Hydraulic valve flow soft measurement method, system and equipment for online prediction of abrasion loss and medium

Abstract

The invention discloses a hydraulic valve flow soft measurement method for on-line prediction of abrasion loss, which comprises the following steps of: acquiring operation data of a hydraulic valve under different working conditions as input characteristics, and training a neural network soft measurement model; predicting the valve port flow according to the trained flow soft measurement model and the hydraulic valve operation data collected in real time; calculating the effective through-flow area of a valve port, the average speed of a nozzle and the mass flux of particles in unit area; the particle impact probability is calculated according to the particle impact characteristics, the total abrasion loss is distributed to the valve seat edge and the valve element edge according to the particle impact probability, and the erosive abrasion rate is calculated; and outputting the valve port abrasion loss according to the erosive wear rate, updating the valve port overflowing area according to the valve port abrasion loss, and calibrating the flow soft measurement model on line. The valve port abrasion can be estimated in real time, the full-period flow change can be tracked, performance degradation caused by abrasion can be recognized in time, predictive maintenance and fault diagnosis are supported, the failure risk caused by excessive abrasion is reduced, the flow control precision is improved, and the service life of equipment is prolonged.

Description

Technical Field

[0001] This invention relates to the field of hydraulic valve flow rate soft measurement and wear technology, specifically to a hydraulic valve flow rate soft measurement method, system, equipment, and medium for online prediction of wear. Background Technology

[0002] As the core control component of a hydraulic system, the geometric fit accuracy and valve port condition of a hydraulic valve directly affect the system's efficiency, stability, and lifespan. Minor geometric changes at the valve port, such as increased fillet radius, rolled cutting edges, or widened throttling clearance, can cause drift in flow rate, differential pressure, and opening characteristics, further leading to actuator hysteresis, oscillation, and decreased control accuracy. Therefore, online sensing and quantitative assessment of valve port wear conditions are crucial for achieving highly reliable hydraulic systems and intelligent operation and maintenance.

[0003] In existing engineering projects, valve wear is often assessed through periodic shutdowns for disassembly and inspection, microscopic measurements, or offline three-dimensional re-measurement to obtain geometric dimensional changes. A few solutions use empirical coefficients for correction or indirectly reflect wear through fixed threshold alarms, but these are difficult to correlate with specific geometric quantification indicators. Another approach is based on computational fluid dynamics and experimental erosion life estimation, which typically requires complete boundary conditions and long-term testing, making it difficult to update on-site according to operating conditions and time. These methods are either visible but not online, or online but not quantitative, failing to meet the need for continuous tracking of wear amount, wear process, and its impact on flow characteristics.

[0004] Wear-related online information is available within the system. Process parameters such as valve cross-sectional pressure difference, valve opening, oil temperature, and system flow rate are widely measurable. Existing soft-sensor research often establishes flow models based on static geometry as a default assumption, failing to incorporate the fact that geometry evolves with service degradation. As service time increases, valve fillet radii and throttling areas change, causing soft-sensor flow rates based on fixed geometric assumptions to gradually become inaccurate. This makes it impossible to deduce wear levels or provide quantitative data for maintenance, such as when and to what extent maintenance should be performed. Summary of the Invention

[0005] To address the shortcomings of existing technologies, embodiments of the present invention provide a hydraulic valve flow rate soft measurement method, system, device, and medium for online prediction of wear. This method can estimate valve port wear in real time, track flow rate changes throughout the entire cycle, and promptly identify performance degradation caused by wear. It supports predictive maintenance and fault diagnosis, reduces the risk of failure due to excessive wear, improves flow control accuracy, and extends equipment life.

[0006] In a first aspect, the first embodiment of the present invention provides a soft measurement method for hydraulic valve flow rate to predict wear online, comprising:

[0007] The operating data of hydraulic valves under different working conditions are collected as input features to train a neural network soft measurement model for flow prediction, thus obtaining a trained flow soft measurement model.

[0008] Predict valve orifice flow rate based on a trained flow soft measurement model and real-time collected hydraulic valve operating data;

[0009] The effective flow area, average nozzle velocity, and particle mass flux per unit area of ​​the valve orifice are calculated based on the predicted orifice flow rate and valve orifice geometric parameters.

[0010] The particle impact probability is calculated based on the particle impact characteristics. The total area wear rate is calculated based on the average nozzle velocity and the particle mass flux per unit area. The total area wear rate is allocated to the valve seat edge and the valve core edge based on the particle impact probability. The erosion wear rate is calculated in conjunction with the erosion model.

[0011] The valve orifice wear amount is output based on the erosion wear rate, the valve orifice flow area is updated based on the valve orifice wear amount, and the flow soft measurement model is calibrated online based on the updated valve orifice flow area.

[0012] Furthermore, the flow soft measurement model adopts a feedforward regression network structure. The input layer incorporates three features: oil temperature, valve opening, and pressure difference across the valve. Several fully connected layers are set in the middle and ReLU activation function is used. The output layer is a single linear node to output continuous flow.

