A data-driven servo actuator system model identification and fault monitoring method

Abstract

The present invention belongs to the field of aircraft servo actuation systems, and relates to a data-driven servo actuation system model identification and fault monitoring method, including step 1, mathematical modeling of an aircraft steering gear electro-hydraulic servo actuation system; step 2, using the BP neural network method, using the collected test flight data, obtaining the input and output data of the aircraft steering gear electro-hydraulic servo actuation system through mathematical calculation of the aircraft steering gear electro-hydraulic servo actuation system, identifying the aircraft steering gear electro-hydraulic servo actuation system, and obtaining a neural network model of the aircraft steering gear electro-hydraulic servo actuation system with more accurate output; step 3, designing a corresponding servo actuation system model monitoring module, implementing real-time monitoring, and using the established dynamic model to continuously evaluate the system status. The present invention can achieve more accurate monitoring and fault detection, timely identify potential problems, reduce the risk of system failures, and can adapt to the dynamic changes of the system in real time, thereby improving overall safety and reliability.

Description

Technical Field

[0001] The invention belongs to the field of aircraft servo actuation systems, and relates to a data-driven servo actuation system model identification and fault monitoring method. Background Art

[0002] Servo actuators are widely used in a variety of fields such as aerospace, automated production, robotics, automotive control, and medical equipment. These systems ensure efficient operation of equipment by precisely controlling motion and position. For example, in the aerospace field, servo actuators are used to control the control surfaces and propulsion systems of aircraft to achieve precise flight attitude control. In automated production, they are used for precise positioning and motion control of robotic arms, thereby improving production efficiency and product quality. The reliability and accuracy of servo actuators directly affect the performance and safety of the entire system, so its fault detection and model monitoring are particularly important.

[0003] Although there are many fault detection and model monitoring methods, there are still some shortcomings in the existing technologies. Traditional fault detection methods often rely on empirical rules, which are prone to missed detections and false alarms. In addition, existing model monitoring methods usually use static models and cannot effectively cope with the dynamic changes of the system under different working conditions. This limits the fault detection and monitoring capabilities of the system in complex environments, especially under high load or extreme conditions, and increases potential risks.

[0004] With the development of data acquisition technology and the popularization of big data analysis tools, data-driven methods are gradually emerging in the engineering field. These methods can more accurately reflect the status and operating characteristics of the system by analyzing large amounts of real-time data. Data-driven fault detection and model monitoring methods can not only improve the accuracy of detection, but also adapt to the dynamic changes of the system in real time. For example, using machine learning algorithms, features can be extracted from historical test flight data to establish a more accurate dynamic model, thereby improving the sensitivity and reliability of fault detection. This trend has shown good prospects and effects in applications in aerospace, industrial automation and other fields. Summary of the invention

[0005] The present invention aims to provide a data-driven servo actuator system model identification and fault monitoring method. This method makes full use of real-time data obtained during the test flight, including sensor data, control signals and system status information. These data are real, dynamic and diverse, and can reflect the real performance of the system in actual operation. By modeling these data using neural network methods and establishing a high-precision mathematical model, the fault detection capability and model monitoring effect of the servo actuator system can be improved, and more accurate monitoring and fault detection can be achieved, potential problems can be identified in time, the risk of system failure can be reduced, and overall safety and reliability can be improved. The innovation lies in the use of data-driven methods, combined with neural network identification technology, to design a model monitoring framework that can adapt to the dynamic changes of the system in real time, overcome the shortcomings of traditional methods, and reduce false alarm rates and missed detection rates.

[0006] The technical solution of the present invention:

[0007] A data-driven servo actuator system model identification and fault monitoring method comprises the following steps:

[0008] Step (1) Mathematical modeling of aircraft steering gear electro-hydraulic servo actuation system

[0009] (1.1) Basic equations of aircraft servo electro-hydraulic servo actuation system

[0010] The aircraft steering gear electro-hydraulic servo actuation system mainly consists of two parts: the electro-hydraulic servo valve and the hydraulic servo actuator.

[0011] (1.1.1) Transfer function equation of electro-hydraulic servo valve

[0012] The electro-hydraulic servo valve controls the hydraulic flow through electrical signals to achieve precise control of the hydraulic system.

[0013] The valve core displacement can be obtained by the driving current. The relationship between the driving current and the valve core displacement can be regarded as a proportional link. The transfer function is:

[0014] (1)

[0015] In the formula, is the valve core displacement; is the current; is the current-spool displacement gain.

[0016] (1.1.2) Sliding valve flow equation

[0017] (2)

[0018] In the formula, is the flow rate of oil in the slide valve, is the valve core displacement, is the pressure of the oil in the valve core, is the flow gain; is the flow-pressure coefficient.

[0019] The dynamic analysis of the position servo system is often carried out under zero-position working conditions, when the increment and variable are equal. Due to the influence of the external leakage and compressibility of the hydraulic servo actuator, the flow rate into the oil inlet chamber of the hydraulic servo actuator is The flow rate out of the hydraulic servo actuator return oil chamber Not equal, define load flow for

[0020] (3)

[0021] (1.1.3) Flow continuity equation for hydraulic servo actuator

[0022] It is usually assumed that: the connecting pipe between the valve and the hydraulic servo actuator is symmetrical, short and thick, the pressure loss and pipeline dynamics in the pipe can be ignored; the pressure at each location in each working chamber of the hydraulic servo actuator is equal, the oil temperature and bulk elastic modulus are constants; the internal and external leakage of the hydraulic servo actuator are laminar flows.

[0023] Flow rate into the oil inlet chamber of the hydraulic servo actuator for

[0024] (4)

[0025] In the formula, is the pressure of the oil in the oil inlet chamber, It is the oil pressure in the oil return chamber.

[0026] Flow rate into the return oil chamber of the hydraulic servo actuator for

[0027] (5)

[0028] In the formula, is the displacement of the hydraulic servo actuator; is the cross-sectional area of ​​the hydraulic servo actuator; is the internal leakage coefficient of the hydraulic servo actuator; is the external leakage coefficient of the hydraulic servo actuator; is the volume of the oil inlet chamber of the hydraulic servo actuator; The volume of the oil return chamber of the hydraulic servo actuator: is the effective bulk elastic modulus.

[0029] In formula (4) and formula (5), is the flow rate required to drive the servo actuator. is the internal leakage flow of the oil inlet chamber and oil return chamber of the servo actuator. , are the external leakage flow of the oil inlet chamber and the oil return chamber of the servo actuator, and The flow rates required for the oil inlet chamber and the oil return chamber due to oil compression and deformation in the chamber are called the oil inlet chamber compression flow rate and the oil return chamber compression flow rate.

