Fault diagnosis method of electric cylinder of six-degree-of-freedom motion platform

By installing acceleration sensors in the flight simulator and using the FD-ELM algorithm, a fault diagnosis model is generated, and the problem of fault diagnosis of the electric cylinder of the six-degree of freedom motion platform is solved, and rapid and automated fault identification and positioning is achieved, ensuring equipment and personnel Safety and improve detection efficiency.

CN114608817BActive Publication Date: 2024-11-22AVIC AVIATION SIMULATION SYST CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210193786.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-03-01
Publication Date
2024-11-22
Estimated Expiration
2042-03-01

AI Technical Summary

Technical Problem

The prior art cannot effectively diagnose the failure of the six-degree-of-freedom motion platform electric cylinder in flight simulators, especially in the absence of regularity of data and limited sensor installation, making it difficult to achieve efficient fault pattern recognition.

Method used

Install an axial acceleration sensor under the six-degree of freedom motion platform, set up the diagnostic actions of periodic vibration data, and analyze the vibration data using a fixed dictionary extreme learning machine (FD-ELM), generate a fault diagnosis model based on neural network, and realize the implementation of a Automatic identification of electric cylinder failures.

Benefits of technology

By reducing the number of sensors, saving costs, obtaining periodic vibration data, improving the application effect of neural network algorithms in fault diagnosis, achieving rapid fault positioning and automated detection, ensuring the safety of equipment and personnel, improving detection efficiency, high recognition accuracy, and suitable for flight The simulator's electric cylinder fault diagnosis scenario.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114608817B_ABST
    Figure CN114608817B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for diagnosing faults of electric cylinders of a six-degree-of-freedom motion platform, wherein a plurality of acceleration sensors with an axial sampling frequency are installed below the upper platform of the six-degree-of-freedom motion platform; the motion characteristics of the six-degree-of-freedom motion platform are analyzed, and a set of actions that can obtain periodic vibration data are set as diagnostic actions; the diagnostic actions are run on the six-degree-of-freedom motion platform, and the vibration data of the electric cylinder in different states are obtained by the acceleration sensor; the vibration data are learned using FD-ELM and a fault diagnosis model based on a neural network is generated; the diagnostic actions are automatically run when the six-degree-of-freedom motion platform to be diagnosed is turned on and off, and the vibration data of the electric cylinder is obtained by the acceleration sensor; the vibration data is analyzed using a trained fault diagnosis model to determine whether the electric cylinder is in a normal or faulty state. The present invention is conducive to obtaining repairs before the function of the electric cylinder is degraded or fails, and ensures the safety of equipment and personnel.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of flight simulators, and in particular to a fault diagnosis method for an electric cylinder of a six-degree-of-freedom motion platform implemented by using a fixed dictionary extreme learning machine. Background Art

[0002] Flight simulators can simulate most flight scenarios and tasks, which is of great help in improving the skills of trainees. Compared with traditional real-life training methods, training with flight simulators obviously has a broader prospect of use and lower training costs.

[0003] As an important executive component of the six-degree-of-freedom motion platform in the flight simulator, the electric cylinder undertakes the important task of providing dynamic simulation. Its health status directly affects the training effect and life safety of the flight simulator users. However, the existing means of maintaining the working ability of the electric cylinder are mainly preventive maintenance, regular observation and replacement of the lubricating oil in the electric cylinder, and relying on human ears to identify whether there is abnormal noise during operation. This method cannot diagnose the electric cylinder failure mode that affects safety.

[0004] Due to the high frequency of use, long continuous working time and high reliability requirements of electric cylinders, there is a very high demand for their fault diagnosis technology. At the same time, except for the encoder at the rear end of the motor that can obtain the motor speed and the stroke of the electric cylinder, it is difficult to install more sensors on the electric cylinder body, which makes the fault diagnosis of the electric cylinder more difficult. Therefore, new diagnostic methods must be introduced to solve this problem. In the flight simulator, the action of the six-degree-of-freedom motion platform is completely dependent on the operation of the trained pilot. The voltage, current, temperature, encoder feedback speed and other data collected by the built-in sensors are not regular and lack analytical value. The neural network fault diagnosis method, which has been widely used in the field of fault diagnosis in recent years and has good results, is difficult to be effectively applied to this kind of data. Summary of the invention

