Reliability prediction method for fuel injector electromagnetic valve based on multi-physical field coupling

Abstract

The application provides a fuel injector electromagnetic valve reliability prediction method based on multi-physical field coupling, relates to the technical field of predicting electromagnetic valve reliability, and comprises the following steps: collecting real-time operation data and historical operation data, constructing a multi-physical field coupling model of the electromagnetic valve, inputting the historical operation data into the model in sequence, solving the model by using a finite element analysis method, obtaining multi-physical field distribution characteristics of the electromagnetic valve, extracting historical reliability parameters in the multi-physical field distribution characteristics, adding category labels to the historical reliability parameters, inputting real-time operation data into the multi-physical field coupling model, obtaining real-time reliability parameters, inputting the real-time reliability parameters into a BI-LSTM model, and outputting a prediction result. The application combines a multi-physical field model and a machine learning model, can comprehensively utilize the mechanism advantages of the physical model and the data driving advantages of the machine learning model, and thus can more accurately predict the reliability of the electromagnetic valve.

Description

Technical Field

[0001] This application relates to the technical field of predicting the reliability of solenoid valves, and in particular to a method for predicting the reliability of fuel injector solenoid valves based on multi-physics coupling. Background Technology

[0002] In the field of new energy internal combustion engines, whether it is high-pressure common rail diesel fuel injectors and gasoline direct injection (GDI) fuel injectors used in hybrid electric vehicles (including plug-in hybrid electric vehicles, range-extended hybrid electric vehicles, etc.) or gas fuel injectors adapted to clean energy internal combustion engines such as natural gas and hydrogen, the reliability of their solenoid valves is crucial to the stable operation of the entire system.

[0003] The Chinese invention patent with publication number CN110852014A establishes electromagnetic differential equations and valve core dynamic equations for the dynamic process of an electromagnetic valve based on multiple physical fields involved in the valve and using theoretical knowledge. Based on the established mathematical physical model, some physical field input parameters required in the equations are calculated using finite element software. Then, the mathematical model of the electromagnetic valve is jointly simulated using mathematical calculation software to further study the performance law of the electromagnetic valve.

[0004] The aforementioned patent relies solely on specific software to calculate and obtain limited physical field parameters, failing to consider that in new energy internal combustion engines, the reliability of the fuel injector solenoid valve depends not only on its own physical structure and working principle, but also on the operating status of the entire internal combustion engine system.

[0005] Because this patent does not take into account the operating conditions of new energy internal combustion engines, it is difficult to accurately predict the reliability of fuel injector solenoid valves in actual use, and it cannot provide effective support for the maintenance and fault prevention of new energy internal combustion engines. Summary of the Invention

[0006] To improve the accuracy of reliability prediction for fuel injector solenoid valves in new energy internal combustion engines, this application provides a reliability prediction method for fuel injector solenoid valves based on multi-physics coupling.

[0007] The technical solution adopted in this application is as follows:

[0008] A reliability prediction method for fuel injector solenoid valves based on multiphysics coupling includes the following steps:

[0009] Data acquisition: Collect real-time operating data of the solenoid valve and multiple sets of historical operating data of the same type of solenoid valve to construct a multi-physics coupling model of the solenoid valve. The operating data includes electromagnetic force data, temperature data, fluid pressure data, vibration data, operating voltage and operating current.

[0010] Extracting reliability parameters: Input each set of historical operating data into the multiphysics coupling model in sequence, and solve the multiphysics coupling model using the finite element analysis method to obtain the multiphysics distribution characteristics of the solenoid valve. Extract historical reliability parameters from the multiphysics distribution characteristics and add category labels to the historical reliability parameters. The historical reliability parameters include: temperature gradient of the solenoid coil, stress concentration factor of the valve core, and pressure pulse amplitude of fluid flow.

[0011] Prediction: Input real-time running data into the multiphysics coupling model to obtain real-time reliability parameters, construct a BI-LSTM model using machine learning algorithms, train the BI-LSTM model using historical reliability parameters and category labels to obtain the trained BI-LSTM model, input the real-time reliability parameters into the trained BI-LSTM model, and output the prediction results.

[0012] This application first collects real-time operating data of the solenoid valve (electromagnetic force, temperature, fluid pressure, vibration, operating voltage, and operating current) and multiple sets of historical operating data for similar solenoid valves. Then, this application uses the collected data to construct a multiphysics coupling model, which comprehensively considers the interactions between multiple physical fields such as electromagnetic, thermal, fluid, and mechanical fields, more realistically reflecting the actual working condition of the solenoid valve. Subsequently, this application solves the multiphysics coupling model using the finite element method to obtain the multiphysics distribution characteristics of the solenoid valve. Historical reliability parameters are extracted from these multiphysics distribution characteristics and category labels are added to the historical reliability parameters. Next, this application inputs the real-time operating data into the multiphysics coupling model to obtain real-time reliability parameters. Then, a BI-LSTM (Bidirectional Long Short-Term Memory) model is constructed using a machine learning algorithm. The BI-LSTM model is trained using the historical reliability parameters and category labels, enabling the BI-LSTM model to learn patterns and features from historical data. Finally, this application inputs the real-time reliability parameters into the trained BI-LSTM model and outputs the prediction result, i.e., whether the solenoid valve is reliable under the current conditions. This application combines a multiphysics model and a machine learning model, which can comprehensively utilize the mechanistic advantages of the physical model and the data-driven advantages of the machine learning model, thereby more accurately predicting the reliability of the solenoid valve.

[0013] Optionally, the method further includes:

[0014] Simulation: Construct a digital twin model, embed a multiphysics coupling model into the digital twin model, and input historical operating data into the digital twin model in chronological order of acquisition time, so that the digital twin model can simulate the degradation process of the solenoid valve and obtain the simulated degradation data of the solenoid valve at different times.