[0013] Furthermore, the flow soft measurement model predicts flow based on the following formula:

[0014]

[0015] in, It is the flow coefficient. It is the effective flow area of ​​the valve port. It is the density of the oil. It is the pressure difference across the valve.

[0016] Furthermore, specific methods for calculating the effective flow area, average nozzle velocity, and particle mass flux per unit area of ​​the valve orifice based on the predicted orifice flow rate and valve orifice geometric parameters include:

[0017] Calculate the effective flow area of ​​the valve port based on the nominal radius of the valve seat, the assembly offset, the valve port opening, and the fillet radii of the valve seat edge and the valve core edge.

[0018] Calculate the average nozzle velocity based on the predicted flow rate and effective flow area;

[0019] Calculate the particle mass flux per unit area based on the solid mass fraction, mixture density, predicted flow rate, and effective flow area.

[0020] Furthermore, the particle impact probability is calculated based on the particle impact characteristics, and the total area wear rate is calculated based on the average nozzle velocity and the particle mass flux per unit area. The total area wear rate is then allocated to the valve seat edge and the valve core edge based on the particle impact probability, specifically including:

[0021] Based on the circumferential impactable projection length of the valve seat edge and the valve core edge, the range of the neutral zone, the characteristic particle size, and the valve opening, the probability of a particle impacting the valve seat edge and the probability of impacting the valve core edge are calculated respectively.

[0022] Based on the nozzle velocity, particle mass flux per unit area, and calibrated erosion model parameters, the total wear rate per unit area is calculated, and the total wear rate is distributed to the valve seat edge and valve core edge according to the probability of impacting the valve seat edge and the probability of impacting the valve core edge.

[0023] Furthermore, the valve orifice wear amount is output based on the erosion wear rate, the valve orifice flow area is updated based on the valve orifice wear amount, and the flow rate soft measurement model is calibrated online based on the updated valve orifice flow area, specifically including:

[0024] The total wear rate per unit area is integrated over time within one operating cycle to obtain the cumulative wear volume increment at the valve orifice as the wear amount.

[0025] Substitute the wear volume increment into the erosion-filler geometric evolution relationship model to solve for the updated valve seat fillet radius and valve core fillet radius;

[0026] Calculate the updated effective flow area of ​​the valve port based on the updated fillet radius of the valve seat and the fillet radius of the valve core;

[0027] The updated effective flow area is introduced as a geometric constraint into the flow soft measurement model, and the flow soft measurement model is calibrated online to correct the flow prediction output.

[0028] Furthermore, the online calibration employs a weighted ridge regression method to linearly correct the output of the flow soft measurement model.

[0029] Secondly, another embodiment of the present invention provides a hydraulic valve flow rate soft measurement system for online wear prediction, used to implement the hydraulic valve flow rate soft measurement method for online wear prediction described in the first embodiment of the present invention, including:

[0030] The model training module is used to collect the operating data of the hydraulic valve under different working conditions as input features, train the neural network soft measurement model for flow prediction, and obtain the trained flow soft measurement model.

[0031] The flow prediction module is used to predict the valve orifice flow based on the trained flow soft measurement model and real-time collected hydraulic valve operating data.

[0032] The first calculation module is used to calculate the effective flow area of ​​the valve orifice, the average velocity of the nozzle, and the particle mass flux per unit area based on the predicted valve orifice flow rate and valve orifice geometric parameters.

[0033] The second calculation module is used to calculate the particle impact probability based on the particle impact characteristics, calculate the total area wear rate based on the average nozzle velocity and the particle mass flux per unit area, allocate the total area wear rate to the valve seat edge and the valve core edge based on the particle impact probability, and calculate the erosion wear rate in conjunction with the erosion model.

[0034] The online calibration module is used to output the valve port wear amount based on the erosion wear rate, update the valve port flow area based on the valve port wear amount, and perform online calibration of the flow soft measurement model based on the updated valve port flow area.

[0035] Thirdly, another embodiment of the present invention provides an electronic device including a processor, an input device, an output device, and a memory, wherein the processor, input device, output device, and memory are interconnected, the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the hydraulic valve flow soft measurement method for online prediction of wear amount described in the first embodiment of the present invention.

[0036] Fourthly, an embodiment of the present invention provides a computer-readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, cause the processor to perform the soft measurement method for online prediction of wear amount of hydraulic valve flow described in the first embodiment of the present invention.

