characteristics directly reflect the real-time transmission performance of the system. In the actual operation of the pump motor, both the pump's output flow and the motor's input flow are influenced by the combined effects of the input flow and leakage flow, exhibiting a dynamic coupling relationship. Specifically, the actual inlet and outlet flow of the variable pump can be constructed as a system function coupled with multiple physical parameters, and its mathematical expression is:

[0093] (1)

[0094] (2)

[0095] In the formula, This refers to the pump's input flow rate; This refers to the pump's output flow rate; This represents the maximum displacement of the variable pump. The swashplate inclination angle; This refers to the input speed of the variable pump. This refers to the leakage rate of the variable pump;

[0096] The inlet and outlet flow rates of the motor can also be constructed as a system function coupled with multiple physical parameters, and its mathematical expression is:

[0097] (3)

[0098] (4)

[0099] In the formula, This refers to the input flow rate of the motor. This refers to the motor's output flow rate; Motor displacement; This refers to the input speed of the motor; This refers to the leakage rate of the motor.

[0100] (2) Motor speed and torque;

[0101] Motor speed characterizes the actual rotational speed of the output shaft in a hydraulic pump motor system. Its value can be calculated and determined based on the input speed of the hydraulic pump and the overall volumetric efficiency of the system. Specifically, under ideal conditions, the motor speed should have a fixed proportional relationship with the pump's input speed. However, in actual systems, efficiency losses due to factors such as internal leakage necessitate the introduction of a volumetric efficiency parameter to correct the theoretical speed, thereby establishing an accurate speed transmission model. The specific expression is as follows:

[0102] (5)

[0103] In the formula, This refers to the volumetric efficiency of the pump.

[0104] Motor torque directly reflects the torque output capability of the hydraulic pump motor system in driving the output shaft. Its magnitude mainly depends on the pressure difference between the high-pressure and low-pressure sides of the system and the theoretical displacement of the motor. Under ideal conditions, the output torque has a linear relationship with the pressure difference and displacement. In actual systems, the combined effects of factors such as mechanical efficiency and friction loss must also be considered. The physical relationship can be expressed as follows:

[0105] (6)

[0106] In the formula, This refers to the motor torque; Motor displacement This refers to the oil pressure on the high-pressure side of the pump motor. This refers to the oil pressure on the low-pressure side of the pump motor.

[0107] (3) Hydraulic pump motor replenishment flow rate;

[0108] To ensure timely compensation for flow losses caused by component lubrication and oil leakage during the operation of the hydraulic pump motor closed circuit, and to maintain sufficient oil supply and pressure stability in the system, a dedicated replenishing pump is typically configured in the system. The compensation flow provided by this replenishing pump must be able to match the dynamic losses of the system in real time. Its theoretical replenishing flow rate can be characterized by a system leakage model and dynamic demand analysis, and the specific expression can be established as follows:

[0109] (7)

[0110] In the formula, To replenish fuel flow; The flow-pressure gradient of the check valve; To replenish oil pressure.

[0111] (4) Flow balance equation;

[0112] To accurately determine the dynamic pressure distribution within each chamber of the hydraulic pump motor system, a corresponding dynamic balance equation for the circuit pressure-flow rate needs to be established. In the high-pressure side circuit, the hydraulic fluid flows from the pump outlet through the high-pressure pipeline into the motor's working chamber, completing the energy transfer. Based on the principle of flow continuity, the dynamic flow balance relationship of the high-pressure side circuit can be expressed as:

[0113] (8)

[0114] In the formula, This refers to the total volume of a chamber within a high-pressure pipeline. This refers to the elastic modulus of the oil.

[0115] In the low-pressure side circuit, oil flows into the low-pressure chamber through the motor outlet and then returns through the pump inlet; simultaneously, compensating oil from the make-up pump is continuously injected into the low-pressure chamber to maintain flow balance in the circuit. The flow continuity equation for the low-pressure side circuit can be expressed as:

[0116] (9)

[0117] In the formula, This refers to the total volume of a chamber in a low-pressure pipeline.