[0030] The volume of the oil inlet chamber and oil return chamber of the hydraulic servo actuator is

[0031] (6)

[0032] (7)

[0033] In the formula, is the initial volume of the oil inlet chamber; is the initial volume of the oil return chamber;

[0034] From equations (3), (4), (5), (6) and (7), the flow continuity equation is:

[0035] (8)

[0036] In the formula, the external leakage flow of the oil inlet chamber and the oil return chamber is and Usually very small and can be ignored. and oil return chamber compression flow If they are equal, Because the valve is matched symmetrically, the flow through the oil inlet valve port and the oil return valve port is also equal. Therefore, the total pressure in dynamic state is The difference between the oil inlet pressure and the oil return pressure, i.e. the pressure drop ,so , . Thus,

[0037] (9)

[0038] To make the compression flow equal, the initial volume of the two chambers of the hydraulic servo actuator should be and Equal, that is

[0039] (10)

[0040] In the formula, It is the average volume of the oil inlet chamber and the oil return chamber of the hydraulic servo actuator; is the total compressed volume.

[0041] because , , then formula (8) can be simplified to

[0042] (11)

[0043] In the formula, is the total leakage coefficient of the hydraulic servo actuator, .

[0044] In formula (11), It is the flow rate required to drive the hydraulic servo actuator. is the total leakage flow, is the total compressed flow.

[0045] (1.1.4) Load balance equation of hydraulic servo actuator

[0046] The dynamic characteristics of the hydraulic servo actuator power element are affected by the load characteristics. The load force generally includes inertia force, viscous damping force, elastic force and any external load force.

[0047] The balance equation between the output force and load force of the hydraulic servo actuator is:

[0048] (12)

[0049] In the formula, is the output force of the hydraulic servo actuator, It is the total mass of the hydraulic servo actuator and the load converted to the actuator; is the viscous damping coefficient of the hydraulic servo actuator and the load; is the elastic stiffness of the load; It is the external load force acting on the hydraulic servo actuator.

[0050] In addition, there are nonlinear loads caused by Coulomb friction, etc., but compared with the external load torque, the influence of these factors is very small. When using linearization methods to analyze the dynamic characteristics of the system, these nonlinear loads can be ignored.

[0051] (1.2) Transfer function derivation

[0052] Taking the Lagrange transformation of equation (2), equation (11) and equation (12), we can get

[0053] (13)

[0054] (14)

[0055] (15)

[0056] In formula (14) and formula (15), is a complex variable, usually expressed as ,in is the real part, is the imaginary part, is an imaginary unit, which can be regarded as a differential operator here. Differentiation in the time domain corresponds to multiplication in the complex frequency domain. In formula (13), (14), (15) , , , In the time domain , , , frequency domain form.

[0057] Equations (13), (14), and (15) fully describe the dynamic characteristics of the servo actuator system. Eliminating intermediate variables and , the valve core displacement can be obtained and external load When acting simultaneously, the total output displacement of the servo actuator It is expressed as:

[0058] (16)

[0059] In the formula, is the total flow pressure coefficient. Valve core displacement is the input quantity of the system, the external load torque is the disturbance quantity, and the external disturbance quantity is also an input quantity, but its effect is opposite to that of the input quantity. From formula (16), we can see the response characteristics of the hydraulic servo actuator to the displacement of the valve and the external load force input.

[0060] (1.3) Transfer function simplification

[0061] In formula (16), factors such as inertial load, viscous friction load, elastic load, oil compressibility, and leakage of hydraulic servo actuator are taken into account. The load of the actual system may be relatively simple and can be ignored, further simplifying the transfer function.

[0062] In most cases, the servo actuator system is mainly based on inertial load, so the elastic load and the influence of viscous damping can be ignored, that is,

[0063] (17)

[0064] In the formula, is the natural frequency of hydraulic pressure; is the hydraulic damping ratio.

[0065] Usually the viscosity coefficient Very small, so .

[0066] Displacement of hydraulic servo actuator The transfer function of the valve core displacement is:

[0067] (18)

[0068] Step (2) Neural network identification method of aircraft servo actuation system model based on test flight data

[0069] In order to obtain an accurate model of the aircraft steering gear electro-hydraulic servo actuation system, the BP neural network method is adopted. The test flight data collected by the aircraft steering gear electro-hydraulic servo actuation system sensor is used. The input and output data of the aircraft steering gear electro-hydraulic servo actuation system are obtained through the mathematical calculation of the aircraft steering gear electro-hydraulic servo actuation system established in step (1). The aircraft steering gear electro-hydraulic servo actuation system is identified to obtain a neural network model of the aircraft steering gear electro-hydraulic servo actuation system with more accurate output. The specific process is as follows:

[0070] (2.1) Flight test data collection

[0071] Data collection is usually carried out in multiple flight phases, including take-off, cruise, maneuvering and landing, to ensure that all possible working conditions are covered.

[0072] The test flight data of the aircraft steering gear electro-hydraulic servo actuation system mainly include the following:

[0073] Aircraft steering gear control instructions, electro-hydraulic servo valve control instructions, electro-hydraulic servo valve current collection value, electro-hydraulic servo valve voltage collection value, electro-hydraulic servo valve spool displacement, and actuator-controlled rudder surface deflection angle.

[0074] (2.2) Data processing

[0075] After filtering and resampling the valve core output signal of the electro-hydraulic servo valve, the output displacement signal of the actuator and other data, neural network identification training is performed.

[0076] The filtering method adopted is moving average filtering. Moving average filtering is a commonly used signal processing technology, especially playing an important role in smoothing and noise suppression of time series data.

[0077] After the data passes through a moving average filter, a resampling operation is performed to reduce the error caused by the reduction in frequency; then a moving average filter is performed again to suppress the jitter of the data and reduce the error.

[0078] (2.3) Neural network model construction and identification

[0079] Model: Aircraft steering gear electro-hydraulic servo actuation system model.

[0080] The model is divided into two parts, namely the electro-hydraulic servo valve neural network model and the actuator neural network model.

[0081] The input layer of the electro-hydraulic servo valve neural network model is: rudder control instructions and electro-hydraulic servo valve instructions, and the output layer is the valve core displacement.

[0082] The input layer of the actuator neural network model is: rudder surface control instructions, electro-hydraulic servo valve instructions and electro-hydraulic servo valve core displacement, and the output layer is the rudder surface deflection angle output by the actuator.

