A real-time online monitoring method and device for electrostatic hydraulic actuator

Abstract

The present application provides a method and device for real-time online monitoring of an electrostatic hydraulic actuator, which belongs to the field of aviation electromechanical. The method provided in the present application includes: obtaining normal performance data of each working condition, and establishing a nonlinear real-time mapping model based on the normal performance data and a neural network, the nonlinear real-time mapping model is used to receive control instructions and predict the output of the actuator; training an SVDD abnormality detection model according to the predicted output and the output of the actuator after receiving the control instruction; and monitoring the state of the electrostatic hydraulic actuator in real time based on the trained SVDD abnormality detection model. The method and device for real-time online monitoring of an electrostatic hydraulic actuator provided in the present application can establish a detection model based on data under normal working conditions, so as to realize accurate identification of fault conditions under various working conditions.

Description

Technical Field

[0001] The present application relates to the field of aviation electromechanical technology, and in particular to a real-time online monitoring method and device for an electrostatic hydraulic actuator. Background Art

[0002] With the development of aviation multi-electric technology, electric drive products are more widely used. When the controller components in the system are abnormal, the system may not be able to accurately track the control instructions, resulting in control function degradation, abnormal operation and other problems, which seriously affect flight safety. The electric drive control system has coupling and nonlinear characteristics in various links. In engineering practice, linearizing the components in complex systems will ignore the impact of nonlinear factors in the system on fault performance, making it difficult to establish an accurate mathematical theoretical model, which reduces the accuracy of state recognition. In addition, the actual system operation process is mostly normal data, and there is a lack of fault data samples. Fault identification is mainly based on empirical thresholds and BIT (Built-In Test), which is heavily dependent on the experience of professionals, has low accuracy, and is prone to false alarms and omissions. Summary of the invention

[0003] In view of this, the present application provides a real-time online monitoring method and device for an electrostatic hydraulic actuator, which can establish a detection model based on data under normal working conditions to achieve accurate identification of fault conditions.

[0004] Specifically, the present application is implemented through the following technical solutions:

[0005] The first aspect of the present application provides a real-time online monitoring method for an electrostatic hydraulic actuator, the method comprising:

[0006] Acquire normal performance data of each working condition, and establish a nonlinear real-time mapping model based on the normal performance data and a neural network, wherein the nonlinear real-time mapping model is used to receive control instructions and predict the output of an actuator;

[0007] The SVDD anomaly detection model is trained based on the predicted output and the output of the actuator after receiving the control command;

[0008] The SVDD anomaly detection model is trained with the error value between the normal working state data of the electrostatic hydraulic actuator and the predicted output, wherein the error value includes the error value under different working conditions, and the SVDD parameters are adjusted based on the error values ​​under different working conditions, so as to make the SVDD parameters adaptively adjusted under different feature dimensions according to the sensitive correlation between the feature and the fault;

[0009] The state of the electrostatic hydraulic actuator is monitored in real time based on the trained SVDD anomaly detection model.

[0010] A second aspect of the present application provides a real-time online monitoring device for an electrostatic hydraulic actuator, the device comprising a building module, a training module and a monitoring module;

[0011] The establishment module is used to obtain normal performance data of each working condition, and establish a nonlinear real-time mapping model based on the normal performance data and the neural network, and the nonlinear real-time mapping model is used to receive control instructions and predict the output of the actuator;

[0012] The training module is used to train the SVDD anomaly detection model according to the predicted output and the output of the actuator after receiving the control instruction;

[0013] The SVDD anomaly detection model is trained with the error value between the normal working state data of the electrostatic hydraulic actuator and the predicted output, wherein the error value includes the error value under different working conditions, and the SVDD parameters are adjusted based on the error values ​​under different working conditions, so as to make the SVDD parameters adaptively adjusted under different feature dimensions according to the sensitive correlation between the feature and the fault;

[0014] The monitoring module is used to monitor the state of the electrostatic hydraulic actuator in real time based on the trained SVDD anomaly detection model.

