CN115270591A - Method and system for monitoring health state of electromechanical actuator based on composite space - Google Patents

Abstract

The invention discloses a method and a system for monitoring the health state of an electromechanical actuator based on a composite space, which relate to the field of dynamic monitoring, diagnosis and maintenance of mechanical systems and mainly comprise the following steps: acquiring operation data of the target electromechanical actuator at the previous moment and acquiring an actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment; predicting a fault monitoring quantity reference value of the target electromechanical actuator at the current moment based on the operating data of the target electromechanical actuator at the previous moment and a machine learning algorithm; and monitoring the fault degree of each layer in the composite health state space of the target electromechanical actuator in a step-by-step mapping mode based on the fault monitoring amount reference value of the target electromechanical actuator at the current moment and the fault monitoring amount actual value of the target electromechanical actuator at the current moment so as to obtain the health state of the target electromechanical actuator. The invention can monitor the health state of each part of the electromechanical actuator and the health state of the whole electromechanical actuator in real time.

Description

A method and system for monitoring the health status of electromechanical actuators based on composite space

technical field

The invention relates to the field of dynamic monitoring, diagnosis and maintenance of mechanical systems, in particular to a method and system for monitoring the health status of an electromechanical actuator based on a composite space.

Background technique

Electro-mechanical actuators (Electro-mechanical Actuators, EMA) is a general term for a class of actuators that directly or indirectly control the motion of the load by controlling the motion of the motor to achieve position or pressure servo control. It is widely used in aerospace, military, transportation, fields of industrial and agricultural production. With the advent of next-generation aerospace systems equipped with fly-by-wire controls, electromechanical actuators are rapidly becoming safety-critical components of aerospace vehicles, driving objects such as aircraft, spacecraft, ground vehicles, and controlling their position , height, etc.

As a key component in the system, electromechanical actuators often work under various complex working conditions, and are prone to various forms of failures such as short circuit, deformation, and stagnation. If the electromechanical actuator fails, it will lead to performance degradation. It may affect other devices in the system, and even cause the entire system to fail to work in severe cases. Therefore, it is of great significance to evaluate the health status of electromechanical actuators for their safe operation.

From the actual working process of the electromechanical actuator, it can be seen that, except for some sudden failures, most of the failures of the electromechanical actuator are the process of accumulation, and the overall health status can be accurately controlled. When its performance declines, an alarm is raised to the upper-level control unit in time, and corresponding adjustments or fault-tolerant control can be carried out, which can well prevent the occurrence of serious failures. However, an electromechanical actuator is a complex electromechanical system composed of multiple components such as a controller, a drive motor, a reduction gear box, and a ball screw. In different links, each component may also have multiple failure modes at the same time, causing complex effects on the performance of the electromechanical actuator. In addition, due to the limited accumulation of failure data of electromechanical actuators, the failure modes and effects are mostly based on industrial experience, and the interference of nonlinear factors on the system output in the fault state is not clear, which leads to the current health of electromechanical actuators. The difficulty of state assessment is further increased.

Currently available electromechanical actuator health monitoring methods can be divided into two types: model-based methods and data-based methods. Although the above methods can evaluate the health status of the electromechanical actuator to a certain extent, they cannot evaluate the health status of each component and the overall health status of the electromechanical actuator in real time.

Contents of the invention

The purpose of the present invention is to provide a method and system for monitoring the health status of an electromechanical actuator based on a composite space, which can monitor the health status of each component and the overall health status of the electromechanical actuator in real time.

To achieve the above object, the present invention provides the following scheme:

In the first aspect, the present invention provides a composite space-based electromechanical actuator health status monitoring method for monitoring the health status of the target electromechanical actuator; the target electromechanical actuator includes a plurality of target components, and each The target component corresponds to one or more fault modes, and each of the fault modes corresponds to a fault monitoring quantity; the composite health state space of the target electromechanical actuator includes a bottom layer, a middle layer and a top layer; the bottom layer includes the All failure modes of the target electromechanical actuator; the middle layer includes all target components of the target electromechanical actuator; the top layer is the target electromechanical actuator; the electromechanical actuator health status monitoring method, include:

Obtaining the operating data of the target electromechanical actuator at the previous moment, and obtaining the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment;

Based on the operating data of the target electromechanical actuator at the previous moment and the machine learning algorithm, predict the fault monitoring quantity reference value of the target electromechanical actuator at the current moment;

Based on the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment, the composite health state of the target electromechanical actuator is calculated by using a step-by-step mapping method The fault degree of each layer in the space is monitored to obtain the health status of the target electromechanical actuator; the health status includes the fault degree of each failure mode in the target electromechanical actuator, the target electromechanical actuator The failure degree of each target component in the actuator and the overall failure degree of the target electromechanical actuator.

Optionally, also include:

Perform fault feature analysis on each of the fault modes in the target electromechanical actuator, and determine a fault monitoring quantity used for monitoring the fault modes.

Optionally, it also includes: constructing a regression model based on a long short-term memory neural network.

Optionally, the constructing a regression model based on the long-short-term memory neural network specifically includes:

Build a regression neural network; the regression neural network includes a long short-term memory neural network layer and a fully connected neural network layer;

Constructing a training sample set; the training sample set includes a plurality of sample input data and label data corresponding to each of the sample input data; the sample input data is the operating data of the target electromechanical actuator in a normal state; The tag data is the actual value of the fault monitoring quantity of the target electromechanical actuator under normal conditions;

The regression neural network is trained based on the training sample set to obtain a regression model.