[0083] The two neural network models are trained separately, and both use the back-propagation neural network algorithm of a two-layer feedforward network with hidden neurons with activation functions and linear output neurons.

[0084] Step (3) Design of dynamic monitoring structure of aircraft steering gear electro-hydraulic servo actuation system based on test flight data

[0085] For the electro-hydraulic servo valve neural network model and actuator neural network model obtained in step (2), design the corresponding servo actuator system model monitoring module, implement real-time monitoring, use the established dynamic model to continuously evaluate the system status, and set monitoring indicators and thresholds. The specific process is as follows:

[0086] (3.1) Model monitoring external loop structure design

[0087] The outer loop of the aircraft steering gear electro-hydraulic servo actuation system model monitoring module is designed for the aircraft electro-hydraulic servo actuation system model established in step (2). The input and output of the model are monitored and compared with the input and output data of the servo actuation system on the actual aircraft to achieve the function of monitoring the status of the servo actuation system during the flight of the aircraft. At the same time, the output data of the two are compared and analyzed to realize the fault detection of the aircraft servo actuation system.

[0088] The outer loop model monitoring is to monitor the entire aircraft steering gear electro-hydraulic servo actuation system model, that is, the loop input is the aircraft control command signal, and the output is the output displacement or deflection angle of the actuation system. The model in the outer loop corresponds to the actuator neural network model in step (2).

[0089] The fault judgment conditions of the mathematical model output in model monitoring and the data collected from the flight test data are as follows:

[0090] (19)

[0091] In the formula, Output of flight test data or output of real actuation system; It is the output of mathematical model in model monitoring; is the fault threshold. When this value is exceeded, it can be determined that the actual actuation system has a fault.

[0092] Formula (19) is a simple model monitoring method based on threshold and residual analysis. It does not consider the time required for the model monitoring system to calculate and the time consumed in the data transmission process. Therefore, it is necessary to consider the time difference between the outputs of the two. Therefore, on this basis, the time judgment condition is added as follows:

[0093] (20)

[0094] In the formula, For never being satisfied To meet time; To satisfy To not satisfied moment; is the time threshold. It means that when Too long, more than , it is considered to be faulty.

[0095] (3.2) Model monitoring inner loop structure design

[0096] The outer loop model monitoring is to monitor the entire servo actuation system model, and the inner loop model monitoring is to monitor each component in the servo actuation system. The inner loop can describe the intermediate process of the outer loop, adding a monitoring link for the intermediate data.

[0097] The inner loop model monitoring is divided into three parts: control electric signal model monitoring, electro-hydraulic servo valve model monitoring and actuator model monitoring. In the inner loop, the control electric signal module reference model is the coil resistance of the torque motor in the electro-hydraulic servo valve, and its rated value is taken as the reference model. The models in the electro-hydraulic servo valve model monitoring and the actuator model monitoring correspond to the electro-hydraulic servo valve neural network model and the actuator neural network model in step (2), and the input and output of the two are taken as the monitoring model.

[0098] Control electrical signal model monitoring object: control voltage and control current .

[0099] Fault judgment conditions for the coil resistance of the torque motor in the aircraft steering gear electro-hydraulic servo valve:

[0100] (twenty one)

[0101] In the formula, is the coil resistance of the torque motor, , is the rated value of the torque motor coil resistance, is the resistance fault threshold.

[0102] The monitoring objects of the electro-hydraulic servo valve model: electro-hydraulic servo valve control signal and valve core displacement.

[0103] Fault judgment conditions for electro-hydraulic servo valve spool displacement:

[0104] (twenty two)

[0105] In the formula, The valve core displacement collected by the aircraft servo sensor. is the valve core displacement output by the neural network model, is the valve core displacement fault threshold.

[0106] Actuator model monitoring object: the deflection angle of the rudder surface driven by the actuator.

[0107] Fault judgment conditions of actuator:

[0108] (twenty three)

[0109] In the formula, is the deflection angle of the rudder surface collected by the aircraft servo sensor. is the rudder deflection angle output by the neural network model, is the fault threshold of the rudder surface deflection angle. The relationship between the rudder surface deflection angle and the actuator output displacement is:

[0110] (twenty four)

[0111] In the formula, is the actuator displacement-angle proportionality coefficient.

[0112] The above monitoring object is built through the neural network model in step (2), which is compared with the test flight data and the residual situation between the two is analyzed to realize the fault monitoring function.

[0113] Beneficial effects of the present invention:

[0114] The present invention firstly identifies the neural model of the electro-hydraulic servo actuation system of an aircraft steering gear by analyzing a large amount of real-time data on the basis of the traditional servo actuation system model, which can more accurately reflect the state and operation characteristics of the system, making the model more accurate and closer to the real system. On this basis, a monitoring module of the electro-hydraulic servo actuation system model of an aircraft steering gear is established, and a fault diagnosis algorithm is designed. By comparing the output of the neural network model, the output state of the real system can be fed back in time, and the fault diagnosis of the electro-hydraulic servo actuation system of the real aircraft steering gear can be realized, thereby improving the sensitivity and reliability of fault detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0115] Figure 1 It is the overall block diagram of data-driven servo actuation system model monitoring;

[0116] Figure 2 It is a schematic block diagram of the servo actuator system model;

[0117] Figure 3 It is a flowchart of the method for reducing the error of flight test data;

[0118] Figure 4 This is a simulation diagram of the flight test data before and after the error is reduced;

[0119] Figure 5 It is the monitoring block diagram of the inner loop current model of the servo actuator system;

[0120] Figure 6 It is the monitoring block diagram of the electro-hydraulic servo valve model in the inner ring of the servo actuation system;

[0121] Figure 7 It is the monitoring block diagram of the inner loop actuator model of the servo actuator system;

[0122] Figure 8 It is a monitoring simulation result diagram of the servo actuator system outer loop servo actuator system model;

[0123] Fig. 9 It is a partial enlarged view of the simulation results of the servo actuator system outer loop model monitoring;

[0124] Fig.10 It is the residual diagram of the simulation of the servo actuator system outer loop servo actuator system model monitoring;

[0125] Fig.11 It is the residual diagram of simulation output of the monitoring system of the inner loop current model of the servo actuator system;

[0126] Fig.12 It is the residual diagram of the monitoring simulation output of the inner loop servo valve model of the servo actuator system;

[0127] Fig.13 It is the residual diagram of the simulation output of the inner-loop actuator model monitoring of the servo actuator system;

[0128] Fig.14 It is a simulation diagram of a failure in the flight test data of the servo actuator system;

[0129] Fig.15 It is a simulation diagram of fault detection and residual comparison of the servo actuator system model. DETAILED DESCRIPTION

[0130] The specific implementation of the present invention is further described below in conjunction with the accompanying drawings and technical solutions.