[0015] The real-time online monitoring method and device of the electrostatic hydraulic actuator provided by the present application obtains the normal performance data of each working condition and establishes a nonlinear real-time mapping model in combination with the nonlinear fitting ability of the neural network, solves the problem of distortion of the traditional simplified model in the nonlinear section, can accurately predict the output of the actuator, and by combining the SVDD abnormal detection model, can more accurately detect minor abnormalities and faults, achieve real-time monitoring and rapid response, and reduce the probability of false alarms and missed alarms. In addition, in the face of the characteristics of the electrostatic hydraulic actuator under different working conditions, the simulation model that changes with the working condition is corrected by the difference in the working condition, avoiding repeated modeling and improving the accuracy of the simulation model prediction output. It should also be noted that the abnormal detection model can be established based only on the data under the normal working state data, that is, based on the zero-fault sample, which solves the problem of lack of fault data and changeable working conditions in engineering applications, without relying on professional experience and fault tests, with good flexibility, low cost, high fault coverage, and can make up for the shortcomings of traditional BIT technology. In addition, the error between the model prediction amount and the actual output amount is used as the input of the input abnormal detection model, and the abnormal detection model is trained by single performance data to improve the fault identification ability, as well as the overall performance and reliability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flow chart of a first embodiment of a real-time online monitoring method for an electrostatic hydraulic actuator provided by the present application;

[0017] Figure 2A simulation model block diagram of an electrostatic hydraulic actuator shown as an exemplary embodiment of the present application;

[0018] Figure 3 A schematic diagram of a nonlinear real-time mapping model prediction output shown in an exemplary embodiment of the present application;

[0019] Figure 4 A schematic diagram of a nonlinear real-time mapping model prediction output shown in another exemplary embodiment of the present application;

[0020] Figure 5 for Figure 3 Calculation result diagram after error calculation;

[0021] Figure 6 for Figure 4 Calculation result diagram after error calculation;

[0022] Fig. 7A A schematic diagram of the original space in the training SVDD anomaly detection model shown in an exemplary embodiment of the present application;

[0023] Figure 7B A schematic diagram of a training SVDD anomaly detection model according to an exemplary embodiment of the present application;

[0024] Figure 7C For this purpose Figure 7B Schematic diagram of adaptive adjustment;

[0025] Fig. 8A The fault data obtained for the electrostatic hydraulic actuator shown in an exemplary embodiment of the present application when the controller is disconnected in a certain working condition;

[0026] Figure 8B Fault data obtained when a controller of an electrostatic hydraulic actuator under a certain working condition is short-circuited, as shown in an exemplary embodiment of the present application;

[0027] Figure 8C Fault data obtained when an IGBT of a controller fails in a certain working condition of an electrostatic hydraulic actuator shown in an exemplary embodiment of the present application;

[0028] Fig.9A A schematic diagram of predicted output and actual output of direct-axis voltage according to an exemplary embodiment of the present application;

[0029] Fig. 9B A schematic diagram of the error value between the predicted output and the actual output of the direct-axis voltage according to an exemplary embodiment of the present application;

[0030] Fig. 10A A schematic diagram of predicted output and actual output of quadrature-axis voltage according to an exemplary embodiment of the present application;

[0031] Fig. 10B A schematic diagram of the error value between the predicted output and the actual output of the quadrature-axis voltage shown in an exemplary embodiment of the present application;

[0032] Fig.11 A schematic diagram showing a fault detected by an abnormality detection model according to an exemplary embodiment of the present application;

[0033] Fig.12 A schematic diagram showing a fault detected by an abnormality detection model shown in another exemplary embodiment of the present application;

[0034] Fig.13 This is a structural schematic diagram of Example 1 of the real-time online monitoring device for an electrostatic hydraulic actuator provided in the present application. DETAILED DESCRIPTION

[0035] Here, exemplary embodiments are described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application.

[0036] The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in this article refers to and includes any or all possible combinations of one or more associated listed items.

[0037] It should be understood that although the terms first, second, third, etc. may be used in the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0038] Specific embodiments are given below to introduce the technical solution of the present application in detail.

[0039] Figure 1 This is a flow chart of the first embodiment of the real-time online monitoring method for electrostatic hydraulic actuators provided in this application. Figure 1 , the method provided in this embodiment may include:

[0040] S101, obtaining normal performance data of each working condition, and establishing a nonlinear real-time mapping model based on the normal performance data and a neural network, wherein the nonlinear real-time mapping model is used to receive control instructions and predict outputs of actuators.

[0041] It should be noted that obtaining simulation data includes:

[0042] (1) According to the characteristics and working requirements of the electrostatic-hydraulic actuator, a simulation model of the electrostatic-hydraulic actuator and an electrostatic-hydraulic actuator test bench are established respectively.