Optionally, the predicting the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment based on the operating data of the target electromechanical actuator at the previous moment and the machine learning algorithm specifically includes:

Preprocessing the operating data of the target electromechanical actuator at the previous moment; the preprocessing includes data splicing operations, normalization operations, and sliding window cutting operations;

Based on the preprocessed operating data of the target electromechanical actuator at the previous moment and the regression model, predict the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment.

Optionally, based on the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment, the target electromechanical actuator is mapped using a step-by-step mapping method. The fault degree of each layer in the compound health state space of the actuator is monitored to obtain the health state of the target electromechanical actuator, which specifically includes:

calculating the fault degree of each of the fault modes based on the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment;

Based on the composite health state space of the target electromechanical actuator, determine the relationship between the failure mode and the target component, and based on the relationship between the failure mode and the target component, and the failure of each of the failure modes degree, determining the degree of failure of each of said target components;

An overall failure degree of the target electromechanical actuator is determined based on the failure degree of each of the target components.

In the second aspect, the health status monitoring system of an electromechanical actuator based on a composite space provided by the present invention is used to monitor the health status of a target electromechanical actuator; the target electromechanical actuator includes a plurality of target components, and each The target component corresponds to one or more fault modes, and each of the fault modes corresponds to a fault monitoring quantity; the composite health state space of the target electromechanical actuator includes a bottom layer, a middle layer and a top layer; the bottom layer includes the All failure modes of the target electromechanical actuator; the middle layer includes all target components of the target electromechanical actuator; the top layer is the target electromechanical actuator; the electromechanical actuator health status monitoring system, include:

A data acquisition module, configured to acquire the operating data of the target electromechanical actuator at the previous moment, and acquire the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment;

The fault monitoring quantity reference value prediction module is used to predict the fault monitoring quantity reference value of the target electromechanical actuator at the current moment based on the operating data of the target electromechanical actuator at the previous moment and the machine learning algorithm;

The electromechanical actuator health status determination module is configured to use a step-by-step mapping method based on the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment. The fault degree of each layer in the composite health state space of the target electromechanical actuator is monitored to obtain the health state of the target electromechanical actuator; the health state includes each of the target electromechanical actuators The failure degree of the failure mode, the failure degree of each target component in the target electromechanical actuator, and the overall failure degree of the target electromechanical actuator.

Optionally, it also includes: a regression model building module, which is used to build a regression model based on a long-short-term memory neural network.

Optionally, the fault monitoring quantity reference value prediction module specifically includes:

A preprocessing unit, configured to preprocess the operating data of the target electromechanical actuator at the previous moment; the preprocessing includes data splicing operations, normalization operations, and sliding window cutting operations;

A fault monitoring quantity reference value prediction unit, configured to predict the fault monitoring quantity reference value of the target electromechanical actuator at the current moment based on the preprocessed operating data of the target electromechanical actuator at the previous moment and the regression model .

Optionally, the health state determining module of the electromechanical actuator specifically includes:

The failure degree calculation unit of the failure mode is used to calculate each failure mode based on the reference value of the failure monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the failure monitoring quantity of the target electromechanical actuator at the current moment the degree of failure;

A failure degree calculation unit of the target component, configured to determine the relationship between the failure mode and the target component based on the composite health state space of the target electromechanical actuator, and based on the relationship between the failure mode and the target component , and the failure degree of each of the failure modes, determining the failure degree of each of the target components;

The overall failure degree determination unit of the target electromechanical actuator is configured to determine the overall failure degree of the target electromechanical actuator based on the failure degree of each of the target components.

According to the specific embodiments provided by the invention, the invention discloses the following technical effects:

Compared with the previous electromechanical actuator health state monitoring method, the electromechanical actuator health state monitoring method and system based on the composite space provided by the present invention can obtain the fault monitoring quantity reference value of the electromechanical actuator in real time through the machine learning algorithm; Through the composite health state space of the electromechanical actuator, the reference value of the fault monitoring quantity, the actual value of the fault monitoring quantity, and step by step calculation, not only the failure degree of the typical failure mode of the electromechanical actuator can be analyzed, but also the fault degree of the electromechanical actuator can be analyzed. The failure degree of each key component in the actuator is analyzed, and on this basis, the overall health status index of the electromechanical actuator is obtained, that is, the failure degree, which can better reflect the actual operating status of the electromechanical actuator.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required in the embodiments. Obviously, the accompanying drawings in the following description are only some of the present invention. Embodiments, for those of ordinary skill in the art, other drawings can also be obtained according to these drawings without paying creative labor.

Fig. 1 is a schematic flow chart of a method for monitoring the health state of an electromechanical actuator based on a composite space in the present invention;

Fig. 2 is a specific flow chart of a method for monitoring the health status of an electromechanical actuator based on a composite space in the present invention;

Fig. 3 is a schematic diagram of the complex health state space constructed by the present invention for electromechanical actuators;

Fig. 4 is a schematic diagram of the training results of the present invention based on LSTM training of a certain type of electromechanical actuator output displacement reference model; Fig. 4 (a) is the test set 1 training result of the present invention based on LSTM training of a certain type of electromechanical actuator output displacement reference model Schematic diagram; Fig. 4 (b) is the test set 2 training result schematic diagram of the present invention based on LSTM training certain model electromechanical actuator output displacement reference model;

Fig. 5 is the schematic diagram of the training result of the control panel current reference model of the present invention based on LSTM training of a certain type of electromechanical actuator motor; Fig. 5 (a) is the control panel current reference model of the present invention based on LSTM training of a certain type of electromechanical actuator motor A schematic diagram of the test set 1 training results; Fig. 5 (b) is a schematic diagram of the test set 2 training results of the control panel current reference model of a certain type of electromechanical actuator motor in the present invention based on LSTM training;