[0015] Simulation judgment: Obtain the real degradation data at the same time as the simulation degradation data, calculate the difference between the simulation degradation data and the real degradation data, and record it as the first data. Determine whether the first data is greater than the preset difference threshold. If so, the digital twin model is corrected, and the simulation degradation data is re-obtained based on the corrected digital twin model. If not, the step of calculating the degree of degradation is executed.

[0016] Calculate the degree of degradation: Obtain the running data for a preset period of time before the current moment, and record it as the second data. Input the second data and the real-time running data into the digital twin model in the order of collection time to obtain the predicted degradation data. Calculate the predicted degree of degradation based on the predicted degradation data.

[0017] Degradation judgment: Determine whether the predicted degradation degree is greater than the preset degradation degree threshold. If so, issue a degradation alarm signal; if not, repeat this step after a preset interval.

[0018] This application creates a virtual solenoid valve model by constructing a digital twin model and embedding a multiphysics coupling model within it. This model can simulate the physical characteristics and behavior of the solenoid valve under various operating conditions. The digital twin model possesses high flexibility and scalability, easily integrating different physical models and data sources, providing a powerful platform for solenoid valve degradation simulation. Subsequently, this application inputs historical operating data into the digital twin model in chronological order of acquisition, enabling the digital twin model to simulate the degradation process of the solenoid valve during actual operation, obtaining simulation degradation data from different periods. By adopting the above scheme, this application fully considers the historical operating information of the solenoid valve, making the simulation results closer to reality.

[0019] This application also acquires real degradation data at the same time as the simulated degradation data and calculates the difference between the simulated degradation data and the real degradation data. It then determines whether this difference exceeds a preset threshold. By adopting this scheme, this application can promptly detect errors in the digital twin model. By adjusting the parameters or structure of the digital twin model, it can make the digital twin model more accurately reflect the actual degradation of the solenoid valve, improving the reliability and accuracy of the digital twin model. When the difference exceeds the preset threshold, the digital twin model is corrected, and simulated degradation data is reacquired based on the corrected digital twin model. This process continuously optimizes the model, making the simulation results of the digital twin model gradually approach the real situation. Subsequently, this application acquires operating data (i.e., second data) for a preset duration prior to the current time. The second data and real-time operating data are input into the digital twin model in the order of acquisition time to obtain predicted degradation data. Based on the predicted degradation data, the predicted degradation degree is calculated. Then, it is determined whether the predicted degradation degree exceeds a preset degradation degree threshold. When it exceeds the preset degradation degree threshold, a degradation alarm signal is issued, enabling personnel to promptly detect potential faults and degradation risks of the solenoid valve.

[0020] Optionally, the method further includes:

[0021] Update the model: Increase the grid density of the target region in the digital twin model to obtain a new digital twin model. Extract a new multiphysics coupling model from the new digital twin model and perform the step of extracting reliability parameters.

[0022] In digital twin models, mesh density directly affects simulation accuracy. By increasing the mesh density of the target region, this application can more finely divide the computational units of the target region, capturing more subtle physical phenomena and changes. The target region is often where problems are prone to occur or where key performance indicators are concentrated during the operation of the solenoid valve. Increasing the mesh density of this region enhances the simulation and analysis capabilities for these local areas, allowing for a deeper understanding of their physical characteristics and behavioral patterns. The new digital twin model obtained after increasing the mesh density is an optimization and improvement of the original model. While retaining the overall framework and functionality of the original model, the new digital twin model performs a more refined simulation of the target region, better reflecting the actual operation of the solenoid valve. Because the mesh density of the target region changes in the new digital twin model, the distribution and interaction of its internal physical fields also change accordingly. Therefore, this application also extracts a new multiphysics coupling model from the new digital twin model, enabling the new multiphysics coupling model to reflect these changes in a timely manner, improving the timeliness and accuracy of the analysis results. Subsequently, this application uses the new multiphysics coupling model to extract reliability parameters, obtaining more accurate results. The new multiphysics coupling model provides a more refined simulation of the solenoid valve's operating state, enabling more accurate calculation of reliability parameters. This improves the accuracy of the BI-LSTM model training samples, thereby enhancing the accuracy of the prediction results.

[0023] Optionally, the method further includes:

[0024] Establish a degradation model: Based on the simulated degradation data, a degradation curve is plotted, and a degradation model is established to fit the degradation curve. The calculation model of the degradation model is as follows:

[0025] ;

[0026] in, The degree of degradation at time t; for The degree of degradation over time; The parameters are determined using the nonlinear least squares method. The value of ;

[0027] Calculate reliability: based on parameters Calculate and output the reliability of the solenoid valve, the reliability... The calculation model is as follows:

[0028] ;

[0029] ;

[0030] Where n is an intermediate variable; This refers to the operating time of the solenoid valve.

[0031] This application constructs degradation curves and establishes a degradation model based on real degradation data, making full use of a large amount of data accumulated during historical operation. This data contains information on the performance changes of the solenoid valve under different operating conditions, accurately reflecting the degradation pattern of the solenoid valve. By establishing the model through a data-driven approach, this application reduces the bias that may arise from relying solely on theoretical assumptions and empirical formulas, improving the accuracy and reliability of the degradation model. Subsequently, this application uses the nonlinear least squares method to determine the parameters in the degradation model, enabling the degradation model to better fit the actual degradation curve. Then, this application calculates the reliability based on the parameters in the degradation model, combining the degree of degradation with the service life of the solenoid valve to reflect the reliability status of the solenoid valve at different stages of use. By calculating the reliability, the probability of the solenoid valve operating normally within a specific time period can be quantified, providing a scientific basis for equipment maintenance decisions.

