Hydraulic system model construction method based on mechanism and data adaptive hybrid drive

By adopting a hybrid drive model of adaptive mechanism and data in the hydraulic system, combined with Kalman filtering and genetic algorithm, the problem of low accuracy and insufficient generalization capability of the hydraulic system's digital twin model is solved, and higher simulation accuracy is achieved and actual system mapping capabilities.

CN119026444APending Publication Date: 2024-11-26CHINESE PEOPLES LIBERATION ARMY UNIT 63729
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410822898.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-06-19
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The digital twin model of hydraulic system has low accuracy and insufficient generalization capabilities, resulting in limited improvement in intelligence level.

Method used

A model construction method based on adaptive hybrid drive of mechanism and data is adopted, and the mechanism simulation model and data-driven model are fused through the Kalman filtering algorithm, and the genetic algorithm is used for iterative updates to realize the adaptive weight adjustment and iterative update of the model.

Benefits of technology

The simulation accuracy and generalization capabilities of the digital twin model of hydraulic system are improved, making it more synchronously mapped actual systems, and enhancing the flexibility and scalability of the model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119026444A_ABST
    Figure CN119026444A_ABST
Patent Text Reader

Abstract

The invention provides a hydraulic system model construction method based on mechanism and data adaptive hybrid drive, which comprises the following steps of: respectively constructing a mechanism simulation model and a data drive model of a hydraulic system based on an operation mechanism and historical operation data of the hydraulic system, and constructing a development environment of a digital twin model of the hydraulic system by applying python; on the basis, effective fusion of the mechanism simulation model and the data driving model is realized by applying a Kalman filtering algorithm, a theoretical derivation value of the mechanism simulation model is corrected through a data driving result, and the Kalman filtering algorithm is improved; the digital twinning model of the hydraulic system can adaptively change the weights of the mechanism model and the data driving model, and the precision and generalization ability of the digital twinning model of the hydraulic system are greatly improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of digital modeling of hydraulic systems, and in particular relates to a method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive. Background Art

[0002] In recent years, with the development of new generation information technologies such as cloud computing and the Internet of Things, the intelligence of mechanical equipment has gradually become a research hotspot. As an important branch of the mechanical field, the research on hydraulic systems is inevitably developing rapidly in the direction of informatization, intelligence, and wisdom. Digital twin technology is an important way to improve the level of intelligence in hydraulic systems. However, the application of digital twin technology in hydraulic systems mainly has the following difficulties. Since hydraulic systems are nonlinear time-varying systems with variable working environments and complex structures, the hydraulic system models built by traditional modeling methods are single in form and lack online real-time interaction with the actual system, resulting in low model simulation accuracy and insufficient generalization ability. Therefore, building a high-fidelity digital twin model with good generalization ability is an important guarantee for the application of digital twin technology in hydraulic systems to improve the level of intelligence. Summary of the invention

[0003] The purpose of the present invention is to provide a hydraulic system model construction method based on mechanism and data adaptive hybrid drive to solve the problems of low accuracy and insufficient generalization ability of existing hydraulic system digital twin models.

[0004] In order to achieve the above-mentioned purpose, the present invention provides a hydraulic system model construction method based on mechanism and data adaptive hybrid drive, which realizes model fusion on the basis of the construction of mechanism simulation model and data-driven model, including: digital twin model development environment construction: using Python language to realize the construction of digital twin model development environment of hydraulic system: model fusion and algorithm improvement: using Kalman filter algorithm to realize the effective fusion of mechanism simulation model and data-driven model, correcting the theoretical derivation value of mechanism simulation model through data-driven results, and improving Kalman filter algorithm so that the digital twin model can adaptively change the weight of mechanism simulation model and data-driven model; model iterative update: selecting genetic algorithm, realizing iterative update of mechanism model by identifying important parameters of mechanism simulation model; updating data-driven model by updating data set, and finally realizing iterative update of digital twin model of hydraulic system.

