A working condition adaptive electromechanical actuator health state evaluation method

Abstract

This invention relates to the field of electromechanical actuator fault diagnosis and provides a condition-adaptive electromechanical actuator health status assessment method. Addressing the inconsistency between the health status assessment model and the actual health status trend under time-varying operating conditions, it proposes an adaptive electromechanical actuator health status assessment optimization method based on model-state linkage updates. This method utilizes filtering to update the EMA health factor state space model. Based on this, an adaptive health status assessment optimization model based on model-state linkage updates is established, enabling accurate online EMA health status assessment under time-varying operating conditions, thereby effectively assessing the actuator's health status under these conditions.

Description

Technical Field

[0001] This invention relates to the field of electromechanical actuator fault diagnosis, and in particular to a method for assessing the health status of an adaptive electromechanical actuator. Background Technology

[0002] Electro-Mechanical Actuators (EMAs) are a class of systems that directly or indirectly control the movement of a load by controlling the motion of an electric motor, achieving position / pressure servo control. They are widely used in aerospace, military, transportation, and industrial and agricultural production. In the aviation field, with the promotion of more-electric and all-electric aircraft, replacing traditional hydraulic actuators with power-driven electric actuators has become an inevitable trend. Power-driven electric actuators mainly come in two forms: electro-hydraulic actuators and EMAs. Compared to electro-hydraulic actuators, EMAs have many advantages such as compact structure, light weight, and ease of maintenance, and are increasingly widely used in more-electric / all-electric aircraft. However, the online operating conditions of the EMA (Electrical Maintenance Module) are typically time-varying, with rich dynamic characteristics such as control modes, loads, and speeds, as well as a wide variety of external loads and environmental factors. Under different operating conditions, the same online monitoring data reflects significantly different health states, leading to substantial differences in the output values ​​of the health status assessment model under the same online monitoring data input. However, the actual health state does not change significantly in a short period of time with changes in operating conditions, resulting in an inconsistency between the health status assessment model and the actual health state under time-varying conditions. Therefore, researching online health status assessment optimization methods to effectively evaluate the online health status of the EMA under the influence of time-varying operating conditions, thereby improving the online health status assessment capability of the EMA—a core component of next-generation aircraft—and promoting its intelligent development, has significant theoretical and engineering value.

[0003] Existing online operating conditions for EMAs are typically time-varying, exhibiting rich dynamic characteristics such as control modes, loads, and speeds, as well as a wide variety of external loads and environmental factors. Under different operating conditions, the same online monitoring data reflects significantly different health states, leading to substantial differences in the output values ​​of health status assessment models under the same online monitoring data input. However, the actual health state does not change significantly in a short period of time with changes in operating conditions, resulting in an inconsistency between the health status assessment model and the actual health state under time-varying operating conditions. Furthermore, existing adaptive health status assessment methods fail to consider time-varying operating condition factors, making it difficult to update the model according to changes in operating conditions and failing to effectively assess the online health status of EMAs under time-varying operating conditions. This results in a model mismatch problem in the health status assessment models for EMAs under time-varying operating conditions. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this application provides a method for assessing the health status of an adaptive electromechanical actuator. This application proposes a modeling and calculation method that updates the model based on monitored online operating parameters and estimates health factors based on the updated model to complete the assessment of the actuator's health status.

[0005] This application provides a method for assessing the health status of an adaptive electromechanical actuator under specific operating conditions, comprising the following steps:

[0006] S10. Construct a state-space model of health factors, which includes a dynamic model and a measurement model;

[0007] ,

[0008] in, represent The estimated vector of health factors at time 1; represent Measurement vectors of health factors at any given time; It is a state transition function that describes the dynamic process of the system; express The parameter vector of the state transition function at each time step. It is the measurement function that describes the measurement system; q k-1 This represents the error vector during the dynamic process of the system. ; This represents the error vector introduced by the measurement system. ; and These represent the error covariance matrix of the measurement system and the error covariance matrix of the system's dynamic process, respectively.

[0009] S20. Obtain the estimated vector x of health factors at time k-1. k-1 The error covariance matrix P at time k-1 k-1 The mean of the estimated vector of health factors at time k is obtained using an unscented Kalman filter. The Kalman gain matrix K at time k k The mean of the health factor measurement estimation vector at time k ;

[0010] S30. Online monitoring of the health factor vector z at time k. k And update the estimated vector x of health factors at time k. k The error covariance matrix P at time k k ;

[0011]

[0012]

[0013] in, It is the autocovariance matrix;

[0014] S40. Obtain online operating condition monitoring parameters at time k. Based on the updated estimated vector x of health factors at time k k And the estimated value of the parameter vector at time k-1. The parameter vector error covariance matrix at time k-1 The parameter vector estimate at time k is obtained using an unscented Kalman filter. and its error covariance matrix ,

[0015] ,

[0016] represent The estimated values ​​of the transition function parameter vector at time t; represent Parameter values ​​are monitored online at all times. It is a state transition function that describes the dynamic process of the parameter vector; It is a mapping function describing the relationship between the parameter vector and the operating condition monitoring parameters; m k-1 This represents the error during the dynamic process of the parameter vector. ; This represents the error introduced by the mapping function. ; and These represent the error covariance of the mapping function and the error covariance matrix of the parameter vector dynamic process, respectively.