[0032] Optionally, the method further includes:

[0033] Constructing the evaluation matrix: Obtain the vector output by the BI-LSTM model after each round of training, construct the evaluation matrix based on the vector, and standardize the rows or columns of the evaluation matrix to obtain the processed evaluation matrix. The element in the i-th row and j-th column of the evaluation matrix represents the probability of misclassifying the i-th class as the j-th class.

[0034] Obtain the dataset: Combine the off-diagonal elements in the rows or columns of the evaluation matrix into a dataset; calculate the difference between any two off-diagonal elements in the dataset, and denote it as the third data. For the third data that is less than a preset threshold, execute the step of generating samples until all datasets have been traversed.

[0035] Sample generation: The training samples corresponding to the third data are generated using the GANs model, and the trained BI-LSTM model is trained again using the training samples to obtain the retrained BI-LSTM model. The training samples include new reliability parameters and new category labels.

[0036] In the prediction step, real-time reliability parameters are input into the retrained BI-LSTM model.

[0037] This application obtains the output vectors of the BI-LSTM model after each training round and constructs an evaluation matrix based on these vectors. The element in the i-th row and j-th column of the evaluation matrix represents the probability of misclassifying class i as class j. This construction method can intuitively reflect the classification performance of the BI-LSTM model among different classes. Subsequently, this application standardizes the rows or columns of the evaluation matrix to make the misclassification probabilities between different classes comparable. Then, this application integrates the off-diagonal elements in the rows or columns of the evaluation matrix into a dataset. The off-diagonal elements represent the model's misclassification among different classes. By integrating these elements, the classification performance of the model among easily confused classes can be analyzed in a focused manner.

[0038] Subsequently, this application calculates the difference between any two off-diagonal elements in the dataset, denoted as the third data. For the third data with a difference less than a preset threshold, a sample generation step is performed to identify cases where the BI-LSTM model has similar classification performance between easily confused categories, i.e., category pairs with a difference less than the preset threshold. When the difference is less than the preset threshold, a GAN model is used to generate training samples for the two categories corresponding to the third data. The GAN model has a powerful generation capability, capable of generating samples similar to the distribution of real data. By generating these samples, the amount of training data for the BI-LSTM model between these easily confused categories can be increased, helping the model to better learn the features of these categories and improve classification performance. Subsequently, this application uses the generated training samples to retrain the trained BI-LSTM model, obtaining a retrained BI-LSTM model. The retraining process allows the BI-LSTM model to adapt to the newly generated samples, further optimizing the model parameters and improving the discrimination of the BI-LSTM model between easily confused categories. The training samples include new reliability parameters and new category labels, which helps the BI-LSTM model learn richer feature information and improve the model's generalization ability. In the prediction step, this application inputs the real-time reliability parameters into the retrained BI-LSTM model. Since the retrained BI-LSTM model has been optimized for easily confused categories, it can more accurately predict the category to which the real-time reliability parameters belong, thus improving the accuracy and reliability of the prediction results.

[0039] Optionally, the diagonal elements of the evaluation matrix are 0.

[0040] Optionally, the method further includes:

[0041] Obtain the vector: Concatenate the vectors into a feature matrix, calculate the maximum eigenvalue of the feature matrix and the corresponding eigenvector, and obtain the vector corresponding to the maximum value of the elements in the feature vector, denoted as the first vector;

[0042] Correct judgment: Determine whether the historical reliability parameter and category label corresponding to the first vector are correct. If yes, divide the evaluation matrix according to the position of the maximum value in the feature vector to obtain two sub-matrices. If no, reset the category label of the historical reliability parameter corresponding to the first vector, retrain the BI-LSTM model, and execute the step of constructing the evaluation matrix.

[0043] Vector filtering: Based on the vectors in the two sub-matrices, obtain the loss value of the trained BI-LSTM model, retain the vectors in the sub-matrices with smaller loss values, and delete the vectors in the sub-matrices with larger loss values.

[0044] This application concatenates the vectors output by the BI-LSTM model into a feature matrix, calculates its largest eigenvalue and corresponding eigenvector, and then obtains the vector corresponding to the maximum value of each element in the feature vector (the first vector). Subsequently, this application determines whether the historical reliability parameter and class label corresponding to the first vector are correct. If the determination is correct, the evaluation matrix is ​​divided into two sub-matrices based on the position of the maximum value in the feature vector; if the determination is incorrect, the class label needs to be reset and the model retrained. Then, this application divides the evaluation matrix into two sub-matrices based on the position of the maximum value in the feature vector, and calculates the loss value of the trained BI-LSTM model based on the vectors in each sub-matrice. Vectors in the sub-matrix with smaller loss values ​​are retained, while vectors in the sub-matrix with larger loss values ​​are deleted. The loss value is an important indicator of how well the BI-LSTM model fits the vector; a smaller loss value indicates a better fit to the vector, and that the vector is more helpful in improving the performance of the BI-LSTM model. This filtering method removes poor-quality vectors and retains good-quality vectors, thereby improving the quality of the evaluation matrix.

[0045] Optionally, the GANs model is trained using historical reliability parameters and category labels to obtain the trained GANs model;

[0046] In the sample generation step, the trained GANs model is used to generate training samples for the two categories corresponding to the third data.

[0047] This application employs historical reliability parameters and category labels to train GANs models, making full use of existing rich historical data resources. Through this approach, GANs models can learn the distribution characteristics and patterns of different categories of samples in historical data, thereby generating high-quality training samples.

[0048] Optionally, the PCA algorithm is used to reduce the dimensionality of the historical reliability parameters and the real-time reliability parameters, and the dimensionality-reduced historical reliability parameters and the real-time reliability parameters are used as new historical reliability parameters and new real-time reliability parameters.