[0118] By simultaneously solving the above equations, key performance parameters of the hydraulic pump motor under ideal operating conditions can be obtained, including system pressure distribution, dynamic flow characteristics, output speed, and torque. However, in actual operation, the amount of oil leakage inside the pump motor is difficult to observe directly. Furthermore, due to the influence of multiple complex factors such as oil temperature changes, external operating condition fluctuations, internal component wear, and dynamic evolution of fluid properties, the calculation results of the theoretical model based on ideal assumptions deviate significantly from the actual system state, making it difficult to accurately reflect the dynamic behavior and performance degradation process of the pump motor in real service environments. Therefore, it is necessary to further construct a high-fidelity mechanism model that integrates real-time state information to improve the accuracy of pump motor operating state characterization and degradation prediction capabilities.

[0119] Furthermore, the process of constructing a mechanism model for internal leakage in the pump motor and obtaining dynamic volumetric efficiency includes:

[0120] Based on oil temperature, oil pressure, viscosity-temperature characteristic parameters and pressure viscosity coefficient, a dynamic equivalent viscosity calculation model for oil is constructed to obtain real-time dynamic viscosity.

[0121] Based on the geometric conductance, shear conductance, effective clearance, inlet pressure and outlet pressure of the mating pair, a leakage calculation model for the mating pair is constructed based on the gap flow theory.

[0122] Based on the sum of the leakage amounts of the plunger and cylinder bore mating pairs, the slipper and swashplate mating pairs, and the distributor plate and cylinder block end face mating pairs, a model of the total leakage amount inside the pump motor is constructed, and the dynamic volumetric efficiency is calculated based on the total leakage amount and the theoretical flow rate.

[0123] Furthermore, the process of constructing a dynamic equivalent viscosity calculation model for oil includes:

[0124] Determine the viscosity-temperature characteristic parameters and viscosity-temperature coefficient based on the type of oil;

[0125] The pressure viscosity coefficient is determined based on the oil pressure to correct for the viscosity increase effect under high load;

[0126] The real-time dynamic viscosity is calculated based on the real-time collected oil temperature, high-pressure side oil pressure, viscosity-temperature characteristic parameters, pressure viscosity coefficient, and viscosity-temperature coefficient.

[0127] like Figure 3 As shown, further, the process of constructing the internal leakage mechanism model of the pump motor in this embodiment includes:

[0128] In solving the theoretical mathematical model of a hydraulic pump motor, volumetric efficiency is a key indicator affecting the performance calculation. However, during actual operation, the volumetric efficiency is primarily affected by oil leakage, essentially manifested as oil leakage from the clearances of key mating parts within the pump motor. This leakage is closely related to operating parameters such as oil temperature and pressure. To compensate for the impact on the accuracy of the pump motor model caused by neglecting the dynamic characteristics of the oil medium, this embodiment incorporates the temperature-viscosity characteristics of the oil and its rheological behavior under unsteady conditions into the modeling framework, thereby more accurately describing the dynamic response of the pump motor in the real working environment. To characterize the combined influence of oil temperature and high pressure on oil viscosity, a dynamic equivalent viscosity model is introduced. The model dynamically updates the dynamic viscosity of the oil using real-time pressure and temperature as variables. :

[0129] (10)

[0130] in, For oil temperature; This refers to the oil pressure of the pump motor. These are viscosity-temperature characteristic parameters based on oil type; The pressure viscosity coefficient is used to correct the leakage reduction effect caused by the increase in oil viscosity under high load; This is the viscosity-temperature coefficient.

[0131] Specifically, this embodiment also considers that oil leakage during actual pump motor operation only occurs from the high-pressure side to the low-pressure side. Therefore, the calculation of the above-mentioned oil dynamic viscosity is mainly based on the high-pressure side pressure parameters during pump motor operation. Meanwhile, oil leakage inside the pump motor is mainly achieved through the clearances of various mating pairs. Therefore, unlike traditional models that only use empirical leakage coefficients, this model establishes a direct mapping between the pump motor leakage and its internal physical clearances based on gap flow theory. The calculation method for leakage caused by various mating pairs inside the pump motor is characterized as follows:

[0132] (11)

[0133] in, For geometric flow guidance; For shear flow conduction; The effective clearance, as a core sensitive parameter characterizing the wear and health status of internal components of the pump motor, can have its initial value determined during the model calibration phase according to relevant standards for pump motor product design. ; To address the pressure on secondary exports; To address the pressure of imports.