Fig. 6 is the training result schematic diagram of the standard deviation reference model of the standard deviation reference model of the present invention based on LSTM training certain type electromechanical actuator ball screw vibration signal; Fig. 6 (a) is the present invention based on LSTM training certain type electromechanical actuator ball screw vibration A schematic diagram of the test set 1 training results of the standard deviation reference model of the signal; Fig. 6 (b) is a schematic diagram of the test set 2 training results of the standard deviation reference model of a certain type of electromechanical actuator ball screw vibration signal based on the LSTM training of the present invention;

Fig. 7 is a schematic diagram of the monitoring results of the present invention when a certain type of electromechanical actuator is in a normal state; Fig. 7 (a) is a schematic diagram of the control board current monitoring results of the present invention when a certain type of electromechanical actuator is in a normal state; Fig. 7 (b) is a schematic diagram of the output displacement monitoring results of the present invention when a certain type of electromechanical actuator is in a normal state; Fig. 7 (c) is the vibration signal standard deviation monitoring result of the present invention when a certain type of electromechanical actuator is in a normal state Schematic diagram; Figure 7 (d) is a schematic diagram of the monitoring results of the transmission mechanism jamming fault degree when a certain type of electromechanical actuator is in a normal state according to the present invention; Figure 7 (e) is a schematic diagram of the present invention for a certain type of electromechanical actuator in a normal state The schematic diagram of the monitoring results of the fault degree of the transmission mechanism when the gap is too large; Fig. 7 (e) is a schematic diagram of the monitoring results of the surface damage of the ball screw when a certain type of electromechanical actuator is in a normal state; Fig. 7 (g) is The present invention is a schematic diagram of the monitoring results of the electromechanical actuator health indicators when a certain type of electromechanical actuator is in a normal state;

Fig. 8 is a schematic diagram of the monitoring results of the present invention when a certain type of electromechanical actuator is injected into the stuck fault state of the transmission mechanism; Fig. 8 (a) is the control of the present invention when a certain type of electromechanical actuator is injected into the stuck fault state of the transmission mechanism Schematic diagram of plate current monitoring results; Figure 8 (b) is a schematic diagram of the output displacement monitoring results of the present invention when a certain type of electromechanical actuator is injected into a transmission mechanism stuck fault state; Figure 8 (c) is a schematic diagram of the present invention for a certain type of electromechanical actuator Schematic diagram of the vibration signal standard deviation monitoring results when the actuator is injected into the stuck fault state of the transmission mechanism; Figure 8 (d) is the monitoring result of the stuck fault degree of the transmission mechanism when a certain type of electromechanical actuator is injected into the stuck fault state of the transmission mechanism according to the present invention Schematic diagram; Figure 8 (e) is a schematic diagram of the monitoring results of the excessive fault degree of the transmission mechanism gap when a certain type of electromechanical actuator is injected into a transmission mechanism stuck fault state; Figure 8 (e) is a schematic diagram of the present invention for a certain type of electromechanical actuator Figure 8(g) is a schematic diagram of the monitoring result of the surface damage of the ball screw when the actuator is injected into the stuck fault state of the transmission mechanism; Figure 8 (g) is the electromechanical actuator of the present invention when a certain type of electromechanical actuator is injected into the stuck fault state of the transmission mechanism Schematic diagram of health indicator monitoring results;

Fig. 9 is a schematic diagram of the monitoring results of the present invention when a certain type of electromechanical actuator is injected into a ball screw surface damage fault state; Fig. 9 (a) is a schematic diagram of the present invention when a certain type of electromechanical actuator is injected into a ball screw surface damage fault state A schematic diagram of the current monitoring results of the control board; Fig. 9 (b) is a schematic diagram of the output displacement monitoring results of the present invention when a certain type of electromechanical actuator is injected into a ball screw surface damage fault state; Fig. 9 (c) is a schematic diagram of the present invention for a certain model Schematic diagram of the vibration signal standard deviation monitoring results when the electromechanical actuator is injected into the ball screw surface damage fault state; Figure 9 (d) is the transmission mechanism card when a certain type of electromechanical actuator is injected into the ball screw surface damage fault state according to the present invention Schematic diagram of monitoring result of hysteresis fault degree; Fig. 9 (e) is a schematic diagram of monitoring result of excessive gap of transmission mechanism when a certain type of electromechanical actuator is injected into ball screw surface damage fault state according to the present invention; Fig. 9 (e) is the present invention Schematic diagram of the monitoring results of the surface damage fault degree of the ball screw when the invention injects a certain type of electromechanical actuator into the ball screw surface damage fault state; Figure 9 (g) is the surface damage of the ball screw injected into a certain type of electromechanical actuator by the present invention Schematic diagram of the monitoring results of the health indicators of the electromechanical actuator in the fault state;

FIG. 10 is a schematic structural diagram of a health status monitoring system for electromechanical actuators based on a composite space according to the present invention.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

In order to solve the need of electromechanical actuators to monitor the health of each part of the system as a whole through a quantitative index under the premise of maintaining high real-time performance, the present invention proposes a health status monitoring of electromechanical actuators based on composite space The method and system provide an electromechanical actuator health state monitoring method and system with high real-time performance, strong versatility, and comprehensive quantitative monitoring of each failure mode of the system, each component and the whole.

The method and system for monitoring the health status of electromechanical actuators based on the composite space provided by the present invention mainly includes two parts: the construction of the composite health status space of the electromechanical actuators, and the monitoring of the health status of the electromechanical actuators based on the composite space.