[0017] S50. Let k = k + 1, return to step S20, and update M in step S40. k Substitute S20.

[0018] The adaptive health status assessment method for electromechanical actuators (EMA) based on the operating conditions described in this application has at least the following advantages compared to existing technologies: This application updates the health factor state model based on monitored time-varying operating conditions and calculates health factors using an unscented Kalman filter method, thereby assessing the actuator's health status. The method of this application can solve the problem of inconsistency between the health status assessment model and the actual health status under time-varying operating conditions, effectively realizing the adaptive health status assessment of EMA under time-varying operating conditions. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the working condition adaptive electromechanical actuator health status assessment method described in Example 1. Detailed Implementation

[0021] The following description provides many different embodiments or examples for implementing various features of the invention. The elements and arrangements described in the specific examples below are only for concise expression of the invention and are merely examples, not intended to limit the invention.

[0022] A method for assessing the health status of an adaptive electromechanical actuator under specific operating conditions, such as Figure 1 As shown, it includes the following steps:

[0023] S10. Construct a state-space model of health factors, which includes a dynamic model and a measurement model;

[0024] ,

[0025] in, represent The estimated vector of health factors at time 1; represent Measurement vectors of health factors at any given time; It is a state transition function that describes the dynamic process of the system; express The parameter vector of the state transition function at each time step. It is the measurement function that describes the measurement system; q k-1 This represents the error vector during the dynamic process of the system. ; This represents the error vector introduced by the measurement system. ; and These represent the error covariance matrix of the measurement system and the error covariance matrix of the system's dynamic process, respectively.

[0026] S20. Obtain the estimated vector x of health factors at time k-1. k-1 The error covariance matrix P at time k-1 k-1 The mean of the estimated vector of health factors at time k is obtained using an unscented Kalman filter. The Kalman gain matrix K at time kk The mean of the health factor measurement estimation vector at time k ;

[0027] Step S20 involves calculations based on an unscented Kalman filter. The calculations based on an unscented Kalman filter specifically include the following steps:

[0028] S21. Based on the estimated vector x of the health factors at time k-1. k-1 The error covariance matrix P at time k-1 k-1 Sigma point sampling is performed to obtain 2n+1 Sigma points and their corresponding weight parameters, forming the first point set. ;

[0029]

[0030] The weights for the mean and variance are:

[0031]

[0032]

[0033] in, It is a parameter that adjusts the distance from the Sigma point to the health factor; and These are the weights for the mean and variance, respectively. i Representation matrix The List;

[0034] S22. Set the first point. Substitute into the dynamic model In the middle, find the second point set. The average value of the second set of points is denoted as... And according to the second point set Obtain prior estimates of health factors and the estimated error covariance matrix The calculation formula is as follows:

[0035]

[0036]

[0037] S23. Set the second point. Substitute into the measurement model To obtain the third set of points The third point set is the health factor measurement value z at time k. k The sampling points were determined, and the mean of the estimated health factor measurements at time k was calculated. Autocovariance matrix and cross-covariance matrix ;

[0038] S24. Calculate the Kalman gain matrix K at time k. k , .

[0039] The average value of the estimated health factors at time k is calculated based on the above steps. The Kalman gain matrix K at time k k The mean of the estimated values ​​of health factor measurements at time k Then, the estimated value x of the health factor at time k is updated based on the actual detected health factor at time k. k That is, execute step S30.

[0040] S30. Online monitoring of the health factor vector z at time k. k And update the estimated vector x of health factors at time k. k The error covariance matrix P at time k k ;

[0041]

[0042]

[0043] in, It is the autocovariance matrix.

[0044] Therefore, steps S10-S30 have completed the estimation vector x of the health factor at the current moment based on the health factor state space model established at the previous moment. k Following the update, the health factor state space model also needs to be updated based on the current online operating conditions, as shown in step S40 below:

[0045] S40. Obtain online operating condition monitoring parameters at time k. Based on the updated estimated vector x of health factors at time k k And the estimated value of the parameter vector at time k-1. The parameter vector error covariance matrix at time k-1 The parameter vector estimate at time k is obtained using an unscented Kalman filter. and its error covariance matrix ,

[0046] ,

[0047] represent The estimated values ​​of the transition function parameter vector at time t; represent Parameter values ​​are monitored online at all times. It is a state transition function that describes the dynamic process of the parameter vector; It is a mapping function describing the relationship between the parameter vector and the operating condition monitoring parameters; m k-1 This represents the error during the dynamic process of the parameter vector. ; This represents the error introduced by the mapping function. ; and These represent the error covariance matrix of the mapping function and the error covariance matrix of the parameter vector dynamic process, respectively.