[0043] It should be noted that the electrostatic hydraulic actuator is a device that converts electrical energy into hydraulic energy and drives the actuator through hydraulics. Its working principle combines the advantages of electric motors and hydraulic systems. It is usually used in application scenarios requiring high precision and high force, such as aerospace, robotics, and precision manufacturing.

[0044] Specifically, in this process, based on the structure and working principle of the electrostatic hydraulic actuator, its mathematical model (electrostatic hydraulic actuator simulation model) is established. The model should include the physical characteristics and dynamic behaviors of components such as motors, hydraulic pumps, hydraulic cylinders, controllers and related sensors. In specific implementation, MATLAB / Simulink or other simulation tools can be used to build a simulation block diagram of the electrostatic hydraulic actuator, and the parameters of each component can be configured in the simulation software to ensure that the model accurately reflects the behavior of the actual electrostatic hydraulic actuator.

[0045] It should also be noted that the test bench is established to obtain real and accurate data. Through the test bench, the actual working performance and stability of the electrostatic hydraulic actuator can be tested.

[0046] (2) Inputting commands under different working conditions into the simulation model and the electrostatic hydraulic actuator test bench to obtain simulation data and test data respectively; the simulation data and test data both include control commands and response outputs.

[0047] Figure 2 This is a simulation model block diagram of an electrostatic hydraulic actuator shown in an exemplary embodiment of the present application, please refer to Figure 2First, different types of input instructions are set, including position signals, speed signals, loads, etc. with different waveforms such as sine wave, square wave, triangle wave, dynamic frequency sweep, etc. Each input instruction represents a working condition, covering various working conditions that the electrostatic hydraulic actuator may encounter in actual operation. Then, the instructions under different working conditions are input into the simulation model and the test bench, and the simulation model and the test bench are run to simulate the response of the electrostatic hydraulic actuator under these conditions, and the control instructions and response output data generated during the simulation and test processes are obtained. Specifically, the control instructions may include direct-axis current instructions (Idcmd) and quadrature-axis current instructions (Iqcmd); the response output may include actual direct-axis current (Id), actual quadrature-axis current (Iq), direct-axis voltage (Ud), quadrature-axis voltage (Uq), and speed (speed).

[0048] As an optional embodiment, the electrostatic hydraulic actuator has different characteristics under different working conditions, therefore, the acquisition of simulation data also includes: establishing an electrostatic hydraulic actuator simulation model based on the first working condition of the electrostatic hydraulic actuator; determining the corresponding second working condition according to the control instruction of the electrostatic hydraulic actuator; calculating a correction coefficient according to the difference between the second working condition and the first working condition; correcting the electrostatic hydraulic actuator simulation model based on the correction coefficient; and calculating simulation data corresponding to the control instruction based on the corrected electrostatic hydraulic actuator simulation model. The method provided by the present invention, in view of the characteristics of the electrostatic hydraulic actuator under different working conditions, corrects the simulation model that changes with the working conditions through the difference in working conditions, thereby avoiding repeated modeling and improving the accuracy of the simulation model prediction output.

[0049] (3) The simulation data and the test data are comprehensively complemented to serve as normal performance data for each operating condition.

[0050] It should be noted that in order to build a nonlinear real-time mapping model, data is needed as support, and its sources include simulation and experiments. Among them, simulation can test more working conditions and save test time, and experiments can obtain real and accurate data. The two can complement each other.

[0051] It should be noted that after obtaining the normal performance data, a nonlinear real-time mapping model is established based on the normal performance data of each working condition and the neural network, including: based on the normal performance data of each working condition, the data under different working conditions are subjected to feature extraction and fusion; the neural network is trained one by one according to the working condition using the feature fusion data, and the network parameters are fine-tuned using the last training result in each training process, and the neural network is used to describe the nonlinear mapping relationship between the control command and the response output. In this process, the data obtained under different working conditions are subjected to feature fusion to form a training set, wherein the purpose of feature fusion is to enable the neural network to learn the nonlinear relationship in the data, and at this time, the BP neural network can be selected for training. Specifically, the neural network can be trained using the successive method, by gradually increasing the amount of training data or gradually increasing the complexity of the network, so as to observe its effect. In addition, the hybrid method can also be used for training.

[0052] Optionally, after obtaining the trained neural network, a pruning method can be used to reduce redundant neurons and connections in the neural network to improve the real-time performance of the model.