In order to make the above objects, features and advantages of the present invention more comprehensible, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

Embodiment one

An electromechanical actuator health status monitoring method based on a composite space provided by an embodiment of the present invention is used to monitor the health status of a target electromechanical actuator; the target electromechanical actuator includes a plurality of target components, and each of the The target component corresponds to one or more failure modes, and each failure mode corresponds to a fault monitoring quantity; the composite health state space of the target electromechanical actuator includes the bottom layer, the middle layer and the top layer; the bottom layer includes the target electromechanical all failure modes of the actuator; the middle layer includes all target components of the target electromechanical actuator; the top layer is the target electromechanical actuator.

As shown in Figure 1, the method for monitoring the health status of the electromechanical actuator provided by the embodiment of the present invention includes:

Step 101: Obtain the operating data of the target electromechanical actuator at the previous moment, and obtain the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment.

Step 102: Based on the operating data of the target electromechanical actuator at the previous moment and the machine learning algorithm, predict the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment.

Step 103: Based on the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment, use a step-by-step mapping method to map the target electromechanical actuator The fault degree of each layer in the composite health state space is monitored to obtain the health state of the target electromechanical actuator; the health state includes the fault degree of each failure mode in the target electromechanical actuator, the The failure degree of each target component in the target electromechanical actuator and the overall failure degree of the target electromechanical actuator.

On the basis of the embodiment described in FIG. 1 , the method for monitoring the health status of the electromechanical actuator provided by the embodiment of the present invention further includes performing a fault characteristic analysis on each of the failure modes in the target electromechanical actuator, A fault monitoring quantity for use in monitoring the fault mode is determined.

On the basis of the embodiment described in FIG. 1 , the method for monitoring the health state of the electromechanical actuator provided by the embodiment of the present invention further includes: constructing a regression model based on a long-short-term memory neural network.

Wherein, the construction process of the regression model is, specifically including:

Step A: constructing a recurrent neural network; the recurrent neural network includes a long short-term memory neural network layer and a fully connected neural network layer.

Step B: Construct a training sample set; the training sample set includes a plurality of sample input data and label data corresponding to each of the sample input data; the sample input data is the target electromechanical actuator in a normal state Operation data; the tag data is the actual value of the fault monitoring quantity of the target electromechanical actuator in a normal state.

Step C: Preprocessing the data in the training sample set. The preprocessing includes data splicing operations, normalization operations and sliding window cutting operations.

Step D: Train the regression neural network based on the preprocessed training sample set to obtain a regression model.

Further, on the basis of the regression model, based on the operation data of the target electromechanical actuator at the previous moment and the machine learning algorithm, predict the fault monitoring quantity reference value of the target electromechanical actuator at the current moment, specifically include:

First, preprocessing is performed on the operating data of the target electromechanical actuator at the previous moment; the preprocessing includes data splicing operations, normalization operations, and sliding window cutting operations. Secondly, based on the operating data of the target electromechanical actuator at the previous moment after preprocessing and the regression model, predict the fault monitoring quantity reference value of the target electromechanical actuator at the current moment.

On the basis of the embodiment described in Figure 1, the reference value of the fault monitoring quantity based on the target electromechanical actuator at the current moment and the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment are mapped step by step The method monitors the fault degree of each layer in the composite health state space of the target electromechanical actuator to obtain the health state of the target electromechanical actuator, specifically including:

Step a: Calculate the failure degree of each failure mode based on the reference value of the failure monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the failure monitoring quantity of the target electromechanical actuator at the current moment.

Step b: Determine the relationship between the failure mode and the target component based on the composite health state space of the target electromechanical actuator, and based on the relationship between the failure mode and the target component, and each of the faults The degree of failure of the mode determines the degree of failure of each of said target components.

Step c: Determine the overall failure degree of the target electromechanical actuator based on the failure degree of each of the target components.

Embodiment two

An embodiment of the present invention provides a method for monitoring the health status of an electromechanical actuator based on a composite space; wherein the electromechanical actuator is a complex electromechanical system composed of multiple components such as a controller, a drive motor, a transmission mechanism, and a ball screw .

As shown in FIG. 2 , the composite space-based electromechanical actuator health status monitoring method provided by the embodiment of the present invention includes the following steps.

Step 201: First, select several typical faults that are common or have great influence when the electromechanical actuator is working, and then analyze the characteristics of the typical faults of the electromechanical actuator to obtain the system state quantity that can represent the degree of the fault as the fault monitoring quantity. The steps are specifically:

Based on the generation mechanism of several common failure modes during the operation of electromechanical actuators, a certain failure characteristic analysis is carried out to determine the fault monitoring quantity used to evaluate the failure mode, that is, the current failure mode can be represented during its operation. Several state parameters for the degree of failure, and the failure threshold when a failure occurs. The greater the deviation between the measured fault monitoring quantity and its reference value when the electromechanical actuator is normal, the more serious the fault degree of the corresponding fault mode is.

The typical failure modes of the electromechanical actuators selected in the embodiment of the present invention include motor winding turn-to-turn short circuit faults, motor bearing surface damage faults, ball screw surface damage faults, transmission mechanism stuck faults, and transmission mechanism clearance faults. in,

The fault monitoring quantity corresponding to the inter-turn short circuit fault of the motor winding is the difference of the three-phase current of the motor;

The fault monitoring quantity corresponding to the motor bearing surface damage fault is the vibration amplitude at the fault characteristic frequency of the motor bearing;

The fault monitoring quantity corresponding to the surface damage fault of the ball screw is the standard deviation of the vibration signal of the ball screw (indicating the vibration intensity);

The fault monitoring quantity corresponding to the stuck fault of the transmission mechanism is the current of the control board of the motor;

The fault monitoring quantity of the transmission mechanism clearance is too large is the output displacement of the electromechanical actuator/rudder surface deflection angle.