[0005] The above-mentioned hydraulic system model construction method based on mechanism and data adaptive hybrid drive, wherein the construction of the digital twin model development environment includes: first, calling the API program interface of the multi-domain modeling and simulation platform through python to realize the development programming of the mechanism simulation model, and then using the python language to build a data-driven model and realize subsequent model fusion work.

[0006] The above-mentioned hydraulic system model construction method based on mechanism and data adaptive hybrid drive, wherein the effective fusion of the mechanism simulation model and the data driven model includes the following steps:

[0007] A priori estimate:

[0008] Y c,t =Y m,t (2)

[0009] A priori error covariance:

[0010]

[0011] Kalman gain:

[0012]

[0013] Digital twin model output:

[0014] Y f,t =Y c,t +K t (Y d,t -Y c,t ) (5)

[0015] Update the error covariance:

[0016]

[0017] Y m,t is the mechanism simulation model; Y d,t Y is a data-driven model; f,t For digital twin models; is the prior error covariance; P t is the initial value of the prior error covariance; W t is the process noise variance; V t is the observation noise variance.

[0018] In the above-mentioned method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive, the improvement of the Kalman filter algorithm includes: using a moving window technique to estimate the process noise variance and the observation noise variance online while performing recursive estimation, and correcting the Kalman gain;

[0019] Assuming the window length is n, the specific calculation formula for updating the process noise variance and the observation noise variance at time t is:

[0020] Process noise variance:

[0021]

[0022] Observation noise variance:

[0023]

[0024] The above-mentioned hydraulic system model construction method based on mechanism and data adaptive hybrid drive, in which, when the training data set samples are sufficient, the data-driven model has higher accuracy, smaller observation noise variance, larger Kalman gain, larger proportion weight of the data-driven model, and the digital twin model is closer to the data-driven model; when the training data set samples are insufficient, the data-driven model has lower accuracy, larger observation noise variance, smaller Kalman gain, smaller proportion weight of the data-driven model, smaller proportion weight of the mechanism simulation model, and larger proportion weight of the mechanism simulation model, and the digital twin model is closer to the mechanism simulation model.

[0025] The above-mentioned hydraulic system model construction method based on mechanism and data adaptive hybrid drive, wherein a multi-domain modeling and simulation platform is used to build a mechanism simulation model of the hydraulic system, and the genetic algorithm in the optimization toolbox Design Exploration that comes with the multi-domain modeling and simulation platform is used to realize the iterative update of the mechanism simulation model.

[0026] The above-mentioned method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive, wherein the iterative update of the mechanism simulation model includes: first, performing a sensitivity analysis on the micro hydraulic system test bench to determine the important parameters affecting the hydraulic system: such as load, reversing valve wear clearance, oil viscosity, etc.; secondly, taking the error between the calculation result of the mechanism simulation model and the measurement parameters of the actual measurement and control system sensor as the fitness function of the genetic algorithm, and through iteration, making the fitness function converge to the minimum value, thereby obtaining the optimal parameter combination, and realizing the iterative update of the mechanism simulation model.

[0027] Compared with the prior art, the beneficial technical effects of the present invention are:

[0028] (1) The Python language is used to build the mechanism simulation model and data-driven model development environment. The programming language can be used to realize the secondary development of the model, which enhances the flexibility and scalability of the model.

[0029] (2) The improved Kalman filter algorithm is used to realize the adaptive hybrid drive of the mechanism model and the data model. Compared with the traditional modeling method, this method model improves the simulation accuracy and generalization ability of the model;

[0030] (3) Through the dual update mode of mechanism model and data model, the iterative update of the hybrid model is realized, ensuring the synchronous mapping between the model and the actual system. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive of the present invention is given by the following embodiments and drawings.

[0032] Figure 1 This is a flow chart of a method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive according to an embodiment of the present invention.

[0033] Figure 2 This is a flowchart of iterative update of the digital twin model in an embodiment of the present invention.

[0034] Figure 3 This is a flowchart of iterative updating of the mechanism simulation model in an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined Figure 1 to Figure 3 The hydraulic system model construction method based on mechanism and data adaptive hybrid drive of the present invention is further described in detail.