[0037] The beneficial effects of this invention are:

[0038] The present invention provides a hydraulic valve flow rate soft measurement method, system, equipment, and medium for online prediction of wear, which has the following advantages:

[0039] 1. Based on the "erosion-geometry-soft measurement" closed loop, accurate predictive maintenance is achieved. The wear rate and wear volume per unit area are calculated in real time, the valve port flow area is updated online and the flow rate soft measurement model is calibrated. It can issue early warnings before the performance deteriorates significantly, scientifically plan the maintenance time, significantly reduce the risk of unplanned downtime caused by sudden valve port failure, and ensure the continuity and stability of production. 2. By adopting neural network flow soft measurement technology, the valve orifice flow can be calculated with high accuracy using only conventional and easily measurable data such as oil temperature, valve orifice opening, and pressure difference. This avoids dependence on expensive high-bandwidth flow sensors and reduces system hardware complexity and deployment costs. When equipped with sensors, it can also be fused and cross-calibrated with measured data to further improve measurement reliability and robustness. 3. By integrating data-driven approaches with physical mechanisms, the nonlinear fitting capability of neural networks is combined with the valve erosion model. The online update of the flow area is used as a geometric constraint to ensure the model's adaptability to complex working conditions such as pressure fluctuations, temperature and viscosity changes, and surface roughness evolution. This also ensures that the prediction results have clear physical meaning and traceability, making wear and flow prediction more accurate and reliable. 4. Supports dynamic optimization of the control system, updating key geometric parameters and soft measurement models of the valve port in real time according to the wear status, enabling the controller to implement adaptive feedforward / feedback compensation, reducing overshoot and oscillation, maintaining stable flow characteristics and control performance throughout the valve's entire life cycle, significantly extending equipment service life and improving overall operating efficiency. Attached Figure Description

[0040] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0041] Figure 1 A flowchart of a soft measurement method for online prediction of wear amount of hydraulic valve flow provided in the first embodiment of the present invention is shown;

[0042] Figure 2 A schematic diagram of a hydraulic valve flow rate soft measurement system for online wear prediction provided by another embodiment of the present invention is shown. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0045] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0046] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes such combinations.

[0047] like Figure 1 The diagram shows a flowchart of a hydraulic valve flow rate soft measurement method for online prediction of wear amount provided by the first embodiment of the present invention. The method includes the following steps:

[0048] Step S1: Collect the operating data of the hydraulic valve under different working conditions as input features for the neural network soft measurement model for flow prediction, and obtain the trained flow soft measurement model.

[0049] Step S2: Predict the valve orifice flow rate based on the trained flow soft measurement model and the real-time collected hydraulic valve operating data;

[0050] Step S3: Calculate the effective flow area of ​​the valve orifice, the average nozzle velocity, and the particle mass flux per unit area based on the predicted valve orifice flow rate and valve orifice geometric parameters.

[0051] Step S4: Calculate the particle impact probability based on the particle impact characteristics, calculate the total area wear rate based on the average nozzle velocity and the particle mass flux per unit area, allocate the total area wear rate to the valve seat edge and valve core edge based on the particle impact probability, and calculate the erosion wear rate in conjunction with the erosion model.

[0052] Step S5: Output the valve port wear amount based on the erosion wear rate, update the valve port flow area based on the valve port wear amount, and perform online calibration of the flow soft measurement model based on the updated valve port flow area.

[0053] As can be seen from the above embodiments, the present invention combines the flow prediction capability of the neural network soft sensor model with the wear calculation logic of the erosion model in the valve port wear prediction method. Based on different operating condition data obtained during the normal service phase of the hydraulic valve, including temperature, pressure, valve port opening, etc., the neural network soft sensor model is trained using these as input features to obtain a trained flow soft sensor model. The real-time flow rate at the valve port is predicted according to the trained flow soft sensor model. Combining the hydraulic system medium characteristics and valve port geometric parameters, the fluid velocity and particle mass flux at the nozzle are calculated. The particle impact probability is calculated based on the particle impact characteristics. The total area wear rate is calculated based on the average nozzle velocity and the particle mass flux per unit area. The total area wear rate is then allocated to the valve seat edge and valve core edge according to the particle impact probability. Finally, the total area wear rate is allocated to the corresponding regions and the erosion wear rate is calculated.

[0054] Based on the calculated valve port wear amount, and combined with the correspondence between the wear amount and the valve port fillet radius, the current valve port fillet radius parameter is updated. This updated fillet radius serves as both the valve port geometric feedback data for the next prediction cycle and a basis for hydraulic valve maintenance decisions, such as maintenance timing and methods. In this way, even during long-term service and gradual wear of the valve port, the valve port wear state can be dynamically and accurately predicted, achieving continuous monitoring without disassembling or replacing the valve body.

[0055] In step S1 above, key input features are collected in real time from the hydraulic system, including oil temperature T(t) and valve opening. (t), pressure difference across the valve and valve flow This real-time data provides the necessary input information for flow forecasting. The input data is used to train a neural network model designed to predict valve flow, based on the following flow forecasting formula:

[0056] ;

[0057] in, It is the flow coefficient. It is the effective flow area of ​​the valve port. It is the density of the oil. It is the pressure difference across the valve.