Step 202: Carry out an experiment on a fully normal electromechanical actuator that has not failed, and collect its operating data and system state quantities under various input signals and different working conditions; the operating data includes sine waves, square waves, trapezoidal Electromechanical actuator command signals with different waveforms, different frequencies, and different amplitudes such as waves, various input signals, motor control input voltages under different working conditions, and load data under different load conditions. The system state quantity includes motor control board current, motor three-phase current, motor vibration signal, lead screw vibration signal and output displacement.

Step 203: Construct a regression model based on the long-short-term memory neural network, and then input the preprocessed input signal into the regression model to predict the reference state quantity corresponding to each fault monitoring quantity at the next moment.

Among them, according to the collected experimental data of different input signals and different working conditions, the regression network is trained to obtain a regression model, which is used as a reference model for normal electromechanical actuators. The regression model consists of a long short-term memory neural network layer and a fully connected neural network layer. The input data is the command signal of the electromechanical actuator corresponding to the current moment, the motor control input voltage, and the load data, and the output data is the control board current of the motor at the next moment, the difference between the three-phase currents of the motor, and the vibration at the characteristic frequency of the motor bearing fault The amplitude, standard deviation of the ball screw vibration signal, and the output displacement of the electromechanical actuator.

The training process of the regression model is:

For the data of different channels collected in step 202, that is, the data of different components collected by different sensors, the input data is spliced according to the collection time. Assuming that the length of the collected data is t, the spliced signal can be regarded as an n-dimensional time-domain data, which includes information of n sensor channels. For the input signal of the regression network, n=3, including the command signal of the electromechanical actuator, the input voltage of the motor control, and the load data. The data of different sensor channels have different dimensions, so it is necessary to normalize the n-dimensional time-domain data through maximum and minimum normalization. In order to increase the number of samples, the normalized n-dimensional time domain data is cut through the sliding window. Each sliding window contains time domain data within a short period of time, and the data is divided into 0-1s, 0.01-1.01s, 0.02 -1.02s, and so on, to obtain the training set of the regression network, which includes the processed electromechanical actuator command signal, motor control input voltage, load data, that is, the input sample, and the motor control board current, motor three The difference between the phase currents, the vibration amplitude at the fault characteristic frequency of the motor bearing, the standard deviation of the vibration signal of the ball screw, and the output displacement of the electromechanical actuator, that is, the label data.

The model is trained through the training set; since the data in the training set are the experimental data of the normal electromechanical actuator working under different experimental conditions, these data represent the working model of the normal electromechanical actuator. Using the trained model, that is, the regression model, as the reference model of the electromechanical actuator, the reference value of the fault monitoring quantity of each channel when the electromechanical actuator is not faulty can be calculated through the input data, that is, through the input within a period of time, it can be Obtain the value that should be output by the electromechanical actuator fault monitoring quantity under the normal state at the next moment.

Step 204: Based on the constructed complex health state space of electromechanical actuators, through the collected multi-channel state data of electromechanical actuators and the reference model obtained from training, the fault degree of each layer is calculated layer by layer according to the method of step-by-step mapping Evaluation, and finally comprehensively consider the overall health index of the electromechanical actuator. The steps are specifically:

First, measure the data of each channel when the electromechanical actuator to be evaluated is working, and the data type of each channel measured is the same as the data collected in step 202 .

Then, based on the electromechanical actuator reference model trained in step 203, by using the processed electromechanical actuator command signal, motor control input voltage, and load data as input data, the reference for the control board current of the motor at the next moment is calculated. Value, the reference value of the difference between the three-phase current of the motor, the reference value of the vibration amplitude at the fault characteristic frequency of the motor bearing, the reference value of the standard deviation of the vibration signal of the ball screw, and the reference value of the output displacement of the electromechanical actuator.

Finally, the reference value at the next moment is compared with the actual measurement value at the next moment, and the fault degree of each layer is evaluated layer by layer according to the level-by-level mapping method, and finally the overall health index of the electric actuator is calculated.

Among them, the health assessment method of layer-by-layer mapping is divided into three layers. The failure mode layer of the bottom layer is based on the analysis of step 201, and the inter-turn short circuit of the motor winding is selected, the motor bearing surface is damaged, the ball screw surface is damaged, and the transmission clearance is too large. , the transmission mechanism is stuck, five typical failure modes of electromechanical actuators, as the five subspaces of the bottom layer; among them, the fault monitoring quantity corresponding to the inter-turn short circuit fault of the motor winding is the difference between the three-phase current of the motor, the motor bearing surface damage fault The corresponding fault monitoring quantity is the vibration amplitude at the fault characteristic frequency of the motor bearing, the fault monitoring quantity corresponding to the surface damage fault of the ball screw is the standard deviation of the vibration signal of the ball screw (indicating the vibration intensity), and the corresponding fault of the transmission mechanism is stuck. The fault monitoring quantity is the control board current of the motor, and the fault monitoring quantity of the transmission mechanism gap is too large is the output displacement/rudder surface deflection angle of the electromechanical actuator. After determining the fault monitoring quantity of each failure mode, according to the deviation between the fault monitoring quantity and the reference value, the failure degree of the corresponding failure mode is obtained, and the key component layer of the upper layer is mapped to the calculation formula of the failure degree of the failure mode for:

Among them, u is the actual value of the fault monitoring quantity measured by the sensor on the electromechanical actuator to be evaluated, u _nom is the reference value of the fault monitoring quantity output through the regression model, and u _max is each fault monitoring quantity determined in step 201 The failure threshold at the time of failure, that is, the value at which the deviation from the normal value is the largest.