[0049] Historical and real-time reliability parameters typically contain multiple feature dimensions. The PCA algorithm projects the original data into a new coordinate system through linear transformation, selecting the principal components with the largest variances as new feature dimensions, thus effectively reducing the dimensionality of the data. Dimensionality reduction reduces redundant information, retains the main features of the data, and makes the data simpler and easier to process.

[0050] In summary, this application includes at least one of the following beneficial technical effects:

[0051] 1. This application combines a multiphysics model and a machine learning model, which can comprehensively utilize the mechanistic advantages of the physical model and the data-driven advantages of the machine learning model, thereby more accurately predicting the reliability of the solenoid valve.

[0052] 2. By constructing a digital twin model and embedding a multiphysics coupling model within it, this application is able to create a virtual solenoid valve model that can simulate the physical characteristics and behavior of the solenoid valve under various operating conditions. Attached Figure Description

[0053] Figure 1 This is a flowchart of Embodiment 1 of this application;

[0054] Figure 2 This is a flowchart of the S21 simulation to S25 degradation judgment in Embodiment 2 of this application;

[0055] Figure 3 This is a flowchart of Embodiment 3 of this application. Detailed Implementation

[0056] The following combination Figures 1 to 3 This application will be described in further detail.

[0057] Example 1: This example discloses a reliability prediction method for fuel injector solenoid valves based on multiphysics coupling, referring to... Figure 1The method includes: S11 data acquisition, S12 reliability parameter extraction, and S13 prediction. First, real-time operating data of the solenoid valve and multiple sets of historical operating data of the same type (including electromagnetic force, temperature, fluid pressure, vibration, operating voltage, and current) are collected to construct a multiphysics coupling model. Next, each set of historical operating data is input into the multiphysics coupling model, and finite element analysis is used to solve for the multiphysics distribution characteristics. Historical reliability parameters such as the solenoid coil temperature gradient, valve core stress concentration factor, and fluid flow pressure pulse amplitude are extracted and labeled. Finally, real-time operating data is input into the model to obtain real-time reliability parameters. A BI-LSTM model is constructed using a machine learning algorithm, trained with historical reliability parameters and labels, and then the real-time reliability parameters are input into the trained model to output the prediction result. The process in this embodiment is as follows:

[0058] S11 data acquisition collects real-time operating data of the solenoid valve and multiple sets of historical operating data of the same type of solenoid valve to construct a multi-physics coupling model of the solenoid valve. The operating data includes electromagnetic force data, temperature data, fluid pressure data, vibration data, operating voltage and operating current.

[0059] Real-time operating data reflects the current working status of the solenoid valve, including electromagnetic force data (showing the magnitude and changes of electromagnetic force when the solenoid valve is working, affecting the movement of the valve core), temperature data (reflecting the thermal state of the solenoid valve as a whole and key parts (such as the solenoid coil); excessively high temperature may affect the performance and lifespan of the solenoid valve), fluid pressure data (related to the fluid system controlled by the solenoid valve; abnormal pressure may lead to seal failure or obstruction of valve core movement), vibration data (may be caused by the movement of internal parts of the solenoid valve, fluid impact, etc.; excessive vibration may cause parts to loosen or fatigue damage), and operating voltage and operating current data (providing energy to the solenoid valve; fluctuations or abnormalities in these values ​​will affect the electromagnetic performance and heat generation of the solenoid valve).

[0060] This step also collected multiple sets of historical operating data for similar solenoid valves. This data can provide long-term operating experience and trends, which helps to uncover the operating patterns of solenoid valves under different working conditions.

[0061] In practical operation, solenoid valves involve complex interactions between multiple physical fields, including electromagnetic, temperature, flow, and structural fields. For example, electromagnetic forces act on the valve core, causing it to move. This movement of the valve core, in turn, alters the flow field, leading to changes in temperature distribution. Simultaneously, temperature changes affect the electromagnetic properties of the solenoid coil, such as its resistance. A multiphysics coupling model comprehensively considers and models these multiple physical fields to more accurately describe the actual operation of the solenoid valve. This embodiment employs professional simulation software (such as ANSYS and COMSOL) based on theories of electromagnetics, thermodynamics, fluid mechanics, and solid mechanics to establish the governing equations for each physical field and couple them together. For instance, in the electromagnetic field calculation, the electromagnetic force is calculated according to Maxwell's equations; in the temperature field calculation, factors such as the heating of the solenoid coil and convective heat transfer of the fluid are considered; in the flow field calculation, fluid flow is simulated based on the Navier-Stokes equations; and in the structural field calculation, the stress distribution of components such as the valve core is analyzed based on the stress-strain relationship.

[0062] S12 extracts reliability parameters, inputs each set of historical operating data into the multiphysics coupling model in sequence, and solves the multiphysics coupling model using the finite element analysis method to obtain the multiphysics distribution characteristics of the solenoid valve. Historical reliability parameters are extracted from the multiphysics distribution characteristics, and category labels are added to the historical reliability parameters. The historical reliability parameters include: the temperature gradient of the solenoid coil, the stress concentration factor of the valve core, and the pressure pulse amplitude of the fluid flow.

[0063] Each set of historical operating data is sequentially input into the constructed multiphysics coupling model. The finite element method discretizes the continuous physical field into a finite number of elements. By calculating and analyzing each element, the distribution of the entire physical field is obtained by combining them. For example, in temperature field analysis, the solenoid valve is divided into many small temperature elements, and the temperature value of each element is calculated to obtain the temperature distribution cloud map of the entire solenoid valve. By solving the multiphysics coupling model through finite element analysis, this embodiment can intuitively understand the multiphysics distribution characteristics of the solenoid valve under different historical operating conditions, such as the distribution of the electromagnetic field, the gradient of the temperature field, the pressure and velocity distribution of the flow field, and the stress distribution of the structural field.