[0058] The flow rate soft measurement model employs a feedforward regression network. The input layer handles three main features: temperature, valve opening, and pressure difference across the valve. Several fully connected layers are configured in the middle and activated using ReLU. The output layer is a single linear node that directly outputs the continuous flow rate. During the training phase, the physical priors of orifice flow are embedded as soft constraints or correction terms. The deviation between the physical flow rate calculated from the priors and the network output is weighted and minimized in the loss function, ensuring that the model follows the mechanistic trend while absorbing assembly errors, leakage, and other non-ideal effects. The size of the hidden layers is designed to achieve stable convergence without overfitting, and the capacity and complexity are controlled through weight decay, early stopping, and piecewise decay of the learning rate.

[0059] The training and validation process employs an adaptive first-order optimization algorithm, setting appropriate initial learning rate, batch size, and maximum training epochs, and decaying the learning rate at fixed epochs. Early stopping is enabled using validation set loss as the monitoring metric; if there is no improvement after several consecutive epochs, training is stopped and rolled back to the optimal weights on the validation set. Before training, inputs and outputs undergo uniform normalization or standardization, and this transformation is persisted to ensure consistency between offline training and subsequent application stages.

[0060] Model performance is evaluated on an independent test set. The report comprehensively measures indicators such as mean absolute error, root mean square error, mean absolute percentage error, and correlation coefficient. It also provides error distribution and single-factor scan results for typical operating conditions to verify the stability and physical rationality of the model in different temperature, opening, and pressure difference ranges, ensuring that the output meets the basic monotonicity requirements.

[0061] Step S1: Collect operating data of the hydraulic valve under different working conditions as input features to train a neural network soft sensing model, specifically including:

[0062] Step S101: In terms of neural network structure design, a deep feedforward regression network is constructed. The input layer is explicitly configured with 3 nodes, corresponding to three key physical characteristics of the hydraulic system: oil temperature measurement range of 20-80℃, valve opening range of 0-100%, and pressure difference across the valve range of 0-31.5MPa. The hidden layers of the network adopt a three-layer fully connected architecture, with 8 neurons in each layer. This configuration has been verified through extensive experiments and can effectively control the risk of overfitting while ensuring the model's fitting ability. All hidden layers use the ReLU activation function, which shows better training stability and prediction accuracy than sigmoid and tanh in flow prediction tasks.

[0063] Regarding the physical prior embedding mechanism, a dedicated physical constraint layer was designed based on the fundamental formula for orifice flow rate. Through a custom regression layer, a physical consistency constraint term was added to the standard mean square error loss function. This constraint term calculates the deviation between the neural network-predicted flow rate and the theoretical physical flow rate, and is weighted into the total loss function with a weighting coefficient of 0.1. This design enables the model to learn complex nonlinear relationships from data while adhering to basic fluid dynamics laws, effectively absorbing non-ideal effects such as assembly errors and leakage in actual systems.

[0064] In terms of regularization strategies, multiple protective measures were adopted. A weight decay mechanism was used to control model complexity, with a gradient decay factor of 0.9 and a second-order moment momentum decay factor of 0.998. An early stopping monitoring strategy was implemented, automatically terminating training and rolling back to the optimal weights on the validation set when the validation set loss showed no improvement for 20 consecutive rounds. A segmented learning rate decay scheme was used, reducing the learning rate to half its original value every 100 training rounds to balance convergence speed and final accuracy.

[0065] Step S102: In the data preprocessing stage, a uniform normalization process is performed on all input and output data. The mapminmax function (normalization function) is used to linearly transform each feature to the [0,1] interval. This method can eliminate numerical differences caused by different physical units and accelerate network convergence. Crucially, the normalization parameters (inputps, outputps) determined during the training stage are persistently saved to ensure complete consistency between offline training and online application data processing, avoiding prediction bias caused by inconsistent preprocessing.

[0066] In terms of training set partitioning strategy, a stratified sampling method was adopted. The original normal state dataset was divided into 85% training, 15% validation, and 15% test sets to ensure the consistency of data distribution in each subset. For wear state data, due to the limited number of samples, 112 sets (14%) were randomly selected from 800 sets of data as training samples for transfer learning. This small sample setting is more in line with the current situation of scarce fault data in engineering practice.

[0067] For training optimization, the Adam adaptive first-order optimization algorithm was selected, with an initial learning rate of 0.01 and a batch size of 128. The learning rate scheduling employed a piecewise constant decay strategy, coupled with a fixed decay period of once every 100 epochs. The validation monitoring mechanism was set to evaluate model performance on an independent validation set every 20 epochs. Early stopping was triggered when the validation loss showed no improvement for 20 consecutive epochs, and the optimal weights on the validation set were automatically saved. Visual monitoring throughout the entire training process ensured real-time tracking of the model's convergence state and generalization ability changes.

[0068] Step S103: Regarding the quantitative evaluation of model performance, a comprehensive error index system was calculated on the independent test set, including absolute error indices: sum of squared errors (SSE), mean absolute error (MAE), and root mean square error (RMSE); relative error indices: mean absolute percentage error (MAPE); and correlation indices: Pearson correlation coefficient (R). These indices comprehensively reflect the model's prediction accuracy from different perspectives. Among them, the correlation coefficient focuses on measuring the consistency between the model's predicted trend and actual flow changes, while the mean absolute percentage error more intuitively reflects the level of relative prediction error.