For each key component of the electromechanical actuator in the key component layer, that is, the drive motor, ball screw and transmission mechanism, the different failure modes that may occur are comprehensively considered. For the drive motor, considering the possible occurrence of motor winding turns short circuit and motor bearing surface damage; for the ball screw, the surface damage of the ball screw may occur; for the transmission mechanism, the concept is relatively broad, considering the possible occurrence of the transmission in the entire transmission chain. For related failures, the embodiment of the present invention considers that the transmission gap may be too large or the transmission mechanism is stuck, and the failure degree of the component for each typical failure mode is used as the base of the subspace of the key component layer, and the Euclidean distance The evaluation function is mapped to obtain the fault degree of the component, which is output to the overall layer of the electromechanical actuator.

Among them, t _j is the failure degree of the j-th key component, x _i is the failure degree of the i-th failure mode in the j-th key component, and x _max-i is the corresponding value of the i-th failure mode in the j-th key component The maximum value of the failure degree is set to 1 here, and n represents the sum of the failure modes corresponding to the jth key component.

The overall layer of the electromechanical actuator divides the electromechanical actuator into three key components: the drive motor, the ball screw, and the transmission mechanism. The failure degree of the three key components is used as the base of the layer space, and is evaluated by the Euclidean distance Function mapping yields an overall health indicator for the electromechanical actuator.

Among them, y is the health index of the electromechanical actuator, t _j is the failure degree of the jth key component, and t _max-j is the maximum value of the failure degree of the jth key component, which is set to 1 here; The sum of the key components in the actuator is set to 3 here.

The advantages and positive effects of the embodiments of the present invention are:

First, the data-based reference model construction method does not need to establish a complex electromechanical actuator model, which avoids the interference caused by various nonlinear links and error terms.

Second, the proposed health status assessment method has high real-time performance and can reflect the health indicators of the electromechanical actuator in real time during its operation.

Third, the proposed health status assessment method, compared with the traditional feature extraction and directly using the classifier to classify the current data, can also pass the reference value and actual value of each channel when facing a fault that does not contain prior information. The larger gap before reflects that the system is currently in a more dangerous state.

Fourth, the health assessment of electromechanical actuators not only includes the assessment of the overall health of the system, but also includes the analysis of each key component in the system and the failure degree of typical failure modes, which can better reflect the mechanical and electrical actuators. the actual operating condition of the actuator.

Embodiment three

Since there is no experimental platform for electromechanical actuators temporarily, the method provided by the embodiment of the present invention is implemented based on NASA's open source electromechanical actuator experimental data set, and the design of a method for monitoring the health status of electromechanical actuators based on composite space is as follows:

(1) Analysis of typical fault characteristics of electromechanical actuators

Firstly, the typical faults in the operation of the electromechanical actuator are analyzed. For the motor winding turn-to-turn short circuit fault, the three-phase motor is no longer symmetrical, resulting in the three-phase current is no longer symmetrical; for the motor bearing surface damage fault and the ball screw surface damage Fault, every time the damage point contacts the surface of the component, a sudden impact pulse force will be generated, resulting in an increase in the amplitude of the characteristic frequency of the vibration signal; for faults with excessive transmission gap, the output displacement will be affected by the gap and produce deviations ; For the stuck fault of the transmission mechanism, the stuck torque will lead to an increase in the current of the control board of the motor. According to this, the fault monitoring quantity of each typical fault is determined. The fault monitoring quantity corresponding to the inter-turn short circuit fault of the motor winding is the difference between the three-phase currents of the motor, and the fault monitoring quantity corresponding to the motor bearing surface damage fault is the vibration at the characteristic frequency of the motor bearing fault. Amplitude, the fault monitoring quantity corresponding to the surface damage fault of the ball screw is the standard deviation of the vibration signal of the ball screw (indicating the vibration intensity), the fault monitoring value of the transmission mechanism stuck fault is the control board current of the motor, and the fault of the transmission clearance fault The monitoring quantity is the output displacement/rudder surface deflection angle of the electromechanical actuator.

(2) Construct a complex health state space

Based on the above analysis, the complex health state space of the electromechanical actuator is constructed as shown in Figure 3, and the overall state of the electromechanical actuator is evaluated through the three-layer space of the electromechanical actuator-key components-failure mode, that is, HI _EMA .

(3) Training electromechanical actuator reference model

The method of the present invention is used to evaluate the open-source electromechanical actuator experimental data set of NASA, and train the reference model of the electromechanical actuator through the electromechanical actuator experimental data in a normal state, so as to obtain the reference value of the fault monitoring quantity at the next moment. Due to the flaws in the data collected by the data set itself, the vibration signal of the bearing and the three-phase current of the motor are not included. The training results of the regression model of the three-channel signals of output displacement, control board current and vibration signal standard deviation are shown in Figure 4-6.

(4) Assess the health status of electromechanical actuators

Due to the defects of the data collected by the data set itself, the vibration signal of the bearing and the three-phase current of the motor are not included, so the surface damage fault of the motor bearing and the short circuit fault of the inter-turn winding of the motor are not considered. The reference value of the fault monitoring quantity is obtained through the trained regression model, and the health status assessment is performed through the composite space.

The evaluation objects include the electromechanical actuator in normal state, the electromechanical actuator injected into the fault state of transmission mechanism jamming, and the electromechanical actuator injected into the fault state of the surface damage of the ball screw. The evaluation results are shown in Figure 7-9 shown. It can be observed that for the electromechanical actuator in the normal state, the actual value of each fault monitoring quantity is very close to the reference value; and for the electromechanical actuator injected into the transmission mechanism stuck fault state, it can be observed that there is a significant The standard deviation of the control board current of the motor with deviation and the vibration signal of the ball screw with small deviation; and for the electromechanical actuator injected into the fault state of the surface damage of the ball screw, the vibration of the ball screw with obvious deviation can be observed The standard deviation of the signal. By adopting the method of the invention, the health state evaluation of the electromechanical actuator under different states is realized.