[0048] It is worth noting that the principle and process of calculation using an unscented Kalman filter in step S40 are the same as those in step S20, and will not be repeated here.

[0049] Therefore, step S40 uses the online health monitoring parameters at the current moment and the estimated value x of the health factor at the current moment calculated in the previous steps. k Once the parameter vector estimates for the current time step are calculated and updated... This is then incorporated into the dynamic model, thus completing the update of the health factor state space model.

[0050] S50. Let k = k + 1, return to step S20, and update M in step S40. k Substitute S20.

[0051] Of course, those skilled in the art will understand that the above steps are all iterative calculations based on time. In the first step of the calculation, it is necessary to assign an initial value to the input parameter, namely the estimated value x of the health factor at time k-1. k-1 The error covariance matrix P at time k-1 k-1 The estimated value of the parameter vector at time k-1 The parameter vector error covariance matrix at time k-1 These values ​​all require an initial value. Those skilled in the art can assign initial values ​​based on experience or historical data.

[0052] This application utilizes the dynamic update characteristics of the dual filtering algorithm to achieve iterative updates of the Wiener process state equation and health status assessment model, establishing an adaptive health status assessment model that updates both the state and the operating conditions. This effectively enables dynamic updates of the EMA adaptive health assessment model under time-varying operating conditions, thereby effectively assessing the health status of actuators under time-varying operating conditions.

[0053] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for health condition assessment of a working condition adaptive electro-mechanical actuator, characterized in that, comprising the steps of: S10. constructing a health factor state space model, the health factor state space model comprising a dynamic model and a measurement model; ， wherein represents an estimated vector of health factors at time instant represents a measured vector of health factors at time instant is a state transition function describing the system dynamics; denotes a parameter vector of the state transition function at time instant is a measurement function describing the measurement system; q k-1 represents an error vector in the system dynamics, represents an error vector introduced by the measurement system, and denote the error covariance matrix of the measurement system and the error covariance matrix of the system dynamics, respectively. S20. Obtain the estimation vector x of health factors at time k-1 k-1 , error covariance matrix P at time k-1 k-1 , obtain the mean of the estimation vector x of health factors at time k by using the unscented Kalman filter , Kalman gain matrix K at time k k , the mean of the health factor measurement estimation vector at time k ; S30. Online monitoring of the health factor vector z at time k k and updating the estimate vector x of the health factor at time k k and the error covariance matrix P at time k k ; wherein is the autocovariance matrix; S40. Obtain the online working condition monitoring parameter at time k , according to the updated estimation vector x of the health factor at time k k , and the estimation value of the parameter vector at time k-1 , the parameter vector error covariance matrix at time k-1 , obtain the estimation value of the parameter vector at time k by using the unscented Kalman filter and the error covariance matrix thereof , ， represents the estimated value of the transfer function parameter vector at the time instant; represents the value of the operating condition monitoring parameter at the time instant; is a state transfer function describing the dynamic process of the parameter vector; is a mapping function describing the relationship between the parameter vector and the operating condition monitoring parameter; m k-1 represents the error in the parameter vector dynamic process, ; represents the error introduced by the mapping function, ; and respectively represent the error covariance of the mapping function and the error covariance matrix of the parameter vector dynamic process; S50. Let k = k + 1, return to step S20, and update M k Go to S20.

2. The method according to claim 1, wherein, The step S20 comprises: S21. The estimation vector x of the health factor at time k-1 k-1 and the error covariance matrix P at time k-1 k-1 , Sigma point sampling is performed to obtain 2n+1 Sigma points and corresponding weight parameters, forming a first point set ; The weights of the mean and variance are: wherein, is a parameter that regulates the Sigma point to health factor distance; and are weight values for mean and variance, respectively, i denotes the i-th column of the matrix ; and denotes the i-th column of the matrix S22. The first point set is substituted into the dynamic model to obtain the second point set The average value of the second point set is denoted as The health factor prior estimate value and the estimation error covariance matrix are obtained according to the second point set The calculation formula is as follows: S23. obtaining a second point set substituting the measurement model , obtaining a third point set , the third point set being the sampling points of the health factor measurement value z k at time k, and calculating the health factor measurement estimation value mean , the auto-covariance matrix and the cross-covariance matrix at time k; S24. Calculate the Kalman gain matrix K at time k k , .

3. The method according to claim 1 or 2, wherein At the first calculation, the estimated value x of the health factor at time k-1 is determined empirically k-1 , the error covariance matrix P of the parameter vector at time k-1 k-1 , the estimated value x of the parameter vector at time k-1 , the error covariance matrix P of the parameter vector at time k-1 .

Images