[0069] For physical plausibility verification, a systematic single-factor scan analysis was implemented. With the other two input parameters fixed, one of the factors—temperature, valve opening, or pressure difference—was systematically changed, and the changes in the model output were observed. The focus was on verifying three key physical characteristics: the monotonically increasing flow rate with valve opening, the monotonically increasing flow rate with the pressure difference across the valve, and the physical plausibility of the effect of temperature on flow rate. This analysis ensures that the model not only has good data fit, but more importantly, that its input-output relationship conforms to basic physical laws, avoiding predictions that violate common sense in engineering.

[0070] In terms of operational adaptability assessment, the error distribution characteristics of the model in different operating ranges were analyzed. Particular attention was paid to the predictive stability under extreme conditions such as low temperature and high pressure, and high temperature and low pressure, as well as the error control capability in nonlinear regions such as small opening and low pressure differential. Through error distribution statistics and typical operating condition scanning, a complete model evaluation report was generated, providing a basis for the model's online application and ensuring the reliability and robustness of the soft measurement model in actual hydraulic systems.

[0071] In step S2 above, during the real-time soft measurement execution phase, three key parameters—oil temperature, valve opening, and pressure difference across the valve—are first collected in real time from the hydraulic sensor and assembled into an input vector according to the feature order determined during the training phase. Subsequently, the input vector is standardized on the same scale using the normalization parameters fixed with the model, transforming it to the [0,1] interval to ensure consistency with the distribution of the training data.

[0072] The standardized input vector is fed into a pre-defined neural network for forward computation to obtain the predicted flow rate in the normalized domain. Finally, the output is denormalized to restore the flow rate estimate with physical dimensions, which is then output as the soft measurement result for the current moment. This process is implemented using the apply and reverse operations of mapminmax, ensuring complete consistency between offline training and online application.

[0073] In step S3 above, based on the real-time flow data obtained from soft sensing, a rigorous three-level calculation process is initiated to accurately quantify the two core physical driving factors leading to erosion wear. Specifically, step S3 may include:

[0074] Step S301: Calculate the effective flow area S(t) of the valve orifice using the following formula:

[0075] ;

[0076] in, The nominal radius of the valve seat. For assembly offset or equivalent clearance. For valve opening, Valve seat edge fillet radius, Valve core edge fillet radius. The first item is the annular gap area, and the latter two items are the fillet radius corrections on both sides. A minimum area threshold should be set for the smallest opening to avoid subsequent division by zero.

[0077] Step S302: Calculate the average nozzle velocity using the following formula. :

[0078] ;

[0079] in, The instantaneous volumetric flow rate is obtained from soft measurement. This represents the effective flow area of ​​the valve orifice.

[0080] Step S303: Calculate the particle mass flux per unit area using the following formula. :

[0081] ;

[0082] in, This represents the solid phase mass fraction. For the density of the fluid-particle mixture, This is the impact spot magnification factor. This represents the particle mass flux per unit area.

[0083] In step S4 above, based on the previously calculated particle mass flux per unit area and average nozzle velocity, the wear amount at the valve orifice is quantitatively predicted through precise impact probability analysis and erosion model calculation. First, considering the circumferential impactable projection length of the valve seat and valve core edges, the neutral zone range, the characteristic particle size, and the slot width parameter approximating the valve orifice opening, the probability distribution of particle impacts on the valve seat and valve core edges is calculated using the established impact probability model. This calculation accurately reflects the trajectory characteristics of particles of different sizes flowing through the valve orifice, providing a theoretical basis for wear distribution. Subsequently, the calculated impact probability parameters are substituted into the calibrated erosion wear model to calculate the total wear rate per unit area. This model comprehensively considers multiple key factors, including the comprehensive calibration coefficient, velocity influence index, material hardness ratio and its influence index, and particle incident angle influence function. Among these, the incident angle of the particle relative to the wall normal has a decisive influence on the final wear morphology. Based on the obtained total wear rate per unit area, the total wear rate per unit area is integrated over time with the impacted area to obtain the valve orifice wear volume increment. As a wear measure, the total wear rate per unit area is scientifically allocated to the valve seat edge and valve core edge according to the pre-calculated impact probability, forming an accurate wear distribution prediction, which provides direct data support for subsequent geometric parameter updates and maintenance decisions. Specifically, step S4 may include:

[0084] Step S401: Assume that in the circumferentially impactable projection length, For the corresponding length of the valve seat edge, For the corresponding length of the valve core edge, The length corresponding to the neutral region. The characteristic particle size is represented by the valve opening. Approximate seam width parameters The probability of a particle impacting the valve seat edge is calculated using the following formula. The probability of impacting the valve core edge :

[0085] ;

[0086] ;

[0087] Step S402: Calculate the average nozzle velocity Particle mass flux per unit area Substituting into the erosion model yields the total wear rate per unit area. :

[0088] ;

[0089] in, Comprehensive calibration coefficients, Speed ​​affects the index. Hardness ratio, e-hardness influence index, The influence function of the incident angle, Let be the incident angle of the particle relative to the wall normal. The total wear rate per unit area is calculated. Then, the total wear rate per unit area is distributed to the valve seat edge and valve core edge according to the impact probability, thus obtaining the wear rate of the valve seat edge. and valve core edge wear rate .