[0064] Representative historical reliability parameters are extracted from the multi-physics field distribution characteristics. The temperature gradient of the electromagnetic coil reflects the temperature change within the coil; a large temperature gradient may cause local overheating, affecting its insulation performance and lifespan. The stress concentration factor of the valve core indicates the degree of stress concentration on the valve core; stress concentration areas are prone to fatigue failure. The pressure pulse amplitude of the fluid flow reflects the fluctuation of fluid pressure; excessively large pressure pulses may lead to seal failure or increased valve core vibration.

[0065] Category labels are added to the extracted historical reliability parameters, such as labeling them according to the solenoid valve's operating conditions (e.g., normal operation, light load, heavy load) or fault type (e.g., seal failure, valve core jamming). These labels provide supervision information for the subsequent training of the machine learning model, helping the model learn the characteristics of reliability parameters under different operating conditions and faults.

[0066] S13 prediction involves inputting real-time running data into a multiphysics coupling model to obtain real-time reliability parameters. A BI-LSTM model is then constructed using machine learning algorithms. The BI-LSTM model is trained using historical reliability parameters and category labels to obtain the trained BI-LSTM model. The real-time reliability parameters are then input into the trained BI-LSTM model to output the prediction results.

[0067] In this embodiment, the collected real-time operating data is input into the multiphysics coupling model. The finite element analysis method is also used to solve the multiphysics coupling model to obtain the multiphysics distribution characteristics of the solenoid valve under real-time operating conditions, and the real-time reliability parameters are extracted from it. The real-time reliability parameters reflect the current working reliability and potential risks of the solenoid valve.

[0068] Bidirectional Long Short-Term Memory (BI-LSTM) networks are an improved type of recurrent neural network (RNN) that can simultaneously consider information from the past and future. In solenoid valve reliability prediction, BI-LSTM models can learn the changes in historical reliability parameters over time and the correlations between different parameters. For example, the trend of historical temperature gradient changes may be related to the future stress concentration of the valve core, and the BI-LSTM model can capture this complex temporal relationship.

[0069] This embodiment uses extracted historical reliability parameters and corresponding category labels to train the BI-LSTM model. During training, the BI-LSTM model continuously adjusts its internal parameters to minimize the error between the prediction results of historical data and the actual category labels. By training the BI-LSTM model, the characteristics and variation patterns of the solenoid valve's reliability parameters can be learned.

[0070] In this embodiment, the acquired real-time reliability parameters are input into the trained BI-LSTM model. The BI-LSTM model predicts the reliability of the solenoid valve based on the learned knowledge, and obtains the probability of the solenoid valve failing in the future and the possible types of failures.

[0071] This embodiment first collects real-time operating data of the solenoid valve (including electromagnetic force, temperature, fluid pressure, vibration, operating voltage and current, etc.) and multiple sets of historical operating data of similar solenoid valves. This data is used to construct a coupled model that comprehensively considers the interaction of multiple physical fields, including electromagnetic field, temperature field, flow field, and structural field. Next, finite element analysis is used to input the historical operating data into the coupled model to solve for the multi-physical field distribution characteristics. Historical reliability parameters such as the electromagnetic coil temperature gradient, valve core stress concentration factor, and fluid pressure pulse amplitude are extracted from these parameters and categorized according to operating conditions or fault types. Finally, real-time operating data is input into the coupled model to obtain real-time reliability parameters. A BI-LSTM model is then constructed and trained using the historical reliability parameters and labels. The real-time reliability parameters are then input into the trained model to output prediction results, thus achieving reliability prediction for the fuel injector solenoid valve.

[0072] Example 2: Refer to Figure 2 The difference between this embodiment and Embodiment 1 is that the method further includes:

[0073] S21 simulation constructs a digital twin model, embeds a multiphysics coupling model into the digital twin model, and inputs historical operating data into the digital twin model in chronological order of acquisition time, so that the digital twin model can simulate the degradation process of the solenoid valve and obtain simulated degradation data of the solenoid valve at different times.

[0074] Digital twins fully utilize data such as physical models, sensor updates, and operational history to integrate simulation models from multiple disciplines, physical quantities, scales, and probabilities. Digital twin models can be mapped in virtual space to reflect the entire life cycle of a solenoid valve.

[0075] In actual operation, solenoid valves involve the interaction of multiple physical fields, such as electromagnetic, thermal, and mechanical fields. The electromagnetic field drives the valve core, and this movement generates frictional heat, affecting the valve's temperature distribution. Temperature changes, in turn, can influence the mechanical and electromagnetic properties of the valve material. Multiphysics coupling models (MPCs) are used to describe the interactions and influences between these different physical fields. Embedding this model into a digital twin model allows for a more realistic simulation of the solenoid valve's behavior under complex operating conditions. For example, in simulating the high-frequency switching of a solenoid valve, changes in the electromagnetic field cause rapid valve core movement, generating significant heat. A thermal model can calculate the temperature rise, which in turn affects the elastic modulus of the valve core material, thus influencing its motion characteristics. MPCs accurately reflect these complex interactions and can then be mapped into the digital twin model, enabling the embedding of the MPC model into the digital twin model.