[0134] Specifically, this embodiment also considers the change in pump motor volumetric efficiency caused by oil leakage from key internal components. The key internal components of the pump motor are identified as follows: the piston-cylinder bore mating pair, the slipper-swashplate mating pair, the distributor plate-cylinder end face mating pair, the cylinder block-spindle mating pair, the spindle-bearing pair mating pair, and the control piston-valve bore mating pair of the variable displacement mechanism. Therefore, the total internal leakage during pump motor operation can be mainly characterized as follows:

[0135] (12)

[0136] Therefore, the true volumetric efficiency during pump motor operation can be characterized as:

[0137] (13)

[0138] Furthermore, the process of constructing a calculation model for the actual output torque of the pump motor and obtaining the theoretical output torque reference value includes:

[0139] Based on the pressure difference across the motor, the motor displacement, and the dynamic volumetric efficiency, an ideal output torque equation is constructed.

[0140] Based on real-time dynamic viscosity, motor speed, Coulomb friction coefficient and static friction parameters, a hybrid friction torque model including viscous friction, Coulomb friction and static friction is constructed.

[0141] Based on the equivalent moment of inertia and angular acceleration of the rotating parts of the pump motor, an inertial torque compensation model is constructed.

[0142] Based on the ideal output torque, the mixed friction torque, and the inertial torque, a calculation equation for the theoretical output torque benchmark value is constructed to obtain the theoretical output torque benchmark value.

[0143] like Figure 4 As shown, further, this embodiment constructs a calculation model for the actual output torque of the pump motor to establish a calculation model for the output torque under real operating conditions of the pump motor. This model is used to accurately calculate the theoretical output torque of the pump motor under healthy conditions. Considering the leakage of the mating pairs of key internal components of the pump motor, the total torque generated by the pressure difference on both sides of the motor can be characterized as:

[0144] (14)

[0145] To accurately describe the torque loss caused by mechanical wear and lubrication during pump motor operation, a hybrid friction model incorporating viscous friction and Coulomb friction (characterizing solid contact) is established. Coupled calculations are performed using the obtained dynamic viscosity.

[0146]

[0147] in, The total frictional torque is the primary source of mechanical frictional heat. (First item) This is viscous friction, generated by oil film shear, and is proportional to rotational speed and viscosity. This item reflects the influence of the internal lubrication state of the pump motor on torque loss; the second item... Coulomb friction, also known as Coulomb friction, is generated by microscopic contact in components such as bearings and piston ball joints, and is proportional to the load. The Coulomb coefficient of friction is... It is a core, sensitive physical parameter that directly characterizes the mechanical wear state of components such as bearings and plunger pairs; the third item This is static friction, primarily acting as the sealing resistance during pump motor startup or low-speed operation.

[0148] Furthermore, given the drastic fluctuations in pump motor operating conditions (such as sudden speed changes), it is essential to eliminate the inertial torque during acceleration and deceleration to avoid misdiagnosing it as a resistance fault. The relevant formula expression is as follows:

[0149] (15)

[0150] in, This is the sum of the equivalent moments of inertia of the rotating parts of the pump motor. For drive shaft angular velocity .

[0151] In summary, the theoretical total output torque of a pump motor in a healthy state under complex operating conditions can be characterized as:

[0152] (16)

[0153] By simultaneously solving the above formulas, integrated simulation calculations of the pump motor's internal and external parameters can be performed. Internal parameters include input and output flow rates and oil leakage, while external parameters encompass input and output speeds, torque, and volumetric efficiency. The advantage of this method is its ability to highly replicate actual operating conditions, thereby achieving high-precision calculations of various key parameters.

[0154] Furthermore, the process of obtaining an individualized health benchmark model for the pump motor includes:

[0155] Under the healthy condition of the pump motor, multi-condition operating data covering different speeds, load pressures, and oil temperatures are collected to construct a healthy sample dataset.

[0156] The oil reference viscosity, compressive viscosity coefficient, equivalent effective clearance reference value and geometric flow channel are used to construct a parameter vector to be identified;

[0157] Based on the operating condition data in the health sample dataset, a simulation model containing the parameter vector to be identified is driven, and a multi-objective optimization function is constructed according to the degree of flow fitting error, torque fitting error and parameter deviation from prior knowledge.

[0158] The NSGA-III framework is used for multi-objective optimization. The Gaussian process regression model is used to predict the optimization objective value, and the expected hypervolume improvement is used as the acquisition function to select candidate solutions.

[0159] After obtaining the Pareto optimal solution set, the confidence interval of the parameters is calculated, the globally optimal parameter vector is output and solidified into the internal leakage mechanism model of the pump motor and the actual output torque calculation model of the pump motor, and an individualized health benchmark model of the pump motor is constructed.