[0090] ;

[0091] .

[0092] In step S5 above, the final quantification of wear amount within this cycle and the dynamic update of key geometric parameters are completed, and the valve orifice flow model is corrected in real time accordingly. First, based on the wear rate per unit area calculated in step S4, combined with the erosion area parameter, the cumulative wear volume of the valve orifice is obtained by integrating over one operating cycle. This calculation simultaneously considers material density and position weight coefficients obtained based on impact probability analysis, realizing the transformation of two-dimensional wear rate data into three-dimensional wear volume. Subsequently, the wear volume increment is substituted into the "erosion-rounded corner" geometric evolution relationship model, using the rounded corner radius recorded in the previous cycle as the initial value and applying physical constraints, to solve for the new rounded corner radii of the valve seat edge and valve core edge respectively; and the updated flow area value under the corresponding opening degree is calculated accordingly. On this basis, the flow soft measurement neural network model is corrected online through geometric coupling: the updated value is used as the explicit input or geometric constraint term of the model, and the calibration layer / output layer of the model is updated in small steps to obtain the corrected valve orifice flow prediction result. The updated geometric and model parameters are used as both the wear assessment output for this cycle and the initial conditions for the next calculation cycle, forming a closed-loop geometric feedback and soft measurement model self-calibration mechanism, providing dynamic basis for predictive maintenance and fault diagnosis.

[0093] Step S5 specifically includes:

[0094] Step S501: Within this cycle, first integrate the total wear rate per unit area obtained in step S4 with the impacted area over time to obtain the valve port wear volume increment. As a measure of wear and tear:

[0095] ;

[0096] in, For the impact area, For material density, This represents the impact probability. When calculating the erosion volume at the valve seat edge... Take as When calculating the erosion volume of the valve core edge Take as .

[0097] Step S502: Calculate the obtained wear volume increment. Substituting the geometric relationship of "erosion fillet", solve for the radius of the new fillet:

[0098] ;

[0099] Known Compared with the previous period radius Numerical solution .constraint and If we divide the valve seat into two sides, we can obtain the fillet radii for each side. and valve core edge radius .

[0100] Step S503: Update the flow area with the new fillet radius and form an opening-area mapping to obtain the updated valve seat fillet radius. and valve core edge radius Then, the effective flow area for the current cycle is calculated using the following formula. :

[0101] ;

[0102] in, , , This is the area-corner radius sensitivity coefficient obtained through calibration tests or simulations.

[0103] Step S504: Obtain the effective flow area for the current cycle. The flow rate soft measurement model is introduced as a geometric constraint and lightweight online calibration is performed to make the model adaptive to wear. A multiplicative coupled physical prior is adopted to ensure that geometric changes are reflected in the flow rate in real time. The specific calculation is as follows:

[0104] ;

[0105] in, The basic predicted flow rate after geometric coupling, in m³ / s; The baseline outflow coefficient is dimensionless. , For valve opening, This represents the pressure difference across the valve, in MPa. Oil temperature, in degrees Celsius. Network learning correction factor. Dimensionless; geometrical change is determined by the effective flow area. Direct drive, unit is m 2 ; This refers to the density of the oil, expressed in kg / m³. 3 A pseudo-calibration layer is used to absorb residual error based on multiplicative coupling. Weighted ridge regression is then performed on the most recent N samples.

[0106] ;

[0107] in, To update the output flow rate after soft flow measurement based on the wear model, , For affine calibration parameters, As a proportion, For bias; , The optimal ratio and bias parameters obtained in this calibration; The basic predicted flow for the i-th sample; The ridge regression regularization coefficient; This is a one-time calibration coefficient; The effective flow area curve after wear update for the current period / iteration k+1 at the opening. The value at; Let be the pressure difference across the valve at the i-th sample time. These are the sample weights.

[0108] The hydraulic valve flow rate soft measurement method for online wear prediction provided in this invention has the following advantages:

[0109] 1. Based on the "erosion-geometry-soft measurement" closed loop, accurate predictive maintenance is achieved. The wear rate and wear volume per unit area are calculated in real time, the valve port flow area is updated online and the flow rate soft measurement model is calibrated. It can issue early warnings before the performance deteriorates significantly, scientifically plan the maintenance time, significantly reduce the risk of unplanned downtime caused by sudden valve port failure, and ensure the continuity and stability of production. 2. By adopting neural network flow soft measurement technology, the valve orifice flow can be calculated with high accuracy using only conventional and easily measurable data such as oil temperature, valve orifice opening, and pressure difference. This avoids dependence on expensive high-bandwidth flow sensors and reduces system hardware complexity and deployment costs. When equipped with sensors, it can also be fused and cross-calibrated with measured data to further improve measurement reliability and robustness. 3. By integrating data-driven approaches with physical mechanisms, the nonlinear fitting capability of neural networks is combined with the valve erosion model. The online update of the flow area is used as a geometric constraint to ensure the model's adaptability to complex working conditions such as pressure fluctuations, temperature and viscosity changes, and surface roughness evolution. This also ensures that the prediction results have clear physical meaning and traceability, making wear and flow prediction more accurate and reliable. 4. Supports dynamic optimization of the control system, updating key geometric parameters and soft measurement models of the valve port in real time according to the wear status, enabling the controller to implement adaptive feedforward / feedback compensation, reducing overshoot and oscillation, maintaining stable flow characteristics and control performance throughout the valve's entire life cycle, significantly extending equipment service life and improving overall operating efficiency.

[0110] like Figure 2 The diagram illustrates a schematic of a hydraulic valve flow rate soft measurement system for online wear prediction provided in another embodiment of the present invention. This system is used to implement the hydraulic valve flow rate soft measurement method for online wear prediction described in the first embodiment of the present invention, and includes:

[0111] The model training module is used to collect the operating data of the hydraulic valve under different working conditions as input features, train the neural network soft measurement model for flow prediction, and obtain the trained flow soft measurement model.

[0112] The flow prediction module is used to predict the valve orifice flow based on the trained flow soft measurement model and real-time collected hydraulic valve operating data.

[0113] The first calculation module is used to calculate the effective flow area of ​​the valve orifice, the average velocity of the nozzle, and the particle mass flux per unit area based on the predicted valve orifice flow rate and valve orifice geometric parameters.

[0114] The second calculation module is used to calculate the particle impact probability based on the particle impact characteristics, allocate the total wear amount to the valve seat edge and the valve core edge based on the particle impact probability, and calculate the erosion wear rate in combination with the erosion model.

[0115] The online calibration module is used to output the valve port wear amount based on the erosion wear rate, update the valve port flow area based on the valve port wear amount, and perform online calibration of the flow soft measurement model based on the updated valve port flow area.

[0116] The execution process of each module can be carried out according to the process steps of the hydraulic valve flow soft measurement method for online prediction of wear provided in the first embodiment, and will not be described in detail in this embodiment.

[0117] The hydraulic valve flow rate soft measurement system for online wear prediction provided in this embodiment of the invention is based on the same inventive concept and has the same beneficial effects as the hydraulic valve flow rate soft measurement method for online wear prediction provided in the first embodiment above, and will not be described again here.

[0118] Another embodiment of the present invention provides an electronic device, which includes a processor, an input device, an output device, and a memory. The processor, the input device, the output device, and the memory are interconnected. The memory is used to store a computer program, which includes program instructions. The processor is configured to call the program instructions to execute the method described in the first embodiment above.

[0119] It should be understood that, in the embodiments of the present invention, the processor may be a Central Processing Unit (CPU), but it may also be other general-purpose processors, graphics processing units (GPUs), digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0120] Input devices may include touchpads, microphones, etc., and output devices may include displays (LCDs, etc.), speakers, etc.

[0121] The memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory may also include non-volatile random access memory. For example, the memory may also store information about the device type.

[0122] In specific implementations, the processor, input device, and output device described in the embodiments of the present invention can execute the implementation methods described in the method embodiments of the present invention, or they can execute the implementation methods described in the system embodiments of the present invention, which will not be repeated here.

[0123] The present invention also provides an embodiment of a computer-readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, cause the processor to perform the methods described in the above embodiments.

[0124] The computer-readable storage medium can be an internal storage unit of the terminal described in the foregoing embodiments, such as the terminal's hard drive or memory. The computer-readable storage medium can also be an external storage device of the terminal, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the terminal. Furthermore, the computer-readable storage medium can include both internal storage units and external storage devices of the terminal. The computer-readable storage medium is used to store the computer program and other programs and data required by the terminal. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0125] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0126] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the terminals and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0127] In the several embodiments provided in this application, it should be understood that the disclosed terminals and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices or units, or may be electrical, mechanical or other forms of connection.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A soft measurement method for hydraulic valve flow rate with online wear prediction, characterized in that, include: The operating data of hydraulic valves under different working conditions are collected as input features to train a neural network soft measurement model for flow prediction, thus obtaining a trained flow soft measurement model. Predict valve orifice flow rate based on a trained flow soft measurement model and real-time collected hydraulic valve operating data; The effective flow area, average nozzle velocity, and particle mass flux per unit area of ​​the valve orifice are calculated based on the predicted orifice flow rate and valve orifice geometric parameters. The particle impact probability is calculated based on the particle impact characteristics. The total area wear rate is calculated based on the average nozzle velocity and the particle mass flux per unit area. The total area wear rate is allocated to the valve seat edge and the valve core edge based on the particle impact probability. The erosion wear rate is calculated in conjunction with the erosion model. The valve orifice wear amount is output based on the erosion wear rate, the valve orifice flow area is updated based on the valve orifice wear amount, and the flow soft measurement model is calibrated online based on the updated valve orifice flow area.