[0076] Historical operating data of the solenoid valve is input into the digital twin model in chronological order of acquisition. This historical data includes various parameters of the solenoid valve under different operating conditions, such as voltage, current, pressure, temperature, and valve core displacement. By inputting this data, the digital twin model can simulate the actual operation of the solenoid valve over a period of time, thereby simulating the degradation process of the solenoid valve. For example, based on information such as the opening and closing time and current magnitude of the solenoid valve recorded in the historical data, the digital twin model can simulate the wear of internal parts of the solenoid valve, such as the wear between the valve core and the valve seat, and thus obtain simulated degradation data of the solenoid valve at different periods to reflect the trend of the solenoid valve's performance gradually declining over time.

[0077] S22 Simulation judgment: Obtain the real degradation data at the same time as the simulation degradation data, calculate the difference between the simulation degradation data and the real degradation data, and record it as the first data. Determine whether the first data is greater than the preset difference threshold. If so, correct the digital twin model and re-obtain the simulation degradation data based on the corrected digital twin model. If not, execute S23 to calculate the degree of degradation.

[0078] To obtain real-world degradation data at the same time as the simulated degradation data, actual degradation data can be obtained through practical testing methods. For example, periodically perform performance tests on solenoid valves, measuring their key performance indicators such as flow rate, pressure loss, opening and closing times, and compare these measurements with the performance indicators of the solenoid valve in its initial state. This yields the actual degradation status of the solenoid valve, and the real-world degradation data reflects the true degree of degradation of the solenoid valve in its actual operating environment.

[0079] This embodiment also calculates the difference between the simulated degradation data and the actual degradation data, denoted as the first data. Then, it determines whether the first data is greater than a preset difference threshold. The preset difference threshold is determined using a statistical analysis algorithm and based on factors such as the performance requirements and operational accuracy of the solenoid valve. If the first data is greater than the preset difference threshold, it indicates a significant deviation between the simulation results of the digital twin model and the actual situation, requiring correction of the digital twin model. Correction methods include adjusting model parameters, optimizing the algorithm of the multiphysics coupling model, and improving the boundary conditions of the digital twin model. After correction, the simulated degradation data is re-acquired based on the corrected digital twin model, and this step is repeated to improve simulation accuracy. If the first data is not greater than the preset difference threshold, it indicates that the simulation results of the digital twin model are relatively accurate, and step S23 can be executed to calculate the degree of degradation.

[0080] S24 calculates the degree of degradation, obtains the running data for a preset duration before the current moment, and records it as the second data. The second data and the real-time running data are input into the digital twin model in the order of collection time to obtain the predicted degradation data. The predicted degradation degree is calculated based on the predicted degradation data.

[0081] Retrieve the operating data for a preset period of time up to the current moment, and record it as the second data. The preset period can be determined based on the degradation characteristics of the solenoid valve and actual needs. For example, if the degradation process of the solenoid valve is relatively slow, the preset period can be set to be longer; if the degradation process is relatively fast, the preset period can be appropriately shortened. The operating data contains the operating status information of the solenoid valve over a past period of time.

[0082] In this embodiment, the second data and real-time operating data are input into the digital twin model in chronological order of acquisition time. The digital twin model simulates the future degradation trend of the solenoid valve under its current operating state based on this data, thereby obtaining predicted degradation data. Subsequently, the predicted degradation degree is calculated based on the predicted degradation data. The predicted degradation degree can be quantified in various ways; for example, a key performance indicator of the solenoid valve (such as flow rate) can be compared with its initial performance indicator, and the percentage decrease can be calculated as the predicted degradation degree.

[0083] S25 Degradation Detection: Determine whether the predicted degradation degree is greater than the preset degradation degree threshold. If so, issue a degradation alarm signal; otherwise, repeat this step after a preset interval.

[0084] Determine if the predicted degradation level exceeds a preset degradation level threshold. If the predicted degradation level exceeds the preset degradation level threshold, it indicates that the performance of the solenoid valve has degraded to a point that affects its normal operation. In this case, a degradation alarm signal should be issued so that maintenance measures can be taken in a timely manner, such as replacing the solenoid valve or performing repairs. If the predicted degradation level does not exceed the preset degradation level threshold, it indicates that the performance of the solenoid valve is still within an acceptable range. This step can be repeated after a preset interval to continue monitoring the degradation status of the solenoid valve.

[0085] In other embodiments, the method further includes:

[0086] S26 updates the model by increasing the grid density of the target region in the digital twin model to obtain a new digital twin model. A new multiphysics coupling model is then extracted from the new digital twin model, and S12 is executed to extract reliability parameters.

[0087] Increasing the mesh density in the target region of the digital twin model is crucial. Target regions are typically critical parts of a solenoid valve prone to degradation or significantly impacting performance, such as the contact area between the valve core and seat, and the winding portion of the solenoid coil. Increasing the mesh density improves the computational accuracy of the digital twin model in these areas, allowing for a more accurate simulation of physical phenomena. For example, in simulating the friction and wear of a solenoid valve, increasing the mesh density in the valve core-seat contact area allows for more precise calculation of parameters such as contact stress and wear rate.

[0088] After obtaining the new digital twin model, a new multiphysics coupling model is extracted from it. Since changes in mesh density may affect the interactions between multiphysics fields, the multiphysics coupling model needs to be re-extracted to accurately reflect the physical characteristics of the solenoid valve. Then, step S12 is executed to extract reliability parameters.

[0089] In other embodiments, the method further includes:

[0090] S27. A degradation model is established, and a degradation curve is plotted based on the simulated degradation data. A degradation model for fitting the degradation curve is established, and the calculation model of the degradation model is as follows:

[0091] ;

[0092] in, The degree of degradation at time t; for The degree of degradation over time; The parameters are determined using the nonlinear least squares method. The value of .

[0093] Degradation curves are plotted based on simulated degradation data, which can intuitively show the degradation trend of the solenoid valve. For example, by plotting time on the x-axis and the degree of degradation of the solenoid valve on the y-axis, the resulting degradation curve can clearly reflect the change in the performance of the solenoid valve over time.