[0160] Furthermore, the process of multi-objective optimization using the NSGA-III framework includes:

[0161] The population is initialized by uniform sampling in the parameter space, and the multi-objective values ​​of individuals in the initial population are evaluated using a simulation model to construct the initial Pareto front.

[0162] Offspring are generated based on the selection, crossover, and mutation mechanisms of NSGA-III;

[0163] Using evaluated individuals to train a Gaussian process regression model, we predict the flow fitting accuracy target, torque fitting accuracy target, and model simplicity target, respectively.

[0164] The desired hypervolume improvement is used as the acquisition function, and individuals that improve the current Pareto front are selected from the candidate solutions for real simulation evaluation.

[0165] Repeat the above process until the algorithm converges and outputs a uniformly distributed Pareto optimal solution set.

[0166] like Figure 5 As shown, this embodiment further demonstrates the inversion of key internal parameters of the pump motor based on experimental data. Under different operating conditions, the health baseline of the pump motor often varies significantly. Especially when operating conditions are complex and variable, its performance status and degradation trend are more likely to exhibit highly individualized characteristics, making it difficult to conduct a unified assessment using a single standard. To achieve accurate assessment of the health status of pump motors under different individual conditions and operating conditions, it is necessary to establish individual pump motor health baselines. Therefore, this embodiment conducts the inversion of key internal parameters of the pump motor based on experimental data, thereby constructing an individualized status assessment model. This has significant theoretical and engineering value for achieving precise equipment operation and maintenance and extending service life. The identification problem of key internal parameters of the pump motor is a multi-objective optimization problem. The core idea is: using a constructed pump motor simulation model containing the parameters to be identified, and using intelligent optimization algorithms to automatically adjust the parameters, minimizing the error between the model output (such as output speed and output torque) and the experimental measurement data. Specifically, this includes:

[0167] (1) Constructing a multi-condition health sample dataset for pump motors. With the pump motor confirmed to be healthy, its operation was controlled via an experimental bench, selecting a typical operating condition sequence covering low, medium, and high speeds, different load pressures, and different oil temperatures. Under each operating condition, a defined full-dimensional data vector was synchronously collected and recorded to form a dataset for inverting the internal parameters of the pump motor model:

[0168] (17)

[0169] This dataset It contains all the information needed for model input and output validation.

[0170] (2) Method for identifying internal parameters of pump motor. To prevent optimization divergence caused by multi-parameter coupling, the NSGA-III framework is used to construct the parameter identification problem as a multi-objective optimization problem that simultaneously minimizes flow error, torque error and model complexity (or parameter uncertainty), thereby realizing the identification of internal parameters of pump motor.

[0171] The parameter vector to be identified for the pump motor model is: .in This is the reference viscosity for the oil. The compressive viscosity coefficient, This is the equivalent effective gap reference value under healthy conditions. It is a geometric flow channel.

[0172] When identifying the internal parameters of the pump motor, the optimization objective of the algorithm is:

[0173] Flow fitting accuracy target: ;

[0174] Torque fitting accuracy target: ;

[0175] Model simplicity / robustness objectives: (The degree to which parameters deviate from prior knowledge, to prevent overfitting);

[0176] In the formula: It is the root mean square difference function. This represents the experimental value of the motor's output flow rate. This is the calculated value of the motor's output flow rate. This represents the experimental value of the motor's output torque. This is the calculated value of the motor's output torque. The prior knowledge values ​​of the parameters to be identified.

[0177] Optimization function: based on dataset Operating condition data The driver contains undetermined parameters as input. Fluid model.

[0178] Furthermore, the specific steps for identifying key internal parameters of the pump motor based on the NSGA-III framework are as follows:

[0179] Initialize the population, sample uniformly in the parameter space, and use a realistic simulation model to evaluate multiple target values ​​of individuals in the initial population. .

[0180] Offspring are generated based on the selection, crossover, and mutation mechanisms of NSGA-III.

[0181] Using the assessed individuals Train three independent Gaussian process regression models to predict... .

[0182] The expected hypervolume improvement (EHVI) is used as the acquisition function to select a small number of individuals with the greatest potential to improve the current Pareto front from the candidate solutions and evaluate them through real simulation.

[0183] After the algorithm converges, it outputs a set of uniformly distributed Pareto optimal solutions. For each solution on the frontier, the confidence interval of its parameter θ is calculated using Bootstrap resampling or variance estimation based on the surrogate model to ensure the accuracy and effectiveness of the estimation of the pump motor's internal parameters.