[0131] Servo actuator system Figure 2 As shown in the figure, the main components are electro-hydraulic servo valve and servo actuator, the actuator type is the actuator cylinder, and the output is the output displacement of the actuator cylinder. The application scenario of this servo actuator system is the deflection control of the rudder surface of the aircraft wing. The working principle is that when the system receives the flight instructions from the pilot's control or the automatic control program, the instructions are input into the controller, and then converted into the control electrical signal of the servo valve through the controller. The servo valve receives the electrical signal (such as voltage or current signal) from the controller, which represents the required output position or speed, so that the valve core moves in the valve body, changes the opening of the fluid channel, and thus controls the fluid flow. The flow and direction changes of the fluid cause the connected actuator cylinder to produce corresponding movement. The servo valve can accurately control the flow, so that the movement of the actuator cylinder is smooth and the response is fast. At the same time, the servo valve is usually equipped with a feedback device (such as a position sensor) to monitor the position of the actuator cylinder in real time and send the feedback signal back to the control system to adjust the input signal to ensure the precise control of the system.

[0132] Based on the test flight data of the aircraft, the system is modeled and monitored by identifying and designing monitoring loops, such as Figure 1 As shown, the following is the specific implementation process:

[0133] (1) Set initial conditions and data and establish mathematical modeling of the aircraft servo actuator system

[0134] (1.1) Setting initial conditions and data

[0135] The test flight data is the data of the servo actuation system of the left horizontal tail during the whole process from takeoff to landing. The data categories are as follows in Table 1:

[0136] Table 1 Simulation flight test data

[0137]

[0138] The data length of each data variable in Table 1 is 464675.

[0139] The identified model needs corresponding test data for testing, so the test data in the model monitoring is set to 50,000.

[0140] The time interval of all the above data is consistent with the real system and is set to 0.0125s.

[0141] (1.2) Establish mathematical model of aircraft steering gear electro-hydraulic servo actuation system

[0142] The aircraft steering gear electro-hydraulic servo actuation system mainly consists of two parts: the electro-hydraulic servo valve and the hydraulic servo actuator.

[0143] (1.2.1) Transfer function equation of electro-hydraulic servo valve

[0144] The electro-hydraulic servo valve controls the hydraulic flow through electrical signals to achieve precise control of the hydraulic system.

[0145] The valve core displacement can be obtained by the driving current. The relationship between the driving current and the valve core displacement can be regarded as a proportional link. The transfer function is:

[0146] (1)

[0147] In the formula, is the valve core displacement; is the current; is the current-spool displacement gain.

[0148] (1.2.2) Sliding valve flow equation

[0149] (2)

[0150] In the formula, is the flow rate of oil in the slide valve, is the valve core displacement, is the pressure of the oil in the valve core, is the flow gain; is the flow-pressure coefficient.

[0151] The dynamic analysis of the position servo system is often carried out under zero-position working conditions, when the increment and variable are equal. Due to the influence of the external leakage and compressibility of the hydraulic servo actuator, the flow rate into the oil inlet chamber of the hydraulic servo actuator is The flow rate out of the hydraulic servo actuator return oil chamber Not equal, define load flow for

[0152] (3)

[0153] (1.2.3) Flow continuity equation for hydraulic servo actuator

[0154] Assumptions: The connecting pipe between the valve and the hydraulic servo actuator is symmetrical, short and thick, and the pressure loss and pipeline dynamics in the pipe can be ignored; the pressure in each working chamber of the hydraulic servo actuator is equal, and the oil temperature and bulk elastic modulus are constants; the internal and external leakage of the hydraulic servo actuator are laminar flows.

[0155] Flow rate into the oil inlet chamber of the hydraulic servo actuator for

[0156] (4)

[0157] In the formula is the pressure of the oil in the oil inlet chamber, It is the oil pressure in the oil return chamber.

[0158] Flow rate into the return oil chamber of the hydraulic servo actuator for

[0159] (5)

[0160] In the formula, is the displacement of the hydraulic servo actuator; is the cross-sectional area of ​​the hydraulic servo actuator; is the internal leakage coefficient of the hydraulic servo actuator; is the external leakage coefficient of the hydraulic servo actuator; It is the volume of the oil inlet chamber of the hydraulic servo actuator; The volume of the oil return chamber of the hydraulic servo actuator: is the effective bulk elastic modulus.

[0161] In formula (4) and formula (5), is the flow rate required to drive the servo actuator. is the internal leakage flow of the oil inlet chamber and oil return chamber of the servo actuator. , are the external leakage flow of the oil inlet chamber and the oil return chamber of the servo actuator, and The flow rates required for the oil inlet chamber and the oil return chamber due to oil compression and deformation in the chamber are called the oil inlet chamber compression flow rate and the oil return chamber compression flow rate.

[0162] The volume of the oil inlet chamber and oil return chamber of the hydraulic servo actuator is

[0163] (6)

[0164] (7)

[0165] In the formula, is the initial volume of the oil inlet chamber; is the initial volume of the oil return chamber;

[0166] From equations (3), (4), (5), (6) and (7), the flow continuity equation is:

[0167] (8)

[0168] In the formula, the external leakage flow of the oil inlet chamber and the oil return chamber is and Usually very small and can be ignored. and oil return chamber compression flow If they are equal, Because the valve is matched symmetrically, the flow through the oil inlet valve port and the oil return valve port is also equal. Therefore, the total pressure in dynamic state is The difference between the oil inlet pressure and the oil return pressure, i.e. the pressure drop ,so , . Thus,

[0169] (9)

[0170] To make the compression flow equal, the initial volume of the two chambers of the hydraulic servo actuator should be and Equal, that is

[0171] (10)

[0172] In the formula, It is the average volume of the oil inlet chamber and the oil return chamber of the hydraulic servo actuator; is the total compressed volume.

[0173] because , , then formula (8) can be simplified to

[0174] (11)

[0175] In the formula, is the total leakage coefficient of the hydraulic servo actuator, .

[0176] In formula (11), It is the flow rate required to drive the hydraulic servo actuator. is the total leakage flow, is the total compressed flow.

[0177] (1.2.4) Load balance equation of hydraulic servo actuator

[0178] The dynamic characteristics of the hydraulic servo actuator power element are affected by the load characteristics. The load force generally includes inertia force, viscous damping force, elastic force and any external load force.