[0094] Nonlinear least squares is a commonly used parameter estimation method. It determines the optimal value of parameter a by minimizing the sum of squared errors between simulated degradation data and degradation model predictions, so that the degradation model can better fit the actual degradation curve.

[0095] S28 calculates reliability based on parameters. Calculate and output the reliability of the solenoid valve, the reliability... The calculation model is as follows:

[0096] ;

[0097] ;

[0098] Where n is an intermediate variable; This refers to the operating time of the solenoid valve.

[0099] Reliability reflects the ability of a solenoid valve to perform its intended function within a specified time. By calculating the reliability of a solenoid valve under different usage periods, its reliability level can be assessed, providing important reference for predicting the valve's service life and formulating maintenance plans. For example, when reliability drops to a certain level, the replacement or repair of the solenoid valve can be arranged in advance to reduce production interruptions or safety accidents caused by solenoid valve failure.

[0100] In this embodiment, the performance analysis process of a solenoid valve first constructs a digital twin model, embeds a multiphysics coupling model, and inputs historical operating data to simulate the degradation process and obtain simulated degradation data at different times. Then, it acquires real degradation data at the same moment and calculates the difference between the actual degradation data and the simulated data (first data). If the difference exceeds a preset threshold, the model is corrected and the simulated data is acquired again; otherwise, the predicted degradation degree is calculated, i.e., the operating data for a preset duration before the current moment (second data) and real-time data are input into the model, and the degradation degree is calculated based on the predicted data. Next, it determines whether the predicted degradation degree exceeds a preset threshold; if it does, an alarm is triggered; otherwise, the determination is repeated after a preset interval. Furthermore, the grid density of the target area of ​​the digital twin model can be increased to update the model, extract a new multiphysics coupling model, and extract reliability parameters. It can also plot degradation curves based on simulated degradation data and use nonlinear least squares to determine parameters and establish a degradation model. Finally, based on the parameters, the reliability of the solenoid valve under its service life is calculated, providing a basis for solenoid valve maintenance, life prediction, etc., ensuring its safe and stable operation.

[0101] Example 3: Reference Figure 3 The difference between this embodiment and Embodiment 2 is that the method further includes:

[0102] S31 Construct an evaluation matrix by obtaining the vector output by the BI-LSTM model after each training round (i.e., the vector composed of the probability distributions output by the BI-LSTM model after each training round). Construct an evaluation matrix based on the vector, and standardize the rows or columns of the evaluation matrix to obtain the processed evaluation matrix. The element in the i-th row and j-th column of the evaluation matrix represents the probability of misclassifying the i-th class as the j-th class. The diagonal elements of the evaluation matrix have a value of 0.

[0103] S32 acquires the dataset, integrating the off-diagonal elements in the rows or columns of the evaluation matrix into a dataset (that is, integrating the probability distributions output after the same round of training, excluding the probability values ​​of the correct classification results into a dataset), calculates the difference between any two off-diagonal elements in the dataset, and records it as the third data (the third data represents the difference in the probability of misclassification; when this difference is very small, it means that the BI-LSTM model cannot specifically distinguish which category the current input data belongs to), and executes S33 to generate samples for the third data that is less than a preset threshold, until all datasets have been traversed.

[0104] S33 generates samples, trains the GANs model using historical reliability parameters and category labels to obtain the trained GANs model, generates training samples corresponding to the third data using the trained GANs model, and uses the training samples to train the trained BI-LSTM model again to obtain the retrained BI-LSTM model (the retrained BI-LSTM model strengthens the training on unrecognizable input data and increases the discrimination ability). The training samples include new reliability parameters and new category labels.

[0105] In S13 prediction, real-time reliability parameters are input into the retrained BI-LSTM model.

[0106] In other embodiments, the method further includes:

[0107] S34 obtains a vector, concatenates the vectors into a feature matrix, calculates the maximum eigenvalue of the feature matrix and the corresponding eigenvector, and obtains the vector corresponding to the maximum value of the elements in the feature vector, denoted as the first vector.

[0108] S35 If the judgment is correct, determine whether the historical reliability parameter and category label corresponding to the first vector are correct. If so, use the position of the maximum value in the feature vector as the index, and divide the evaluation matrix according to the index to obtain two sub-matrices (for example, if the maximum value is in the 5th position in the feature vector, then divide it into two sub-matrices in the 5th row or 5th column of the evaluation matrix, and the 5th row or 5th column can be divided in any one of the sub-matrices), and execute S36 Vector Filtering; if not, reset the category label of the historical reliability parameter corresponding to the first vector, retrain the BI-LSTM model, and execute S31 to construct the evaluation matrix.

[0109] S36 Vector Filtering: Based on the vectors in the two sub-matrices, obtain the loss value of the trained BI-LSTM model (i.e. calculate the loss value corresponding to the element in each of the two sub-matrices), retain the vectors in the sub-matrices with smaller loss values ​​(smaller loss values ​​indicate better training performance of the sub-matrices and better classification ability of the BI-LSTM model), delete the vectors in the sub-matrices with larger loss values, and execute S32 to obtain the dataset.

[0110] Example 4: The difference between this example and Example 1 is that the method further includes: using the PCA algorithm to reduce the dimensionality of the historical reliability parameters and the real-time reliability parameters, and using the dimensionality-reduced historical reliability parameters and the real-time reliability parameters as new historical reliability parameters and new real-time reliability parameters.