[0036] The hydraulic system model construction method based on mechanism and data adaptive hybrid drive of the present invention constructs a mechanism simulation model and a data-driven model of the hydraulic system based on the hydraulic system operation mechanism and historical operation data, respectively, uses Python to build a development environment for the digital twin model of the hydraulic system, and on this basis uses the Kalman filter algorithm to realize the effective fusion of the mechanism simulation model and the data-driven model, corrects the theoretical derivation value of the mechanism simulation model through the data-driven result, and improves the Kalman filter algorithm, so that the digital twin model of the hydraulic system can adaptively change the weights of the mechanism model and the data-driven model, thereby greatly improving the accuracy and generalization ability of the digital twin model of the hydraulic system.

[0037] The method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive of the present invention includes:

[0038] 1) Construction of mechanism simulation model

[0039] According to the working principle of the hydraulic system, a mechanism simulation model of the hydraulic system is built on the simulation platform;

[0040] The simulation platform is, for example, a multi-domain modeling and simulation platform, MWORKS.Sysplorer, etc.;

[0041] 2) Data-driven model building

[0042] Combined with the actual hydraulic system historical operation data, the random forest algorithm is used to train the data set, and a hydraulic system historical data regression model is constructed as a data-driven model of the hydraulic system to achieve real-time prediction of the actual hydraulic system;

[0043] 3) Construction of digital twin model development environment

[0044] Use Python to build a digital twin model development environment for hydraulic systems: First, use Python to call the API program interface of the multi-domain modeling and simulation platform to develop and program the mechanism simulation model. Then, use Python to build a data-driven model and implement subsequent model fusion work.

[0045] 4) Model fusion and algorithm improvement

[0046] The Kalman filter algorithm is used to realize the effective integration of the mechanism simulation model and the data-driven model. The theoretical derivation value of the mechanism simulation model is corrected through the data-driven results, and the Kalman filter algorithm is improved so that the digital twin model can adaptively change the weights of the mechanism model and the data-driven model.

[0047] 5) Model iterative update

[0048] As the hydraulic system runs for a long time, the performance of some of its components will degrade. To ensure that the digital twin model of the hydraulic system can achieve real-time mapping of the actual hydraulic system, the digital twin model needs to be iteratively updated accordingly.

[0049] Specifically, the present invention selects a genetic algorithm in the optimization toolbox Design Exploration that comes with the multi-domain modeling and simulation platform, and realizes the iterative update of the mechanism simulation model by identifying important parameters of the mechanism simulation model; updates the data-driven model by updating the data set (i.e., the actual hydraulic system historical operation data set), and finally realizes the iterative update of the digital twin model of the hydraulic system.

[0050] The method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive of the present invention is now described in detail with specific embodiments.

[0051] Figure 1 Shown is a flow chart of a method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive according to an embodiment of the present invention.

[0052] See also Figure 1 The method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive of this embodiment includes:

[0053] 1) Construction of mechanism simulation model

[0054] First, the viscosity-temperature characteristic of the oil is modeled. During the operation of the hydraulic system, the ambient temperature may change, thereby changing the viscosity of the oil and affecting the operating state of the hydraulic system. Therefore, it is necessary to model the viscosity-temperature characteristic of the hydraulic system to ensure the accuracy of the mechanism model. This embodiment uses No. 46 hydraulic oil. By fitting the viscosity of the hydraulic oil at different temperatures with a polynomial, the viscosity-temperature characteristic curve of No. 46 hydraulic oil is obtained. According to the viscosity-temperature characteristic curve, the viscosity-temperature model of the hydraulic system can be constructed as shown in formula (1):

[0055]

[0056] Where, μ is the viscosity coefficient of the hydraulic oil, and x is the temperature;

[0057] Secondly, according to the working principle of the hydraulic system, a mechanism simulation model of the hydraulic system is built on the multi-domain modeling and simulation platform. The main components of the hydraulic system include: hydraulic pump, pilot relief valve, three-position five-way solenoid reversing valve, speed control valve, hydraulic cylinder, etc.