[0179] The balance equation between the output force and load force of the hydraulic servo actuator is:

[0180] (12)

[0181] In the formula, is the output force of the hydraulic servo actuator, It is the total mass of the hydraulic servo actuator and the load converted to the actuator; is the viscous damping coefficient of the hydraulic servo actuator and the load; is the elastic stiffness of the load; It is the external load force acting on the hydraulic servo actuator.

[0182] In addition, there are nonlinear loads caused by Coulomb friction, etc., but compared with the external load torque, the influence of these factors is very small. When using linearization methods to analyze the dynamic characteristics of the system, these nonlinear loads can be ignored.

[0183] (1.3) Transfer function derivation

[0184] Taking the Lagrange transformation of equation (2), equation (11) and equation (12), we can get

[0185] (13)

[0186] (14)

[0187] (15)

[0188] In formula (14) and formula (15), is a complex variable, usually expressed as in is the real part, is the imaginary part, is an imaginary unit, which can be regarded as a differential operator here. Differentiation in the time domain corresponds to multiplication in the complex frequency domain. In formula (13), (14), (15) , , , In the time domain , , , frequency domain form.

[0189] Equations (13), (14), and (15) fully describe the dynamic characteristics of the servo actuator system. Eliminating intermediate variables and , the valve core displacement can be obtained and external load When acting simultaneously, the total output displacement of the servo actuator It is expressed as:

[0190] (16)

[0191] In the formula, is the total flow pressure coefficient. Valve core displacement is the input quantity of the system, the external load torque is the disturbance quantity, and the external disturbance quantity is also an input quantity, but its effect is opposite to that of the input quantity. From formula (16), we can see the response characteristics of the hydraulic servo actuator to the displacement of the valve and the external load force input.

[0192] (1.4) Transfer function simplification

[0193] In formula (16), factors such as inertial load, viscous friction load, elastic load, oil compressibility, and leakage of hydraulic servo actuator are taken into account. The load of the actual system may be relatively simple and can be ignored, further simplifying the transfer function.

[0194] In most cases, the servo actuator system is mainly inertial load, so the elastic load and the influence of viscous damping can be ignored, that is,

[0195] (17)

[0196] In the formula, is the natural frequency of hydraulic pressure; is the hydraulic damping ratio.

[0197] Usually the viscosity coefficient Very small, so .

[0198] Displacement of hydraulic servo actuator The transfer function of the valve core displacement is:

[0199] (18)

[0200] (2) Example of simulation of the neural network identification process of the servo actuator system based on test flight data

[0201] Using the neural network MATLAB toolbox, through the identification of test flight data, a more accurate mathematical model of the aircraft electro-hydraulic servo valve and actuator is obtained. The specific process is as follows:

[0202] (2.1) Flight test data collection

[0203] Data collection is usually carried out in multiple flight phases, including take-off, cruise, maneuvering and landing, to ensure that all possible working conditions are covered.

[0204] The flight test data of the aircraft steering gear electro-hydraulic servo actuation system mainly include the following five:

[0205] Aircraft steering gear control instructions, electro-hydraulic servo valve control instructions, electro-hydraulic servo valve current collection value, electro-hydraulic servo valve spool displacement, and actuator-controlled rudder surface deflection angle.

[0206] (2.2) Data processing

[0207] After analysis, it is found that the control signal input of all the collected flight test data is accurate data, but the output signal of the servo actuator system is affected by external interference and sampling frequency, resulting in errors, making the data inaccurate and showing signs of jitter, such as the valve core output signal of the electro-hydraulic servo valve and the output displacement signal of the actuator. Therefore, it is necessary to filter and resample the data before conducting neural network identification training.

[0208] The filtering method used this time is moving average filtering. Moving average filtering is a commonly used signal processing technology, especially playing an important role in smoothing and noise suppression of time series data.

[0209] Data error and interference processing process Figure 3 As shown in the figure, the data first passes through a moving average filter to filter out interference and fluctuations and smooth the curve, and then a resampling operation is performed, which can reduce the error caused by the reduction in frequency; finally, a moving average filter is performed again to suppress the jitter of the data and reduce the error.

[0210] Taking the servo actuator system output as an example, the processing results are as follows Figure 4 As shown in the figure, the solid line is the output data collected by the sensor, and the dotted line is the curve obtained after data filtering, resampling and filtering. It can be seen that the curve is smoother, the jitter is reduced, and the data is smoothed and noise is suppressed.

[0211] (2.3) Neural network model construction and identification

[0212] It is constructed using the MATLAB neural network toolbox (back propagation neural network), and the neural network model used is the BP neural network, which is a special feedforward neural network. It is a multi-layer feedforward neural network trained according to the error back propagation algorithm, with the square of the network error as the objective function, and the gradient descent method is used to obtain the minimum value of the objective function. It has excellent multi-dimensional function nonlinear mapping capabilities and the ability to classify arbitrarily complex patterns.

[0213] Model: Neural network model of aircraft steering gear electro-hydraulic servo actuation system.

[0214] The whole system consists of two parts, namely the electro-hydraulic servo valve and the actuator. The electro-hydraulic servo valve receives the control electrical signal and converts it into a mechanical motion signal of the valve core displacement; the valve core displacement controls the oil flow into the actuator, thereby controlling the output displacement of the actuator. Equations (1) and (18) can express the mathematical model of the electro-hydraulic servo valve and the actuator, that is, the entire aircraft steering gear electro-hydraulic servo actuation system.

[0215] Formula (1) is the relationship between the current signal and the displacement of the valve core of the electro-hydraulic servo valve, and formula (18) is the displacement of the valve core of the electro-hydraulic servo valve. The identification goal is to obtain accurate neural network models of the electro-hydraulic servo valve and the actuator neural network model, and then obtain accurate valve core displacement output and actuator control surface deflection angle output.

[0216] Generally, the electro-hydraulic servo valve and the actuator are identified separately, that is, the input of the electro-hydraulic servo valve is the control current signal, and the output is the valve core displacement, the input of the actuator is the valve core displacement, and the output is the output displacement of the actuator; or the two are identified as a whole, that is, the overall input is the electro-hydraulic servo valve control current signal, and the output is the actuator output displacement.