[0053] S102: training an SVDD anomaly detection model according to the predicted output and the output of the actuator after receiving the control instruction.

[0054] As an optional embodiment, the change of error value under different working conditions is not the same. If the SVDD anomaly detection model is trained only according to the error value and the anomaly is predicted based on it, it is easy to cause failure in the detection of anomalies. For fault detection, the accuracy requirement is high, and false alarms of anomalies are not allowed for anomaly detection. Therefore, in order to improve the accuracy of the SVDD anomaly detection model, the SVDD anomaly detection model is trained according to the predicted output and the output of the actuator after receiving the control instruction, and also includes: calculating the prediction error according to the output of the actuator after receiving the control instruction and the predicted output; calculating the mapping relationship between the error change value and the working condition corresponding to the prediction error and the control instruction; using the mapping relationship as prompt information; training the SVDD anomaly detection model based on the training sample and the prompt information.

[0055] Among them, the SVDD abnormality detection model is trained with the normal working state data of the electrostatic hydraulic actuator and the error value of the predicted output, and the error value includes the error value under different working conditions. The SVDD parameters are adjusted based on the error values ​​under different working conditions, and the SVDD parameters are adaptively adjusted under different feature dimensions according to the sensitive correlation between the features and the faults.

[0056] It should be noted that before training the SVDD anomaly detection model based on the predicted output and the output of the actuator after receiving the control instruction, it includes: calculating the amplitude error and phase error of the predicted output and the output of the physical actuator after receiving the control instruction; determining the error value based on the error calculation result; and constructing a training sample based on the error value.

[0057] Figure 3 This is a schematic diagram of a nonlinear real-time mapping model prediction output shown in an exemplary embodiment of the present application. Figure 4 This is a schematic diagram of a nonlinear real-time mapping model prediction output according to another exemplary embodiment of the present application. Figure 3 and Figure 4 , the predicted output of the model can be observed, and the error calculation can be performed based on the relationship between the output of the electrostatic hydraulic actuator under normal working conditions.

[0058] Figure 5 for Figure 3 The calculation result diagram after error calculation is shown below. Figure 6 for Figure 4 Please also refer to the calculation result diagram after error calculation. Figure 5 and Figure 6 , the normal working state data of the electrostatic hydraulic actuator under various working conditions are compared with the predicted output to obtain the error value, and the training samples are constructed based on the error value to train the SVDD (support vector data description) anomaly detection model.

[0059] Specifically, the SVDD anomaly detection model is trained according to the predicted output and the output of the actuator after receiving the control instruction, including: calculating the deviation between the predicted output of each working condition and the output of the actuator after receiving the control instruction according to the error value; adjusting the parameters of the SVDD model based on the deviation, and adaptively adjusting it under different feature dimensions according to the sensitive correlation between each feature and the fault; wherein the SVDD model is trained based on normal working state data; and the SVDD model under a single sample is trained based on the adjusted parameters and the error value under normal working state to obtain the SVDD anomaly detection model.

[0060] It should be noted that the deviation can be regarded as a measure of the error value between the predicted output and the actual output, such as the mean square error (MSE).

[0061] Fig. 7A A schematic diagram of the original space in the training SVDD anomaly detection model shown in an exemplary embodiment of the present application; Figure 7B A schematic diagram of a training SVDD anomaly detection model according to an exemplary embodiment of the present application; Figure 7C For this purpose Figure 7BSchematic diagram of adaptive adjustment. Please also refer to Fig. 7A , Figure 7B and Figure 7C , by mapping the sample points in the original space to the high-dimensional feature space, a hypersphere is found in this space. This sphere should be as small as possible and contain normal sample points as much as possible in the hypersphere, while excluding abnormal sample points or exception points outside the hypersphere. Considering that the prediction error is inconsistent under different working conditions, it will make it impossible to identify small faults. The model parameters are optimized by changing different kernel functions and coefficients, and targeted parameter adjustments are made according to the sensitive direction of the features.

[0062] It should be noted that, in actual use, by designing the injection method of the electrostatic hydraulic actuator under different fault modes such as controller open circuit, short circuit, IGBT failure, etc., the fault data is obtained, and then the error value between the model predicted output and the actual output is calculated.