[0111] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A reliability prediction method for fuel injector solenoid valves based on multiphysics coupling, characterized in that, include: Data acquisition: Collect real-time operating data of the solenoid valve and multiple sets of historical operating data of the same type of solenoid valve to construct a multi-physics coupling model of the solenoid valve. The operating data includes electromagnetic force data, temperature data, fluid pressure data, vibration data, operating voltage and operating current. Extracting reliability parameters: Input each set of historical operating data into the multiphysics coupling model in sequence, and solve the multiphysics coupling model using the finite element analysis method to obtain the multiphysics distribution characteristics of the solenoid valve. Extract historical reliability parameters from the multiphysics distribution characteristics and add category labels to the historical reliability parameters. The historical reliability parameters include: temperature gradient of the solenoid coil, stress concentration factor of the valve core, and pressure pulse amplitude of fluid flow. Prediction: Input real-time running data into the multiphysics coupling model to obtain real-time reliability parameters, construct a BI-LSTM model using machine learning algorithms, train the BI-LSTM model using historical reliability parameters and category labels to obtain the trained BI-LSTM model, input the real-time reliability parameters into the trained BI-LSTM model, and output the prediction results. Simulation: Construct a digital twin model, embed a multiphysics coupling model into the digital twin model, and input historical operating data into the digital twin model in chronological order of acquisition time, so that the digital twin model can simulate the degradation process of the solenoid valve and obtain the simulated degradation data of the solenoid valve at different times. Simulation judgment: Obtain the real degradation data at the same time as the simulation degradation data, calculate the difference between the simulation degradation data and the real degradation data, and record it as the first data. Determine whether the first data is greater than the preset difference threshold. If so, the digital twin model is corrected, and the simulation degradation data is re-obtained based on the corrected digital twin model. If not, the step of calculating the degree of degradation is executed. Calculate the degree of degradation: Obtain the running data for a preset period of time before the current moment, and record it as the second data. Input the second data and the real-time running data into the digital twin model in the order of collection time to obtain the predicted degradation data. Calculate the predicted degree of degradation based on the predicted degradation data. Degradation judgment: Determine whether the predicted degradation degree is greater than the preset degradation degree threshold. If so, issue a degradation alarm signal; if not, repeat this step after a preset interval. Model Update: Increase the mesh density of the target region in the digital twin model to obtain a new digital twin model. Extract a new multiphysics coupling model from the new digital twin model and perform the step of extracting reliability parameters. The target region is a critical part of the solenoid valve that is prone to degradation or has a significant impact on performance.

2. The reliability prediction method for fuel injector solenoid valves based on multiphysics coupling according to claim 1, characterized in that, The method further includes: Establish a degradation model: Based on the simulated degradation data, a degradation curve is plotted, and a degradation model is established to fit the degradation curve. The calculation model of the degradation model is as follows: ； in, The degree of degradation at time t; for The degree of degradation over time; The parameters are determined using the nonlinear least squares method. The value of ; Calculate reliability: based on parameters Calculate and output the reliability of the solenoid valve, the reliability... The calculation model is as follows: ； ； Where n is an intermediate variable; This refers to the operating time of the solenoid valve.

3. The reliability prediction method for fuel injector solenoid valves based on multiphysics coupling according to claim 1 or 2, characterized in that, The method further includes: Constructing the evaluation matrix: Obtain the vector output by the BI-LSTM model after each round of training, construct the evaluation matrix based on the vector, and standardize the rows or columns of the evaluation matrix to obtain the processed evaluation matrix. The element in the i-th row and j-th column of the evaluation matrix represents the probability of misclassifying the i-th class as the j-th class. Obtain the dataset: Combine the off-diagonal elements in the rows or columns of the evaluation matrix into a dataset; calculate the difference between any two off-diagonal elements in the dataset, and denote it as the third data. For the third data that is less than a preset threshold, execute the step of generating samples until all datasets have been traversed. Sample generation: The training samples corresponding to the third data are generated using the GANs model, and the trained BI-LSTM model is trained again using the training samples to obtain the retrained BI-LSTM model. The training samples include new reliability parameters and new category labels. In the prediction step, real-time reliability parameters are input into the retrained BI-LSTM model.

4. The reliability prediction method for fuel injector solenoid valves based on multiphysics coupling according to claim 3, characterized in that, The diagonal elements of the evaluation matrix have a value of 0.

5. The reliability prediction method for fuel injector solenoid valves based on multiphysics coupling according to claim 4, characterized in that, The method further includes: Obtain the vector: Concatenate the vectors into a feature matrix, calculate the maximum eigenvalue of the feature matrix and the corresponding eigenvector, and obtain the vector corresponding to the maximum value of the elements in the feature vector, denoted as the first vector; Correct judgment: Determine whether the historical reliability parameter and category label corresponding to the first vector are correct. If yes, divide the evaluation matrix according to the position of the maximum value in the feature vector to obtain two sub-matrices. If no, reset the category label of the historical reliability parameter corresponding to the first vector, retrain the BI-LSTM model, and execute the step of constructing the evaluation matrix. Vector filtering: Based on the vectors in the two sub-matrices, obtain the loss value of the trained BI-LSTM model, retain the vectors in the sub-matrices with smaller loss values, and delete the vectors in the sub-matrices with larger loss values.

6. The reliability prediction method for fuel injector solenoid valves based on multiphysics coupling according to claim 3, characterized in that, The method further includes: training the GANs model using historical reliability parameters and category labels to obtain the trained GANs model; In the sample generation step, the trained GANs model is used to generate training samples corresponding to the third data.

7. The reliability prediction method for fuel injector solenoid valves based on multiphysics coupling according to claim 1, characterized in that, The method further includes: using the PCA algorithm to reduce the dimensionality of historical reliability parameters and real-time reliability parameters, and using the dimensionality-reduced historical reliability parameters and real-time reliability parameters as new historical reliability parameters and new real-time reliability parameters.

Images