Embodiment Four

As shown in Figure 10, a composite space-based electromechanical actuator health status monitoring system provided by an embodiment of the present invention is used to monitor the health status of a target electromechanical actuator; the target electromechanical actuator includes a plurality of target components , and each of the target components corresponds to one or more failure modes, and each of the failure modes corresponds to a fault monitoring quantity; the composite health state space of the target electromechanical actuator includes a bottom layer, a middle layer and a top layer; the The bottom layer includes all failure modes of the target electromechanical actuator; the middle layer includes all target components of the target electromechanical actuator; the top layer is the target electromechanical actuator; the electromechanical actuator is healthy Condition monitoring system, including:

The data acquisition module 1001 is configured to acquire the operating data of the target electromechanical actuator at the last moment, and acquire the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment.

The fault monitoring quantity reference value prediction module 1002 is configured to predict the fault monitoring quantity reference value of the target electromechanical actuator at the current moment based on the operating data of the target electromechanical actuator at the previous moment and the machine learning algorithm.

The electromechanical actuator health state determination module 1003 is configured to adopt a step-by-step mapping method based on the current moment reference value of the target electromechanical actuator’s fault monitoring quantity and the current moment’s actual value of the target electromechanical actuator’s fault monitoring quantity Monitoring the fault degree of each layer in the composite health state space of the target electromechanical actuator to obtain the health state of the target electromechanical actuator; the health state includes each The failure degree of the failure mode, the failure degree of each target component in the target electromechanical actuator, and the overall failure degree of the target electromechanical actuator.

On the basis of the embodiment described in FIG. 10 , the system provided by this embodiment further includes: a regression model construction module, configured to construct a regression model based on a long-short-term memory neural network.

On the basis of the embodiment described in FIG. 10 , the fault monitoring quantity reference value prediction module 1002 specifically includes:

A preprocessing unit is used to preprocess the operating data of the target electromechanical actuator at the previous moment; the preprocessing includes data splicing, normalization and sliding window cutting.

A fault monitoring quantity reference value prediction unit, configured to predict the fault monitoring quantity reference value of the target electromechanical actuator at the current moment based on the preprocessed operating data of the target electromechanical actuator at the previous moment and the regression model .

On the basis of the embodiment described in FIG. 10 , the electromechanical actuator health status determination module 1003 specifically includes:

The failure degree calculation unit of the failure mode is used to calculate each failure mode based on the reference value of the failure monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the failure monitoring quantity of the target electromechanical actuator at the current moment degree of failure.

A failure degree calculation unit of the target component, configured to determine the relationship between the failure mode and the target component based on the composite health state space of the target electromechanical actuator, and based on the relationship between the failure mode and the target component , and the failure degree of each of the failure modes, determining the failure degree of each of the target components.

The overall failure degree determination unit of the target electromechanical actuator is configured to determine the overall failure degree of the target electromechanical actuator based on the failure degree of each of the target components.

The present invention provides a method and system for evaluating the health state of electromechanical actuators based on composite space, which provides an electromechanical actuator with high real-time performance, strong versatility, and comprehensive analysis of each failure mode of the system, each component to A method and system for assessing the health status of electromechanical actuators for comprehensive and complete quantitative assessment. The method and system construct a composite health state space of the electromechanical actuator by analyzing typical failure modes of the electromechanical actuator and extracting corresponding fault features. By conducting experiments on completely normal electromechanical actuators, collecting their experimental data in each channel, and constructing a regression model based on long-term and short-term memory neural networks, as a reference model of electromechanical actuators, the faults of each channel can be calculated through input data The reference value of the monitoring quantity when the electromechanical actuator is not faulty. On this basis, the reference value is compared with the actual measurement value, and the fault degree of each layer is evaluated layer by layer according to the level-by-level mapping method, and finally the overall health index of the electric actuator system is calculated. The method of the present invention realizes the real-time health status evaluation of electromechanical actuators. Based on the constructed composite health state space, it can evaluate the health status of electromechanical actuators from faults to components to the whole level of electromechanical actuators. Decisions provide information. In practical application, it is necessary to obtain the reference value of the fault monitoring quantity, so based on the LSTM neural network, the reference model of the electromechanical actuator is trained through the actual experimental data, and the reference value of each fault monitoring quantity is predicted at the next moment.

Each embodiment in this specification is described in a progressive manner, each embodiment focuses on the difference from other embodiments, and the same and similar parts of each embodiment can be referred to each other. As for the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and for the related information, please refer to the description of the method part.

In this paper, specific examples have been used to illustrate the principle and implementation of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; meanwhile, for those of ordinary skill in the art, according to the present invention Thoughts, there will be changes in specific implementation methods and application ranges. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims (10)