[0063] Fig. 8A The fault data obtained for the electrostatic hydraulic actuator shown in an exemplary embodiment of the present application when the controller is disconnected in a certain working condition; Figure 8B Fault data obtained when a controller of an electrostatic hydraulic actuator under a certain working condition is short-circuited, as shown in an exemplary embodiment of the present application; Figure 8C This is the fault data obtained by the electrostatic hydraulic actuator in a certain working condition when the IGBT controller fails, as shown in an exemplary embodiment of the present application. Fig. 8A , Figure 8B and Figure 8C , the specific fault data of the electrostatic hydraulic actuator under different fault modes such as open circuit, short circuit, IGBT failure, etc. can be obtained.

[0064] Fig.9A A schematic diagram of the predicted output and the actual output of the direct axis voltage shown in an exemplary embodiment of the present application, according to Fig.9A The error between the predicted output and the actual output of the direct axis voltage can be obtained as follows: Fig. 9B shown. Fig. 10A A schematic diagram of the predicted output and the actual output of the quadrature-axis voltage shown in an exemplary embodiment of the present application, according to Fig. 10A The error between the predicted output and the actual output of the quadrature axis voltage can be obtained as Fig. 10B shown.

[0065] S103 . Monitoring the state of the electrostatic hydraulic actuator in real time based on the trained SVDD anomaly detection model.

[0066] Fig.11 This is a schematic diagram of an abnormality detection model detecting a fault according to an exemplary embodiment of the present application. Fig.12 This is a schematic diagram showing another exemplary embodiment of the present application showing a fault detected by an abnormality detection model. Fig.11 and Fig.12 , the error value calculated in real time is substituted into the SVDD model trained based only on normal working state data, and the model automatically determines whether the current state is normal or faulty. This is because when the real object is abnormal or faulty, its error characteristics show obvious differences, so that the error characteristics are obviously different from the normal spatial range after SVDD dimension conversion.

[0067] It should be noted that the state of the electrostatic hydraulic actuator is monitored in real time based on the trained SVDD abnormality detection model, including: the SVDD abnormality detection model continuously compares the error value of the predicted output and the output of the physical actuator after receiving the control instruction; when the SVDD abnormality detection model detects an abnormality, a warning or fault information is issued to the controller of the electrostatic hydraulic actuator; according to the fault information, the fault type and fault location are identified, and the control algorithm is adjusted based on the fault type; wherein the adjustment of the control algorithm at least includes parameter adjustment, switching control mode and compensation control; the controller receives the feedback information of the SVDD abnormality detection model in real time, and adjusts the control algorithm based on the feedback information. In this way, by real-time monitoring of the error value, detecting abnormalities and issuing warnings, identifying fault types and adjusting control algorithms, and real-time feedback to the controller, it is ensured that the electrostatic hydraulic actuator can quickly identify and adjust the control strategy when a fault occurs, thereby achieving fault-tolerant operation.

[0068] As an optional embodiment, the real-time monitoring of the state of the electrostatic hydraulic actuator based on the trained SVDD anomaly detection model also includes: obtaining the output values ​​of the physical actuator respectively according to control instructions; correcting the electrostatic hydraulic actuator simulation model according to the working conditions corresponding to the control instructions; calculating the predicted output value based on the corrected electrostatic hydraulic actuator simulation model and the control instructions; calculating the error value based on the output value of the physical actuator and the predicted output value, and calculating the prompt information according to the mapping relationship between the error value and the corresponding working condition; and identifying the fault of the electrostatic hydraulic actuator based on the prompt information and the error value.

[0069] The real-time online monitoring method of the electrostatic hydraulic actuator provided in this embodiment first obtains the normal performance data of each working condition, and establishes a nonlinear real-time mapping model based on the normal performance data and the neural network, the nonlinear real-time mapping model is used to receive the control instruction and predict the output of the actuator, and then trains the SVDD abnormality detection model according to the predicted output and the output of the actuator after receiving the control instruction, and finally monitors the state of the electrostatic hydraulic actuator in real time based on the trained SVDD abnormality detection model. In this way, by acquiring simulation data and combining the nonlinear fitting ability of the neural network to establish a nonlinear real-time mapping model, the problem of distortion of the traditional simplified model in the nonlinear section is solved, and the output of the actuator can be accurately predicted. In addition, by combining the SVDD abnormality detection model, it can more accurately detect minor abnormalities and faults, achieve real-time monitoring and rapid response, and reduce the probability of false alarms and missed alarms. In addition, in the face of the characteristics of the electrostatic hydraulic actuator under different working conditions, the simulation model that changes with the working condition is corrected by the difference in the working condition, which avoids repeated modeling and improves the accuracy of the simulation model prediction output. It should also be noted that the anomaly detection model can be established based only on the data under normal working conditions, that is, based on zero-fault samples, which solves the problems of lack of fault data and variable working conditions in engineering applications. It does not need to rely on professional experience and fault tests, has good flexibility, low cost, and high fault coverage, which can make up for the shortcomings of traditional BIT technology. In addition, the error between the model prediction and the actual output is used as the input of the anomaly detection model, and the anomaly detection model is trained by single performance data to improve the fault identification ability, as well as the overall performance and reliability of the model.