[0217] Identification purpose: Combine the relationship between the influence of the aircraft steering gear on the electro-hydraulic servo valve and the actuator, and obtain the valve core displacement of the electro-hydraulic servo valve and the output rudder deflection angle of the actuator at the same time. The aircraft steering gear electro-hydraulic servo actuation system model is divided into two parts: the electro-hydraulic servo valve and the actuator, and identified separately. When identifying the electro-hydraulic servo valve, the influence of the aircraft command is considered; when identifying the actuator, the influence of the electro-hydraulic servo valve and the aircraft command is considered.

[0218] The input layer of the electro-hydraulic servo valve model is: rudder control instructions, electro-hydraulic servo valve instructions, and the output layer is the valve core displacement.

[0219] The input layer of the actuator model is: rudder surface control instructions, electro-hydraulic servo valve instructions, electro-hydraulic servo valve core displacement, and the output layer is the rudder surface deflection angle output by the actuator.

[0220] A back-propagation neural network algorithm with two layers of feed-forward networks consisting of hidden neurons with activation functions and linear output neurons was used. The middle layer was set to 10 layers, with 70% of the training data and 30% of the validation test data.

[0221] Through training with MATLAB's neural network toolbox, the neural network model of the electro-hydraulic servo valve and the neural network model of the actuator are obtained.

[0222] (3) Dynamic monitoring design of servo actuator system based on flight test data

[0223] Based on the mathematical model obtained by the servo actuation system model identification of the flight test data, the corresponding servo actuation system model monitoring module is designed to implement real-time monitoring. The established dynamic model is used to continuously evaluate the system status and set monitoring indicators and thresholds. The specific process is as follows:

[0224] (3.1) Model monitoring external loop structure design

[0225] The external loop of the servo actuator system model monitoring module is mainly designed for the servo actuator system neural network model established in the second chapter. It monitors the input and output of the neural network model and compares it with the input and output data of the servo actuator system on the actual aircraft to achieve the function of monitoring the status of the servo actuator system during the flight of the aircraft. At the same time, the output data of the two are compared and analyzed to realize fault detection of the aircraft servo actuator system.

[0226] like Figure 1 The outer loop shown is a loop composed of the neural network model of the entire aircraft steering gear electro-hydraulic servo actuation system obtained by identification, and the control signals and output signals at both ends. Model monitoring can be a part of the aircraft flight controller. It receives the same control instructions as the aircraft at the same time. After mathematical model calculation, it obtains the ideal output result, which is then compared with the actual output of the actuation system collected by the sensor in the aircraft actuation system. By analyzing the difference between the two, it is determined whether the aircraft actuation system is in a normal operating state.

[0227] The fault judgment conditions of the mathematical model output in model monitoring and the data collected from the flight test data are as follows:

[0228] (19)

[0229] In the formula, Output of flight test data or output of real actuation system; It is the output of mathematical model in model monitoring; is the fault threshold. When this value is exceeded, it can be determined that the actual actuation system has a fault.

[0230] Formula (19) is a simple model monitoring method based on threshold and residual analysis. It does not consider the time required for the model monitoring system to calculate and the time consumed in the data transmission process. Therefore, it is necessary to consider the time difference between the outputs of the two. Therefore, on this basis, the time judgment condition is added as follows:

[0231] (20)

[0232] In the formula, For never being satisfied To meet time; To satisfy To not satisfied moment; is the time threshold. It means that when Too long, more than , it is considered to be faulty.

[0233] At the same time, the model monitoring module can also be a way for the ground to analyze offline flight data. After the aircraft completes a round of test flights, the exported test flight data is put into the model monitoring module for calculation and the output results are compared to help analyze the various conditions that may occur in the aircraft actuation system during the entire test flight. At the same time, the test flight data can be used to further train the mathematical model in the model monitoring to make it closer to the real physical system model.

[0234] (3.2) Model monitoring inner loop structure design

[0235] The outer loop model monitoring is to monitor the entire servo actuation system model, while the monitoring function of each part in the servo actuation system cannot be realized, so it is necessary to design the inner loop model monitoring to realize the monitoring of each component in the servo actuation system. The outer loop model monitoring is to monitor the entire servo actuation system model, and the inner loop model monitoring is to realize the monitoring of each component in the servo actuation system. The inner loop can describe the intermediate process of the outer loop, adding the monitoring link of the intermediate data.

[0236] The inner loop model monitoring is divided into three parts: control electric signal model monitoring, electro-hydraulic servo valve model monitoring and actuator model monitoring. In the inner loop, the control electric signal module reference model is the coil resistance of the torque motor in the electro-hydraulic servo valve, and its rated value is taken as the reference model. The models in the electro-hydraulic servo valve model monitoring and actuator model monitoring are the electro-hydraulic servo valve neural network model and the actuator neural network model, and the input and output of the two are taken as the monitoring reference model.

[0237] Control electrical signal model monitoring object: control voltage and control current .

[0238] Fault judgment conditions for the coil resistance of the torque motor in the aircraft steering gear electro-hydraulic servo valve:

[0239] (twenty one)

[0240] In the formula, is the coil resistance of the torque motor, , is the rated value of the torque motor coil resistance, is the resistance fault threshold.

[0241] The monitoring objects of the electro-hydraulic servo valve model: electro-hydraulic servo valve control signal and valve core displacement.

[0242] Fault judgment conditions for electro-hydraulic servo valve spool displacement:

[0243] (twenty two)

[0244] In the formula, The valve core displacement collected by the aircraft servo sensor. is the valve core displacement output by the neural network model, is the valve core displacement fault threshold.

[0245] Actuator model monitoring object: the deflection angle of the rudder surface driven by the actuator.

[0246] Fault judgment conditions of actuator:

[0247] (twenty three)

[0248] In the formula, is the deflection angle of the rudder surface collected by the aircraft servo sensor. is the rudder deflection angle output by the neural network model, is the fault threshold of the rudder surface deflection angle. The relationship between the rudder surface deflection angle and the actuator output displacement is:

[0249] (twenty four)

[0250] In the formula, is the actuator displacement-angle proportionality coefficient.

[0251] The above-mentioned monitoring object is built through a neural network model, which is compared with the test flight data, and the residual situation between the two is analyzed to realize the fault monitoring function.

[0252] like Figure 1 The inner loop model monitoring shown is divided into three parts, namely: control electrical signal model monitoring, electro-hydraulic servo valve model monitoring and actuator model monitoring.

[0253] like Figure 5 The control electrical signal model monitoring shown is responsible for monitoring the electrical input signal of the servo actuator system, and the electrical signal includes control voltage and control current.

[0254] like Figure 6 The electro-hydraulic servo valve model monitoring module shown is responsible for monitoring the input and output signals at both ends of the electro-hydraulic servo valve.