[0184] Furthermore, the process of assessing the health status of the pump motor based on a comprehensive health index and obtaining the assessment results of the pump motor's health status includes:

[0185] Based on the real-time operating data of the pump motor and the output data of the pump motor's individualized health benchmark model, the residual of the motor output speed and the residual of the motor output torque are calculated.

[0186] The Z-Score method is used to dynamically standardize the torque residual to obtain dimensionless speed anomaly multiples and torque anomaly multiples;

[0187] A sliding time window is used to process abnormal multiples by root mean square or moving average, extracting robust trend features and constructing a multidimensional health feature vector.

[0188] The weighted sum is calculated based on the multidimensional health feature vector and the preset weight coefficient vector, and the weighted sum result is mapped to the [0,1] interval using a mapping function;

[0189] A shortcoming correction term is introduced, and the mapping result is multiplied by the shortcoming correction term to obtain a comprehensive health index. The health level is then classified based on the value of the comprehensive health index.

[0190] Furthermore, the process of introducing a weakness correction term, multiplying the mapping result by the weakness correction term, and obtaining the comprehensive health index includes:

[0191] A multidimensional health feature vector is constructed based on the abnormal speed and abnormal torque ratios;

[0192] Weighting coefficients are assigned to the sensitivity of fault modes based on the characteristics of each dimension. Among them, the weighting coefficient corresponding to the abnormal torque multiple used to characterize the increase in leakage is greater than the weighting coefficient corresponding to the abnormal speed multiple.

[0193] Calculate the health of each subsystem and the average health of all subsystems. Construct a short-board correction term based on the ratio of the health of each subsystem to the average health, where the correction strength coefficient is greater than or equal to 1.

[0194] The comprehensive health index is obtained by multiplying the weighted sum of the multidimensional health feature vectors with the shortcoming correction term.

[0195] Furthermore, the process of classifying health levels based on the comprehensive health index includes:

[0196] When the comprehensive health index is greater than 0.8 and less than or equal to 1, it is judged as a healthy level;

[0197] When the overall health index is greater than 0.6 and less than or equal to 0.8, it is judged as good.

[0198] When the comprehensive health index is greater than 0.4 and less than or equal to 0.6, it is judged as a deterioration level;

[0199] When the comprehensive health index is greater than or equal to 0 and less than or equal to 0.4, it is judged as a fault level.

[0200] like Figure 6 As shown, further, the process of assessing the pump motor condition based on the residuals of health benchmarks and test data in this embodiment includes:

[0201] After identifying the key internal parameters of the pump motor based on the NSGA-III framework, the globally optimal parameter vector of the pump motor is obtained. And perform the following operations to establish a baseline for assessing the health status of the pump motor:

[0202] Parameter solidification: Based on the identified optimal parameter set Once the mechanistic model is constructed, this model becomes the health benchmark model. The performance signal acquired in real time by the pump motor system is used as the input vector. Input value health benchmark model The model's response vector to the pump motor system Solve the problem.

[0203] When evaluating the operating status of the pump motor, the instantaneous deviation between the measured data and the output value of the health benchmark model is calculated based on the measured data of the pump motor:

[0204] Motor output speed residual:

[0205] Motor output torque residual:

[0206] Residual Normalization and Feature Extraction: To address the issue of the different dimensions and difficulty in merging motor output speed (r / min) and output torque (Nm), this embodiment introduces the Z-Score dynamic normalization method. This transforms the two physical residuals into dimensionless outlier factors.

[0207] (18)

[0208] Using a sliding time window ( The instantaneous Z-value is processed by root mean square (RMS) or moving average to filter out high-frequency random noise from the sensor, extract robust trend features, and output a multidimensional health feature vector. :

[0209] (19)

[0210] The multidimensional health feature vectors obtained from the above steps are weighted and fused to output a comprehensive health index, which is used to determine the health level of the pump motor, thereby realizing a quantitative assessment of equipment status and fault early warning.

[0211] First, pre-define the multidimensional health feature vector. Weight coefficient vector and satisfy The weighting coefficients can be allocated according to the sensitivity of different failure modes. Preferably, to highlight the sensitivity to early wear (manifested as increased leakage), a weighting coefficient can be set. .