[0070] Corresponding to the aforementioned embodiment of a method for real-time online monitoring of an electrostatic-hydraulic actuator, the present application also provides an embodiment of a device for real-time online monitoring of an electrostatic-hydraulic actuator.

[0071] Fig.13 This is a structural diagram of the first embodiment of the real-time online monitoring device for electrostatic hydraulic actuators provided by the present application. Fig.13 , the device provided in this embodiment includes an establishment module 1310, a training module 1320 and a monitoring module 1330;

[0072] The establishment module 1310 is used to obtain normal performance data of each working condition, and establish a nonlinear real-time mapping model based on the normal performance data and the neural network, and the nonlinear real-time mapping model is used to receive control instructions and predict the output of the actuator;

[0073] The training module 1320 is used to train the SVDD anomaly detection model according to the predicted output and the output of the actuator after receiving the control instruction;

[0074] The SVDD anomaly detection model is trained with the error value between the normal working state data of the electrostatic hydraulic actuator and the predicted output, wherein the error value includes the error value under different working conditions, and the SVDD parameters are adjusted based on the error values ​​under different working conditions, so as to make the SVDD parameters adaptively adjusted under different feature dimensions according to the sensitive correlation between the feature and the fault;

[0075] The monitoring module 1330 is used to monitor the state of the electrostatic hydraulic actuator in real time based on the trained SVDD anomaly detection model.

[0076] The device of this embodiment can be used to perform Figure 1 The steps, specific implementation principles and implementation processes of the method embodiment shown are similar and will not be repeated here.

[0077] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0078] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present application scheme. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0079] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A real-time online monitoring method for an electrostatic hydraulic actuator, characterized in that: The method comprises: Acquire normal performance data of each working condition, and establish a nonlinear real-time mapping model based on the normal performance data and the neural network, wherein the nonlinear real-time mapping model is used to receive control instructions and predict the output of the actuator; wherein the normal performance data includes simulation data and test data of the electrostatic hydraulic actuator; The SVDD anomaly detection model is trained based on the predicted output and the output of the actuator after receiving the control command; The SVDD anomaly detection model is trained with the normal working state data of the electrostatic hydraulic actuator and the error value of the predicted output, wherein the error value includes the error value under different working conditions, and the SVDD parameters are adjusted based on the error values ​​under different working conditions, and the SVDD parameters are adaptively adjusted under different feature dimensions according to the sensitive correlation between the feature and the fault; the normal working state data includes the state data of the electrostatic hydraulic actuator in a healthy state; The state of the electrostatic hydraulic actuator is monitored in real time based on the trained SVDD anomaly detection model.

2. The method according to claim 1, characterized in that: The obtaining of normal performance data of each working condition includes: According to the characteristics and working requirements of the electrostatic hydraulic actuator, a simulation model of the electrostatic hydraulic actuator and an electrostatic hydraulic actuator test bench are established respectively; Inputting commands under different working conditions into the simulation model and the electrostatic hydraulic actuator test bench to obtain simulation data and test data respectively; the simulation data and test data both include control commands and response outputs; The simulation data and the test data are comprehensively complemented and used as normal performance data for each operating condition.

3. The method according to claim 1, characterized in that The nonlinear real-time mapping model is established based on the normal performance data of each working condition and the neural network, including: Based on the normal performance data of each working condition, feature extraction and fusion of data under different working conditions are performed; The neural network is trained successively according to working conditions using feature fusion data, and the last training result is used in each training process to fine-tune the network parameters. The neural network is used to describe the nonlinear mapping relationship between control instructions and response outputs.