[0058] 2) Data-driven model building

[0059] Combined with the actual hydraulic system historical operation data, the random forest algorithm is used to train the data set, and a hydraulic system historical data regression model is constructed as a data-driven model of the hydraulic system to achieve real-time prediction of the actual hydraulic system;

[0060] 3) Construction of digital twin model development environment

[0061] In this embodiment, since the digital twin model of the hydraulic system is mainly composed of a mechanism simulation model and a data-driven model, and the mechanism simulation model cannot be run independently from the simulation platform, it is necessary to build a development environment to realize the secondary development of the two models. In this embodiment, Python calls the API program interface of the multi-domain modeling and simulation platform to realize the development programming of the mechanism simulation model, and then uses Python language to build a data-driven model and realize the subsequent model fusion work;

[0062] 4) Model fusion and algorithm improvement

[0063] This embodiment uses the Kalman filter algorithm to achieve effective fusion of the mechanism simulation model and the data-driven model. The fusion steps are as follows:

[0064] A priori estimate:

[0065] Y c,t =Y m,t (2)

[0066] A priori error covariance:

[0067]

[0068] Kalman gain:

[0069]

[0070] Digital twin model output:

[0071] Y f,t =Y c,t +K t (Y d,t -Y c,t ) (5)

[0072] Update the error covariance:

[0073]

[0074] Y m,t is the mechanism simulation model (time t); Y d,t is the data-driven model (at time t); Y f,t is the digital twin model (at time t); is the prior error covariance (at time t); P t is the initial value of the prior error covariance; W t is the process noise variance (at time t); V t is the observation noise variance (at time t);

[0075] From the above, we can see that the mechanism simulation model is used as the state value of the Kalman filter algorithm, and the data-driven model is used as the observation value of the Kalman filter algorithm. t The weighted fusion of the two models is realized, and the theoretical derivation value of the mechanism simulation model is corrected by using the data-driven results. The present embodiment then makes the following improvements to the Kalman filter algorithm: the moving window technology is used to estimate the process noise variance and the observation noise variance online while performing recursive estimation, and the Kalman gain is corrected. Assuming the window length is n, the specific calculation formula for updating the process noise variance and the observation noise variance at time t is:

[0076] Process noise variance:

[0077]

[0078] Observation noise variance:

[0079]

[0080] Where Y s,tis the actual sensor data (at time t). From formulas (3) and (4), we can see that the Kalman gain is mainly related to the process noise variance and the observation noise variance. Therefore, the variance between the data-driven model and the actual sensor data is taken as the observation noise variance, and the variance between the mechanism simulation model and the actual sensor data is taken as the process noise variance. When the training data set has sufficient samples, the data-driven model has high accuracy, its observation noise variance is small, the Kalman gain increases, the weight of the data-driven model increases, and the digital twin model is closer to the data-driven model. When the training data set has insufficient samples, the data-driven model has low accuracy, its observation noise variance is large, the Kalman gain decreases, the weight of the data-driven model decreases, the weight of the mechanism simulation model increases, and the digital twin model is closer to the mechanism simulation model. Therefore, the digital twin model can adaptively change the weights of the mechanism simulation model and the data-driven model, which greatly improves the accuracy and generalization ability of the digital twin model.

[0081] 5) Model iterative update

[0082] As the hydraulic system runs for a long time, the performance of some of its components will degrade. In order to ensure that the digital twin model of the hydraulic system can achieve real-time mapping of the actual hydraulic system, the digital twin model needs to be iteratively updated accordingly; Figure 2 In this embodiment, a genetic algorithm is selected in the optimization toolbox Design Exploration that comes with the multi-domain modeling and simulation platform. The important parameters of the mechanism simulation model are identified to realize the iterative update of the mechanism model. The data-driven model is updated by updating the data set, and finally the iterative update of the digital twin model is realized.

[0083] Among them, the specific steps of iterative update of the mechanism simulation model are as follows: Figure 3 As shown, including:

[0084] Firstly, a sensitivity analysis is conducted on the micro hydraulic system test bench to determine three important parameters that affect the hydraulic system: load, reversing valve wear clearance, and viscosity.