[0212] Secondly, calculate the comprehensive health index. To prevent calculation errors caused by the characteristic index Z potentially being negative when healthy, it can be mapped to the [0,1] interval. The mapping method used is as follows:

[0213] (20)

[0214] Among them, This is an adjustment coefficient that reflects the sensitivity to degradation of various characteristics.

[0215] To avoid the severe degradation of a single subsystem (e.g., a sharp increase in friction) being masked by the average degradation of other healthy subsystems, a short-board correction term λ(t) is introduced before weighted fusion:

[0216] (twenty one)

[0217] In the formula, This represents the health status of the currently weakest subsystem; Represents the average health of all subsystems; To correct the strength coefficient (usually set to 1). When In this case, the correction term is the ratio of the weakest link to the average. Increased size leads to stronger penalties for weaknesses.

[0218] Final pump motor overall health index The result is obtained by multiplying the weighted sum of health scores by the weakness correction term:

[0219] (twenty two)

[0220] The theoretical range of HI(t) is [0,1]. The closer the value is to 1, the healthier the device is, and the lower the value is, the more performance has degraded.

[0221] Health level classification and condition assessment. Based on the value of the pump motor comprehensive health index HI(t), the health status of the pump motor is classified into the following discrete levels, and clear maintenance operation guidelines are provided:

[0222] Health: 0.8 < HI ≤ 1; operating normally, in good condition.

[0223] Good: 0.6 < HI ≤ 8; slight degradation, it is recommended to pay attention and arrange maintenance and evaluation.

[0224] Deterioration: 0.4 < HI ≤ 0.6; Significant degradation, requiring immediate repair or operation restriction.

[0225] Fault: 0≤HI≤0.4; serious abnormality or fault, it is recommended to stop the machine for inspection.

[0226] This embodiment innovatively constructs a refined fluid dynamics model that includes dynamic pressure-viscosity correction and physical gap-driven leakage, as well as a mechanical dynamics model that includes hybrid friction and inertia compensation. This model system can accurately calculate the theoretical health status benchmark value that the equipment should possess under corresponding operating conditions based on real-time oil temperature, pressure, speed, and other operating conditions, providing a reliable mechanistic benchmark for high-precision condition assessment under complex operating conditions.

[0227] The evaluation method in this embodiment dynamically generates a health benchmark that changes synchronously with the operating conditions through a mechanistic model, abandoning the traditional fixed alarm threshold and achieving condition-adaptive evaluation thresholds. Simultaneously, a bottleneck correction logic is introduced into the fusion calculation of the comprehensive health index to ensure that the evaluation results can respond sensitively to local severe faults. This overcomes the defect of the traditional weighted average method, which may mask significant degradation in a single dimension. Thus, it achieves a robust and accurate evaluation effect that avoids false alarms under harsh operating conditions, provides early warnings for minor degradation, and guarantees an alarm in the event of a severe fault.

[0228] The method described in this embodiment provides a new generation of technical solutions for condition monitoring, health management, and predictive maintenance of key hydraulic components in high-end equipment, combining physical interpretability and engineering practicality. In terms of assessment accuracy, it achieves adaptive isolation of operating condition influences, significantly reducing false alarm rates. By integrating multi-field coupled models such as oil viscosity-temperature characteristics and wear-deformation effects, the generated theoretical health baseline can automatically adjust in real time to follow changes in operating conditions such as speed, pressure, and temperature. This fundamentally solves the problem of "normal drift false alarms" caused by operating condition fluctuations in the traditional fixed threshold method, shifting the focus of condition assessment from absolute signal values ​​to the residual values ​​with dynamic benchmarks, greatly improving the accuracy of assessment results under complex operating conditions.

[0229] In terms of diagnostic interpretability, it achieves transparency of degradation mechanisms and precise fault location. The abstract "health index" is transformed into key geometric and mechanical parameters with clear physical meaning. Through model inversion technology, the physical relationships between the parameters of key internal components of the pump motor and the model are directly quantified, enabling precise calculation of the contribution of each worn component to overall performance (such as decreased volumetric efficiency and increased vibration).

[0230] Meanwhile, in terms of engineering applicability, this method reduces reliance on historical fault data samples, enhancing its practicality. Based on a physical model constructed using first-principles calculations, this method requires only a small amount of health status data for parameter calibration during initialization, without relying on a massive, complete fault sample library. The model possesses self-learning capabilities through online parameter identification, enabling adaptive adjustments as equipment ages, thus solving the common problem of "scarcity of full lifecycle data" in industrial settings. Furthermore, since the model is essentially a digital expression of physical laws, it possesses natural adaptability and reasoning ability for new operating condition combinations (new speed-pressure-temperature combinations) not present in the training set, overcoming the deficiency of pure data-driven models where generalization performance drops sharply during operating condition extrapolation.