2. The hydraulic valve flow rate soft measurement method for online wear prediction as described in claim 1, characterized in that, The flow rate soft measurement model adopts a feedforward regression network structure. The input layer incorporates three features: oil temperature, valve opening, and pressure difference across the valve. Several fully connected layers are set in the middle and ReLU activation function is used. The output layer is a single linear node to output continuous flow rate.

3. The hydraulic valve flow rate soft measurement method for online wear prediction as described in claim 1, characterized in that, The flow soft measurement model predicts flow based on the following formula: ， in, It is the flow coefficient. It is the effective flow area of ​​the valve port. It is the density of the oil. It is the pressure difference across the valve.

4. The hydraulic valve flow rate soft measurement method for online wear prediction as described in claim 1, characterized in that, The specific method for calculating the effective flow area of ​​the valve orifice, the average nozzle velocity, and the particle mass flux per unit area based on the predicted valve orifice flow rate and valve orifice geometric parameters includes: Calculate the effective flow area of ​​the valve port based on the nominal radius of the valve seat, the assembly offset, the valve port opening, and the fillet radii of the valve seat edge and the valve core edge. Calculate the average nozzle velocity based on the predicted flow rate and effective flow area; Calculate the particle mass flux per unit area based on the solid mass fraction, mixture density, predicted flow rate, and effective flow area.

5. The hydraulic valve flow rate soft measurement method for online wear prediction as described in claim 4, characterized in that, The calculation of the total area wear rate based on the average nozzle velocity and particle mass flux per unit area, and the allocation of the total area wear rate to the valve seat edge and valve core edge based on the particle impact probability, specifically includes: Based on the circumferential impactable projection length of the valve seat edge and the valve core edge, the range of the neutral zone, the characteristic particle size, and the valve opening, the probability of a particle impacting the valve seat edge and the probability of impacting the valve core edge are calculated respectively. Based on the nozzle velocity, particle mass flux per unit area, and calibrated erosion model parameters, the total wear rate per unit area is calculated, and the total wear rate is distributed to the valve seat edge and valve core edge according to the probability of impacting the valve seat edge and the probability of impacting the valve core edge.

6. The hydraulic valve flow rate soft measurement method for online wear prediction as described in claim 5, characterized in that, The process of outputting valve orifice wear amount based on erosion wear rate, updating valve orifice flow area based on valve orifice wear amount, and performing online calibration of flow soft measurement model based on updated valve orifice flow area specifically includes: The total wear rate per unit area is integrated over time within one operating cycle to obtain the cumulative wear volume increment at the valve orifice as the wear amount. Substitute the wear volume increment into the erosion-filler geometric evolution relationship model to solve for the updated valve seat fillet radius and valve core fillet radius; Calculate the updated effective flow area of ​​the valve port based on the updated fillet radius of the valve seat and the fillet radius of the valve core; The updated effective flow area is introduced as a geometric constraint into the flow soft measurement model, and the flow soft measurement model is calibrated online to correct the flow prediction output.

7. The hydraulic valve flow rate soft measurement method for online wear prediction as described in claim 6, characterized in that, The online calibration uses a weighted ridge regression method to linearly correct the output of the flow soft measurement model.

8. A hydraulic valve flow rate soft measurement system for online wear prediction, characterized in that, A soft measurement method for hydraulic valve flow rate as described in any one of claims 1 to 7, comprising: The model training module is used to collect the operating data of the hydraulic valve under different working conditions as input features, train the neural network soft measurement model for flow prediction, and obtain the trained flow soft measurement model. The flow prediction module is used to predict the valve orifice flow based on the trained flow soft measurement model and real-time collected hydraulic valve operating data. The first calculation module is used to calculate the effective flow area of ​​the valve orifice, the average velocity of the nozzle, and the particle mass flux per unit area based on the predicted valve orifice flow rate and valve orifice geometric parameters. The second calculation module is used to calculate the particle impact probability based on the particle impact characteristics, calculate the total area wear rate based on the average nozzle velocity and the particle mass flux per unit area, allocate the total area wear rate to the valve seat edge and the valve core edge based on the particle impact probability, and calculate the erosion wear rate in conjunction with the erosion model. The online calibration module is used to output the valve port wear amount based on the erosion wear rate, update the valve port flow area based on the valve port wear amount, and perform online calibration of the flow soft measurement model based on the updated valve port flow area.

9. An electronic device comprising a processor, an input device, an output device, and a memory, wherein the processor, input device, output device, and memory are interconnected, and the memory is used to store a computer program, the computer program comprising program instructions, characterized in that, The processor is configured to invoke the program instructions to execute the hydraulic valve flow soft measurement method for online prediction of wear as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the soft measurement method for online prediction of wear of hydraulic valve flow as described in any one of claims 1 to 7.

Images