[0085] Secondly, the error between the calculation results of the mechanism simulation model and the measurement parameters of the actual measurement and control system sensors is used as the fitness function of the genetic algorithm. The fitness function is converged to the minimum value through iteration, so as to obtain the optimal parameter combination and realize the iterative update of the mechanism simulation model.

[0086] The above is only a specific embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other application objects and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art do not deviate from the spirit and scope of the present invention, and should be within the scope of protection of the claims attached to the present invention.

Claims

1. A hydraulic system model construction method based on mechanism and data adaptive hybrid drive realizes model fusion on the basis of mechanism simulation model and data driven model construction, which is characterized by: include: Construction of digital twin model development environment: Use Python language to build the digital twin model development environment of hydraulic system: Model fusion and algorithm improvement: The Kalman filter algorithm is used to achieve effective fusion of the mechanism simulation model and the data-driven model. The theoretical derivation value of the mechanism simulation model is corrected through the data-driven results, and the Kalman filter algorithm is improved so that the digital twin model can adaptively change the weights of the mechanism simulation model and the data-driven model. Model iterative update: Select a genetic algorithm to identify important parameters of the mechanism simulation model to achieve iterative update of the mechanism model; update the data-driven model by updating the data set to ultimately achieve iterative update of the digital twin model of the hydraulic system.

2. The method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive according to claim 1, characterized in that: The construction of the digital twin model development environment includes: first, calling the API program interface of the multi-domain modeling and simulation platform through Python to realize the development programming of the mechanism simulation model, and then using the Python language to build a data-driven model and realize subsequent model fusion work.

3. The method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive according to claim 1, characterized in that: The effective integration of mechanism simulation model and data-driven model includes the following steps: A priori estimate: AND c,t =And m,t (2) A priori error covariance: P t - =P t-1 +W t-1 (3) Kalman gain: K t =P t - (P t - +V t-1 ) -1 (4) Digital twin model output: AND f,t =And c,t +K t (AND d,t -AND c,t ) (5) Update the error covariance: P t =(1-K t )P t - (6) Y m,t is the mechanism simulation model; Y d,t Y is a data-driven model; f,t is the digital twin model; P t - is the prior error covariance; P t is the initial value of the prior error covariance; W t is the process noise variance; V t is the observation noise variance.

4. The method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive according to claim 3, characterized in that: The improvements of Kalman filter algorithm include: using moving window technology to estimate process noise variance and observation noise variance online while performing recursive estimation, and correcting Kalman gain; Assuming the window length is n, the specific calculation formula for updating the process noise variance and the observation noise variance at time t is: Process noise variance: Observation noise variance:

5. The method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive according to claim 4, characterized in that: When the training data set has sufficient samples, the data-driven model has higher accuracy, smaller observation noise variance, larger Kalman gain, and greater weight of the data-driven model, and the digital twin model is closer to the data-driven model; when the training data set has insufficient samples, the data-driven model has lower accuracy, larger observation noise variance, smaller Kalman gain, and smaller weight of the data-driven model, and greater weight of the mechanism simulation model, and the digital twin model is closer to the mechanism simulation model.

6. The method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive according to claim 1, characterized in that: A multi-domain modeling and simulation platform is used to build a mechanism simulation model of the hydraulic system, and the genetic algorithm in the optimization toolbox Design Exploration that comes with the multi-domain modeling and simulation platform is used to realize the iterative update of the mechanism simulation model.

7. The method for constructing a hydraulic system model based on mechanism and data adaptive hybrid drive according to claim 1, characterized in that: Iterative updates of the mechanism simulation model include: Firstly, a sensitivity analysis is conducted on the actual hydraulic system test bench to determine the three important parameters that affect the hydraulic system: load, reversing valve wear clearance, and viscosity; Secondly, the error between the calculation results of the mechanism simulation model and the measurement parameters of the actual measurement and control system sensors is used as the fitness function of the genetic algorithm. The fitness function is converged to the minimum value through iteration, so as to obtain the optimal parameter combination and realize the iterative update of the mechanism simulation model.

Citation Information

Cited By

  • Complex system simulation resolving method based on mixed gradient search

    CN119849144A