[0231] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A high-fidelity condition assessment method for pump motors in high-end equipment, characterized in that, include: Based on the structural parameters and operating condition data of the pump motor, an ideal mathematical model of the pump motor is constructed to obtain the theoretical performance parameters of the pump motor. Based on the theoretical performance parameters, a mechanism model for internal leakage of the pump motor is constructed to obtain the dynamic volumetric efficiency. Based on the dynamic volumetric efficiency and friction loss mechanism, a calculation model for the actual output torque of the pump motor is constructed to obtain the theoretical output torque benchmark value. Based on the test data of the pump motor under healthy conditions, a multi-objective optimization algorithm is used to invert the undetermined parameters in the internal leakage mechanism model and the actual output torque calculation model of the pump motor to obtain an individualized health benchmark model of the pump motor. The real-time collected pump motor performance signals are input into the pump motor individualized health benchmark model to obtain the model response output; The residual between the real-time collected pump motor operation data and the model response output is normalized to obtain a dimensionless health feature vector. The dimensionless health feature vector is weighted and fused to obtain a comprehensive health index. The health status of the pump motor is evaluated based on the comprehensive health index to obtain the evaluation result of the pump motor health status. The process of constructing a mechanism model for internal leakage in a pump motor and obtaining dynamic volumetric efficiency includes: Based on oil temperature, oil pressure, viscosity-temperature characteristic parameters and pressure viscosity coefficient, a dynamic equivalent viscosity calculation model for oil is constructed to obtain real-time dynamic viscosity. Based on the geometric conductance, shear conductance, effective clearance, inlet pressure and outlet pressure of the mating pair, a leakage calculation model for the mating pair is constructed based on the gap flow theory. Based on the sum of the leakage amounts of the plunger and cylinder bore mating pairs, the slipper and swashplate mating pairs, and the distributor plate and cylinder block end face mating pairs, a model of the total leakage amount inside the pump motor is constructed, and the dynamic volumetric efficiency is calculated based on the total leakage amount and the theoretical flow rate. The process of obtaining an individualized health benchmark model for the pump motor includes: Under the healthy condition of the pump motor, multi-condition operating data covering different speeds, load pressures, and oil temperatures are collected to construct a healthy sample dataset. The oil reference viscosity, compressive viscosity coefficient, equivalent effective clearance reference value and geometric flow channel are used to construct a parameter vector to be identified; The simulation model containing the parameter vector to be identified is driven by the working condition data in the health sample dataset, and a multi-objective optimization function is constructed based on the degree of flow fitting error, torque fitting error and parameter deviation from prior knowledge. The NSGA-III framework is used for multi-objective optimization. The Gaussian process regression model is used to predict the optimization objective value, and the expected hypervolume improvement is used as the acquisition function to select candidate solutions. After obtaining the Pareto optimal solution set, the parameter confidence interval is calculated, the globally optimal parameter vector is output and fixed into the internal leakage mechanism model of the pump motor and the actual output torque calculation model of the pump motor, and an individualized health benchmark model of the pump motor is constructed.

2. The method according to claim 1, characterized in that, The process of constructing a mathematical model of an ideal pump motor and obtaining its theoretical performance parameters includes: Based on the maximum displacement of the variable pump, the swashplate angle, and the input speed, construct the equations for the pump's input and output flow rates; Based on the motor displacement and input speed, construct the equations for the motor's input flow rate and output flow rate; Based on the pump's volumetric efficiency and speed transmission ratio, an equation for calculating the motor speed is constructed. Based on the pressure difference across the motor and the motor displacement, an equation for calculating the motor torque is constructed. Based on the flow-pressure gradient of the check valve and the replenishment pressure, a replenishment flow equation is constructed. Based on the principle of flow continuity on the high-pressure side and the low-pressure side, flow balance differential equations are constructed respectively; By combining the input and output flow equations of the pump, the input and output flow equations of the motor, the motor speed calculation equation, the motor torque calculation equation, the oil replenishment flow equation, and the flow balance differential equation, an ideal pump motor mathematical model is constructed, and the theoretical performance parameters of the pump motor are obtained. The theoretical performance parameters of the pump motor include system pressure distribution, dynamic flow characteristics, output speed and torque.