[0255] like Figure 7 The actuator model monitoring module shown is responsible for monitoring the input and output signals at both ends of the actuator.

[0256] The data of each of the above modules can be measured by sensors at both ends of each component. The data of the sensors can be compared with the output data of the mathematical models of each component that can be identified, and the interpolation and transformation between the mathematical model and the actual test flight data can be analyzed to monitor each component.

[0257] The simulation results are as follows Figure 8~Figure 10 As shown, Figure 8 Monitor the simulation output of the servo actuator system model during the entire test flight. Fig. 9 for Figure 8 The local enlarged image shows that the output curves of the identification model and the real system (flight test data) are basically consistent, indicating that the model obtained based on historical flight test data has a high degree of accuracy. At the same time, the output of the real system (flight test data) is consistent with the output of the identification model, reflecting that there is no fault in the real system at present.

[0258] The system residual result is as follows Fig.10 As shown in Figure 2, the residuals of the identification system output results and the test flight data are within 0.05°; Fig.11 As shown in Figure 2, the residuals of the inner loop current model are all within 0.1°, and most are within 0.05°; Fig.12 As shown, the residual error of the servo valve is within 0.3°; Fig.13 As shown in the figure, the residual error of the actuator is within 0.03°, which meets the expected ideal target.

[0259] Fig.14 and 15 The simulation results when the servo actuator system simulates a fault, such as Fig.14 As shown in Figure 1, at 1812.5s, the servo actuator system acquired from the flight test data failed, which was quite different from the model output obtained by identification. Fig.15 As shown, the current fault diagnosis threshold is 1°, the residual difference exceeds 1°, and the model monitoring fault diagnosis signal is promptly reported as 1, which can realize fault diagnosis and meet the requirements.