4. The method according to claim 1, characterized in that Before training the SVDD anomaly detection model according to the predicted output and the output of the actuator after receiving the control instruction, the method includes: Calculate the amplitude error and phase error of the predicted output and the output of the physical actuator after receiving the control instruction; Determine the error value according to the error calculation result; A training sample is constructed based on the error value.

5. The method according to claim 4, characterized in that The method of training the SVDD anomaly detection model according to the predicted output and the output of the actuator after receiving the control instruction includes: Calculate the deviation between the predicted output of each working condition and the output of the actuator after receiving the control instruction according to the error value; Based on the deviation, the parameters of the SVDD model are adjusted, and according to the sensitivity correlation between each feature and the fault, the SVDD model is adaptively adjusted under different feature dimensions; wherein the SVDD model is trained based on normal working state data; The SVDD model under a single sample is trained based on the adjusted parameters and the error value under normal working conditions to obtain the SVDD anomaly detection model; the error value under normal working conditions is the difference between the normal working state data and the predicted output.

6. The method according to claim 1, characterized in that The real-time monitoring of the state of the electrostatic hydraulic actuator based on the trained SVDD anomaly detection model includes: The SVDD anomaly detection model continuously compares the error value of the predicted output and the output of the physical actuator after receiving the control instruction; When the SVDD abnormality detection model detects an abnormality, a warning or fault message is sent to a controller of the electrostatic hydraulic actuator; According to the fault information, identify the fault type and the fault location, and adjust the control algorithm based on the fault type; wherein adjusting the control algorithm at least includes parameter adjustment, switching control mode and compensation control; The controller receives feedback information of the SVDD anomaly detection model in real time and adjusts the control algorithm based on the feedback information.

7. The method according to claim 2, characterized in that The obtaining of simulation data further comprises: Establishing a simulation model of the electrostatic hydraulic actuator based on the first working condition of the electrostatic hydraulic actuator; Determining a corresponding second operating condition according to a control instruction of the electrostatic hydraulic actuator; Calculating a correction coefficient according to a difference between the second operating condition and the first operating condition; Correcting the electrostatic hydraulic actuator simulation model based on the correction coefficient; The simulation data corresponding to the control instruction is calculated based on the corrected electrostatic hydraulic actuator simulation model.

8. The method according to claim 1, characterized in that The training of the SVDD anomaly detection model according to the predicted output and the output of the actuator after receiving the control instruction also includes: Calculating a prediction error based on the output of the actuator after receiving the control instruction and the predicted output; Calculate a mapping relationship between an error change value and an operating condition based on the prediction error and the operating condition corresponding to the control instruction; Based on the mapping relationship as prompt information; The SVDD anomaly detection model is trained based on training samples and prompt information.

9. The method according to claim 1, characterized in that: The real-time monitoring of the state of the electrostatic hydraulic actuator based on the trained SVDD anomaly detection model includes: Obtain output values ​​of the physical actuators respectively according to the control instructions; Correct the electrostatic hydraulic actuator simulation model according to the working conditions corresponding to the control instructions; Calculating a predicted output value based on the corrected electrostatic hydraulic actuator simulation model and control instructions; Calculating an error value based on an output value of the physical actuator and a predicted output value, and calculating prompt information according to a mapping relationship between the error value and a corresponding working condition; A fault of the electrostatic hydraulic actuator is identified based on the prompt information and the error value.

10. A real-time online monitoring device for an electrostatic hydraulic actuator, characterized in that: The device comprises a building module, a training module and a monitoring module; wherein, The establishment module is used to obtain normal performance data of each working condition, and establish a nonlinear real-time mapping model based on the normal performance data of each working condition and the neural network, and the nonlinear real-time mapping model is used to receive control instructions and predict the output of the actuator; wherein the normal performance data includes simulation data and test data of the electrostatic hydraulic actuator; The training module is used to train the SVDD anomaly detection model according to the predicted output and the output of the actuator after receiving the control instruction; The SVDD anomaly detection model is trained with the normal working state data of the electrostatic hydraulic actuator and the error value of the predicted output, wherein the error value includes the error value under different working conditions, and the SVDD parameters are adjusted based on the error values ​​under different working conditions, and the SVDD parameters are adaptively adjusted under different feature dimensions according to the sensitive correlation between the feature and the fault; the normal working state data includes the state data of the electrostatic hydraulic actuator in a healthy state; The monitoring module is used to monitor the state of the electrostatic hydraulic actuator in real time based on the trained SVDD anomaly detection model.

Images