3. The method according to claim 1, characterized in that, The process of constructing a dynamic equivalent viscosity calculation model for oil includes: Determine the viscosity-temperature characteristic parameters and viscosity-temperature coefficient based on the type of oil; The pressure viscosity coefficient is determined based on the oil pressure to correct for the viscosity increase effect under high load; The real-time dynamic viscosity is calculated based on the real-time collected oil temperature, high-pressure side oil pressure, viscosity-temperature characteristic parameters, pressure viscosity coefficient, and viscosity-temperature coefficient.

4. The method according to claim 1, characterized in that, The process of constructing a calculation model for the actual output torque of a pump motor and obtaining a theoretical output torque benchmark value includes: Based on the pressure difference across the motor, the motor displacement, and the dynamic volumetric efficiency, an ideal output torque equation is constructed. Based on real-time dynamic viscosity, motor speed, Coulomb friction coefficient and static friction parameters, a hybrid friction torque model including viscous friction, Coulomb friction and static friction is constructed. Based on the equivalent moment of inertia and angular acceleration of the rotating parts of the pump motor, an inertial torque compensation model is constructed. Based on the ideal output torque, the mixed friction torque, and the inertial torque, a calculation equation for the theoretical output torque benchmark value is constructed to obtain the theoretical output torque benchmark value.

5. The method according to claim 1, characterized in that, The process of multi-objective optimization using the NSGA-III framework includes: The population is initialized by uniform sampling in the parameter space, and the multi-objective values ​​of individuals in the initial population are evaluated using a simulation model to construct the initial Pareto front. Offspring are generated based on the selection, crossover, and mutation mechanisms of NSGA-III; Using evaluated individuals to train a Gaussian process regression model, we predict the flow fitting accuracy target, torque fitting accuracy target, and model simplicity target, respectively. The desired hypervolume improvement is used as the acquisition function, and individuals that improve the current Pareto front are selected from the candidate solutions for real simulation evaluation. Repeat the above process until the algorithm converges and outputs a uniformly distributed Pareto optimal solution set.

6. The method according to claim 1, characterized in that, The process of assessing the health status of the pump motor based on the comprehensive health index and obtaining the assessment results of the pump motor health status includes: Based on the real-time operating data of the pump motor and the output data of the pump motor's individualized health benchmark model, the residual of the motor output speed and the residual of the motor output torque are calculated. The torque residual is dynamically standardized using the Z-Score method to obtain dimensionless speed anomaly multiples and torque anomaly multiples; The abnormal multiples are processed by root mean square or moving average using a sliding time window to extract robust trend features and construct a multidimensional health feature vector. The weighted sum is calculated based on the multidimensional health feature vector and the preset weight coefficient vector, and the weighted sum result is mapped to the [0,1] interval using a mapping function; A shortcoming correction term is introduced, and the mapping result is multiplied by the shortcoming correction term to obtain a comprehensive health index. The health level is then classified according to the value of the comprehensive health index.

7. The method according to claim 6, characterized in that, The process of introducing a weakness correction term, multiplying the mapping result by the weakness correction term, and obtaining the comprehensive health index includes: A multidimensional health feature vector is constructed based on the abnormal speed and abnormal torque ratios; Weighting coefficients are assigned to the sensitivity of fault modes based on the characteristics of each dimension. Among them, the weighting coefficient corresponding to the abnormal torque multiple used to characterize the increase in leakage is greater than the weighting coefficient corresponding to the abnormal speed multiple. Calculate the health of each subsystem and the average health of all subsystems, and construct a shortcoming correction term based on the ratio of the health of each subsystem to the average health, wherein the correction strength coefficient is greater than or equal to 1; The comprehensive health index is obtained by multiplying the weighted sum of the multidimensional health feature vectors with the shortcoming correction term.

8. The method according to claim 6, characterized in that, The process of classifying health levels based on the values ​​of the comprehensive health index includes: When the comprehensive health index is greater than 0.8 and less than or equal to 1, it is determined to be a health level; When the comprehensive health index is greater than 0.6 and less than or equal to 0.8, it is judged to be at a good level; When the comprehensive health index is greater than 0.4 and less than or equal to 0.6, it is judged as a deterioration level; When the comprehensive health index is greater than or equal to 0 and less than or equal to 0.4, it is determined to be a fault level.

Images