Claims

1. A data-driven servo actuator system model identification and fault monitoring method, characterized in that: The steps include: Step (1) Mathematical modeling of aircraft steering gear electro-hydraulic servo actuation system (1.1) Basic equations of aircraft servo electro-hydraulic servo actuation system The electro-hydraulic servo actuation system of aircraft steering gear mainly consists of two parts: electro-hydraulic servo valve and hydraulic servo actuator; (1.1.1) Transfer function equation of electro-hydraulic servo valve The electro-hydraulic servo valve controls the hydraulic flow through electrical signals to achieve precise control of the hydraulic system; The valve core displacement is obtained by the driving current. The relationship between the driving current and the valve core displacement is regarded as a proportional link. The transfer function is: (1) In the formula, is the valve core displacement; is the current; is the current-spool displacement gain; (1.1.2) Sliding valve flow equation (2) In the formula, is the flow rate of oil in the slide valve, is the valve core displacement, is the pressure of the oil in the valve core, is the flow gain; is the flow-pressure coefficient; The dynamic analysis of the position servo system is carried out under zero-position working conditions, when the increment and variable are equal; due to the external leakage and compressibility of the hydraulic servo actuator, the flow rate into the oil inlet chamber of the hydraulic servo actuator is The flow rate out of the hydraulic servo actuator return oil chamber Not equal, define load flow for (3) (1.1.3) Flow continuity equation for hydraulic servo actuator The pressure loss and pipeline dynamics in the pipeline are ignored; the pressure in each working chamber of the hydraulic servo actuator is equal, and the oil temperature and bulk elastic modulus are constants; the internal and external leakage of the hydraulic servo actuator are laminar flows; Flow rate into the oil inlet chamber of the hydraulic servo actuator for (4) In the formula, is the pressure of the oil in the oil inlet chamber, is the pressure of the oil in the oil return chamber; Flow rate into the return oil chamber of the hydraulic servo actuator for (5) In the formula, is the displacement of the hydraulic servo actuator; is the cross-sectional area of ​​the hydraulic servo actuator; is the internal leakage coefficient of the hydraulic servo actuator; is the external leakage coefficient of the hydraulic servo actuator; It is the volume of the oil inlet chamber of the hydraulic servo actuator; The volume of the oil return chamber of the hydraulic servo actuator: is the effective bulk elastic modulus; In formula (4) and formula (5), is the flow rate required to drive the servo actuator. is the internal leakage flow of the oil inlet chamber and the oil return chamber of the servo actuator. , are the external leakage flow of the oil inlet chamber and the oil return chamber of the servo actuator, and The flow rates required by the oil inlet chamber and the oil return chamber due to oil compression and deformation in the chamber are called the oil inlet chamber compression flow rate and the oil return chamber compression flow rate respectively; The volume of the oil inlet chamber and oil return chamber of the hydraulic servo actuator is (6) (7) In the formula, is the initial volume of the oil inlet chamber; is the initial volume of the oil return chamber; From equations (3), (4), (5), (6) and (7), the flow continuity equation is: (8) In the formula, the external leakage flow of the oil inlet chamber and the oil return chamber is and Ignore; if the oil inlet cavity compression flow and oil return chamber compression flow If they are equal, ; Because the valve is matched symmetrically, the flow through the oil inlet valve port and the oil return valve port is also equal; therefore, the total pressure in dynamic Still applies; due to the difference between the oil inlet chamber pressure and the oil return chamber pressure, that is, the pressure drop ,so , ; thus there is (9) To make the compression flow equal, the initial volume of the two chambers of the hydraulic servo actuator should be and Equal, that is (10) In the formula, It is the average volume of the oil inlet chamber and the oil return chamber of the hydraulic servo actuator; is the total compression volume; because , , then formula (8) is simplified to (11) In the formula, is the total leakage coefficient of the hydraulic servo actuator, ; In formula (11), It is the flow rate required to drive the hydraulic servo actuator. is the total leakage flow, is the total compression flow; (1.1.4) Load balance equation of hydraulic servo actuator The balance equation between the output force and load force of the hydraulic servo actuator is: (12) In the formula, is the output force of the hydraulic servo actuator, It is the total mass of the hydraulic servo actuator and the load converted to the actuator; is the viscous damping coefficient of the hydraulic servo actuator and the load; is the elastic stiffness of the load; is the external load force acting on the hydraulic servo actuator; (1.2) Transfer function derivation Taking the Lagrange transformation of equation (2), equation (11) and equation (12), we can get (13) (14) (15) In formula (14) and formula (15), is a complex variable, expressed as ,in is the real part, is the imaginary part, is an imaginary unit, which is considered as a differential operator here. Differentiation in the time domain corresponds to multiplication in the complex frequency domain. ; In formula (13), (14), (15) , , , In the time domain , , , The frequency domain form of Equations (13), (14), and (15) fully describe the dynamic characteristics of the servo actuator system; eliminating intermediate variables and , find the valve core displacement and external load When acting simultaneously, the total output displacement of the servo actuator It is expressed as: (16) In the formula, is the total flow pressure coefficient; valve core displacement is the input quantity of the system, the external load torque is the disturbance quantity. The external disturbance quantity is also an input quantity, but its effect is opposite to that of the input quantity. From formula (16), we can see the response characteristics of the hydraulic servo actuator to the displacement of the valve and the external load force input; (1.3) Transfer function simplification The servo actuator system is mainly based on inertial load, so the elastic load and the influence of viscous damping are ignored, that is, (17) In the formula, is the natural frequency of hydraulic pressure; is the hydraulic damping ratio; Viscosity coefficient Very small, so ; Displacement of hydraulic servo actuator The transfer function of the valve core displacement is: (18) Step (2) Neural network identification method of aircraft servo actuation system model based on test flight data In order to obtain an accurate model of the aircraft steering gear electro-hydraulic servo actuation system, the BP neural network method is adopted. The test flight data collected by the aircraft steering gear electro-hydraulic servo actuation system sensor is used. The input and output data of the aircraft steering gear electro-hydraulic servo actuation system are obtained through the mathematical calculation of the aircraft steering gear electro-hydraulic servo actuation system established in step (1). The aircraft steering gear electro-hydraulic servo actuation system is identified to obtain a neural network model of the aircraft steering gear electro-hydraulic servo actuation system with more accurate output. The specific process is as follows: (2.1) Flight test data collection Data collection is performed during multiple flight phases, including takeoff, cruise, maneuvering and landing, to ensure coverage of all possible operating conditions; The test flight data of the aircraft steering gear electro-hydraulic servo actuation system include the following: Aircraft steering gear control instructions, electro-hydraulic servo valve control instructions, electro-hydraulic servo valve current collection value, electro-hydraulic servo valve voltage collection value, electro-hydraulic servo valve valve core displacement, actuator-controlled rudder surface deflection angle; (2.2) Data processing After filtering and resampling the valve core output signal of the electro-hydraulic servo valve and the output displacement signal data of the actuator, neural network identification training is performed; (2.3) Neural network model construction and identification Model: Aircraft steering gear electro-hydraulic servo actuation system model; The model is divided into two parts, namely the neural network model of the electro-hydraulic servo valve and the neural network model of the actuator; The input layer of the electro-hydraulic servo valve neural network model is: control surface control instructions and electro-hydraulic servo valve instructions, and the output layer is the valve core displacement; The input layer of the actuator neural network model is: control command of the rudder surface, command of the electro-hydraulic servo valve and displacement of the spool of the electro-hydraulic servo valve, and the output layer is the deflection angle of the rudder surface output by the actuator. The two neural network models were trained separately, both using the back-propagation BP neural network algorithm with two layers of feed-forward networks of hidden neurons with activation functions and linear output neurons; Step (3) Design of dynamic monitoring structure of aircraft steering gear electro-hydraulic servo actuation system based on test flight data For the electro-hydraulic servo valve neural network model and actuator neural network model obtained in step (2), design the corresponding servo actuator system model monitoring module, implement real-time monitoring, use the established dynamic model to continuously evaluate the system status, and set monitoring indicators and thresholds. The specific process is as follows: (3.1) Model monitoring external loop structure design The outer loop of the aircraft steering gear electro-hydraulic servo actuation system model monitoring module is designed for the aircraft electro-hydraulic servo actuation system model established in step (2), monitors the input and output of the model, and compares it with the input and output data of the servo actuation system on the actual aircraft, so as to achieve the function of monitoring the state of the servo actuation system during the flight of the aircraft, and at the same time compares and analyzes the output data of the two to realize fault detection of the aircraft servo actuation system; The outer loop model monitoring is to monitor the entire aircraft steering gear electro-hydraulic servo actuation system model, that is, the loop input is the aircraft control command signal, and the output is the output displacement or deflection angle of the actuation system; the model in the outer loop corresponds to the actuator neural network model in step (2); The fault judgment conditions of the mathematical model output in model monitoring and the data collected from the flight test data are as follows: (19) In the formula, Output of flight test data or output of real actuation system; It is the output of mathematical model in model monitoring; is the fault threshold. When this value is exceeded, it is judged that the actual actuation system has a fault. Formula (19) does not take into account the time required for the model monitoring system to calculate and the time consumed in the data transmission process, so it is necessary to consider the time difference between the outputs of the two; therefore, the time judgment condition is added as follows: (20) In the formula, For never being satisfied To meet time; To satisfy To not satisfied moment; is the time threshold; meaning is, when Too long, more than , it is considered to be faulty; (3.2) Model monitoring inner loop structure design The inner loop model monitoring is divided into three parts, namely: control electric signal model monitoring, electro-hydraulic servo valve model monitoring and actuator model monitoring; in the inner loop, the control electric signal module reference model is the coil resistance of the torque motor in the electro-hydraulic servo valve, and its rated value is taken as the reference model; the models in the electro-hydraulic servo valve model monitoring and the actuator model monitoring correspond to the electro-hydraulic servo valve neural network model and the actuator neural network model in step (2), and the input and output of the two are taken as the monitoring model; Control electrical signal model monitoring object: control voltage and control current ; Fault judgment conditions for the coil resistance of the torque motor in the aircraft steering gear electro-hydraulic servo valve: (21) In the formula, is the coil resistance of the torque motor, , is the rated value of the torque motor coil resistance, is the resistance fault threshold; The monitoring objects of the electro-hydraulic servo valve model: electro-hydraulic servo valve control signal and valve core displacement; Fault judgment conditions for electro-hydraulic servo valve spool displacement: (22) In the formula, The valve core displacement collected by the aircraft servo sensor. is the valve core displacement output by the neural network model, is the valve core displacement fault threshold; Actuator model monitoring object: deflection angle of the rudder surface driven by the actuator; Fault judgment conditions of actuator: (23) In the formula, is the deflection angle of the rudder surface collected by the aircraft servo sensor. is the rudder deflection angle output by the neural network model, is the fault threshold of the rudder surface deflection angle; the relationship between the rudder surface deflection angle and the actuator output displacement is: (24) In the formula, is the actuator displacement-angle proportionality coefficient; The above monitoring object is built through the neural network model in step (2), which is compared with the test flight data and the residual situation between the two is analyzed to realize the fault monitoring function.

Images