1. A method for monitoring the state of health of an electromechanical actuator based on a composite space, characterized in that, the method for monitoring the state of health of an electromechanical actuator is used to monitor the state of health of a target electromechanical actuator; the target electromechanical actuator It includes a plurality of target components, and each of the target components corresponds to one or more failure modes, and each of the failure modes corresponds to a fault monitoring quantity; the composite health state space of the target electromechanical actuator includes the bottom layer and the middle layer and a top layer; the bottom layer includes all failure modes of the target electromechanical actuator; the middle layer includes all target components of the target electromechanical actuator; the top layer is the target electromechanical actuator; the A method for monitoring the health status of an electromechanical actuator, including: Obtaining the operating data of the target electromechanical actuator at the previous moment, and obtaining the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment; Based on the operating data of the target electromechanical actuator at the previous moment and the machine learning algorithm, predict the fault monitoring quantity reference value of the target electromechanical actuator at the current moment; Based on the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment, the composite health state of the target electromechanical actuator is calculated by using a step-by-step mapping method The fault degree of each layer in the space is monitored to obtain the health status of the target electromechanical actuator; the health status includes the fault degree of each failure mode in the target electromechanical actuator, the target electromechanical actuator The failure degree of each target component in the actuator and the overall failure degree of the target electromechanical actuator. 2. A kind of electromechanical actuator health state monitoring method based on composite space according to claim 1, is characterized in that, also comprises: Perform fault feature analysis on each of the fault modes in the target electromechanical actuator, and determine a fault monitoring quantity used for monitoring the fault modes. 3. A method for monitoring the health status of an electromechanical actuator based on a composite space according to claim 1, further comprising: constructing a regression model based on a long-short-term memory neural network. 4. a kind of electromechanical actuator health status monitoring method based on composite space according to claim 3, is characterized in that, described construction is based on the regression model of long short-term memory neural network, specifically comprises: Build a regression neural network; the regression neural network includes a long short-term memory neural network layer and a fully connected neural network layer; Constructing a training sample set; the training sample set includes a plurality of sample input data and label data corresponding to each of the sample input data; the sample input data is the operating data of the target electromechanical actuator in a normal state; The tag data is the actual value of the fault monitoring quantity of the target electromechanical actuator under normal conditions; The regression neural network is trained based on the training sample set to obtain a regression model. 5. A method for monitoring the health state of an electromechanical actuator based on a compound space according to claim 3, wherein, based on the operating data and machine learning algorithms of the target electromechanical actuator at the last moment, predicting The reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment includes: Preprocessing the operating data of the target electromechanical actuator at the previous moment; the preprocessing includes data splicing operations, normalization operations, and sliding window cutting operations; Based on the preprocessed operating data of the target electromechanical actuator at the previous moment and the regression model, predict the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment. 6. A kind of electromechanical actuator health state monitoring method based on compound space according to claim 1, it is characterized in that, the fault monitoring quantity reference value based on the target electromechanical actuator at the current moment and the current moment The actual value of the fault monitoring quantity of the target electromechanical actuator, and the level-by-level mapping method is used to monitor the fault degree of each layer in the composite health state space of the target electromechanical actuator to obtain the target electromechanical actuator. Health status, including: calculating the fault degree of each of the fault modes based on the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment; Based on the composite health state space of the target electromechanical actuator, determine the relationship between the failure mode and the target component, and based on the relationship between the failure mode and the target component, and the failure of each of the failure modes degree, determining the degree of failure of each of said target components; An overall failure degree of the target electromechanical actuator is determined based on the failure degree of each of the target components. 7. An electromechanical actuator health status monitoring system based on composite space, characterized in that, the electromechanical actuator health status monitoring system is used to monitor the health status of the target electromechanical actuator; the target electromechanical actuator It includes a plurality of target components, and each of the target components corresponds to one or more failure modes, and each of the failure modes corresponds to a fault monitoring quantity; the composite health state space of the target electromechanical actuator includes the bottom layer and the middle layer and a top layer; the bottom layer includes all failure modes of the target electromechanical actuator; the middle layer includes all target components of the target electromechanical actuator; the top layer is the target electromechanical actuator; the Electromechanical actuator health status monitoring system, including: A data acquisition module, configured to acquire the operating data of the target electromechanical actuator at the previous moment, and acquire the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment; The fault monitoring quantity reference value prediction module is used to predict the fault monitoring quantity reference value of the target electromechanical actuator at the current moment based on the operating data of the target electromechanical actuator at the previous moment and the machine learning algorithm; The electromechanical actuator health status determination module is configured to use a step-by-step mapping method based on the reference value of the fault monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the fault monitoring quantity of the target electromechanical actuator at the current moment. The fault degree of each layer in the composite health state space of the target electromechanical actuator is monitored to obtain the health state of the target electromechanical actuator; the health state includes each of the target electromechanical actuators The failure degree of the failure mode, the failure degree of each target component in the target electromechanical actuator, and the overall failure degree of the target electromechanical actuator. 8 . The composite space-based electromechanical actuator health status monitoring system according to claim 7 , further comprising: a regression model building module for building a regression model based on a long-short-term memory neural network. 9. A kind of electromechanical actuator health state monitoring system based on compound space according to claim 8, is characterized in that, described fault monitoring quantity reference value prediction module, specifically comprises: A preprocessing unit, configured to preprocess the operating data of the target electromechanical actuator at the previous moment; the preprocessing includes data splicing operations, normalization operations, and sliding window cutting operations; A fault monitoring quantity reference value prediction unit, configured to predict the fault monitoring quantity reference value of the target electromechanical actuator at the current moment based on the preprocessed operating data of the target electromechanical actuator at the previous moment and the regression model . 10. A composite space-based electromechanical actuator health status monitoring system according to claim 7, wherein the electromechanical actuator health status determination module specifically includes: The failure degree calculation unit of the failure mode is used to calculate each failure mode based on the reference value of the failure monitoring quantity of the target electromechanical actuator at the current moment and the actual value of the failure monitoring quantity of the target electromechanical actuator at the current moment the degree of failure; A failure degree calculation unit of the target component, configured to determine the relationship between the failure mode and the target component based on the composite health state space of the target electromechanical actuator, and based on the relationship between the failure mode and the target component , and the failure degree of each of the failure modes, determining the failure degree of each of the target components; The overall failure degree determination unit of the target electromechanical actuator is configured to determine the overall failure degree of the target electromechanical actuator based on the failure degree of each of the target components.

Images