[0005] In order to meet the fault diagnosis needs of the electric cylinder of a six-degree-of-freedom motion platform, the present invention proposes a fault diagnosis method for the electric cylinder of a six-degree-of-freedom motion platform. An axial acceleration sensor is installed under the upper platform of the six-degree-of-freedom motion platform of a flight simulator, and a diagnostic action is set to obtain vibration data with periodic characteristics; then a fixed dictionary extreme learning machine (FD-ELM) with a single hidden layer feedforward neural network algorithm with low complexity and extremely fast running speed is used to analyze the vibration data obtained by the sensor to achieve the purpose of electric cylinder fault diagnosis.

[0006] The object of the present invention is achieved through the following technical solutions:

[0007] A method for diagnosing faults of an electric cylinder of a six-degree-of-freedom motion platform comprises the following steps:

[0008] Step 1: Install several acceleration sensors with axial sampling frequency below the upper platform of the six-degree-of-freedom motion platform;

[0009] Step 2: Analyze the motion characteristics of the six-degree-of-freedom motion platform and set a set of actions that can obtain periodic vibration data as diagnostic actions;

[0010] Step 3: Run the diagnostic action on the six-degree-of-freedom motion platform, and obtain the vibration data of the electric cylinder in different states by the acceleration sensor;

[0011] Step 4: Use FD-ELM to learn the vibration data and generate a neural network-based fault diagnosis model;

[0012] Step 5: When the six-degree-of-freedom motion platform to be diagnosed is turned on and off, the diagnostic action is automatically performed and the vibration data of the electric cylinder is obtained by the acceleration sensor;

[0013] Step 6: Use the trained fault diagnosis model to analyze the vibration data to determine whether the electric cylinder is in a normal or faulty state. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 Principles of fault diagnosis methods;

[0015] Figure 2 Sensor installation location;

[0016] Figure 3 Diagnostic action diagram;

[0017] Figure 4 Vibration signal acquisition example;

[0018] Figure 5 Fault diagnosis flow chart; DETAILED DESCRIPTION

[0019] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0020] like Figure 1 As shown, the fault diagnosis method of the electric cylinder of a six-degree-of-freedom motion platform shown in this embodiment has the following specific steps:

[0021] Step 1: Install several acceleration sensors with axial sampling frequency below the upper platform of the six-degree-of-freedom motion platform to obtain the vibration signal of the electric cylinder.

[0022] In order to save costs, this embodiment installs three acceleration sensors under the upper platform of the six-degree-of-freedom motion platform, such as Figure 2As shown, each acceleration sensor is installed at the junction of two adjacent electric cylinders. According to the general characteristic of mechanical failure that the closer to the fault point, the higher the vibration energy, the position of the faulty electric cylinder can be quickly found by analyzing the energy contained in the vibration signals of the three acceleration sensors. This embodiment can quickly locate the position of the faulty electric cylinder and identify the fault mode while using fewer sensors.

[0023] Step 2: Analyze the motion characteristics of the six-degree-of-freedom motion platform and set a set of actions that can obtain periodic vibration data as diagnostic actions.

[0024] In order to solve the problem that the data obtained by the six-degree-of-freedom motion platform in the flight simulator lacks regularity and cannot be used for fault diagnosis and analysis, such as Figure 3 As shown, in this embodiment, the upper platform of the six-degree-of-freedom motion platform is kept horizontal and moves to the outer diameter limit, and then the upper platform is centered on the vertical axis and The angular velocity is set to move clockwise at a uniform speed, and this action is set as a diagnostic action to obtain vibration signals with analytical value.

[0025] Step 3: Run the diagnostic action described in step 2 on the six-degree-of-freedom motion platform, and obtain the vibration data of the electric cylinder in different states by the acceleration sensor.

[0026] In this embodiment, the acceleration sensor obtains vibration data of the electric cylinder in four states: normal, joint hinge wear, lead screw wear, and telescopic rod crack. The collected vibration data is as follows: Figure 4 shown.

[0027] Step 4: Use FD-ELM to learn the vibration data obtained in step 3 and generate a neural network-based fault diagnosis model.

[0028] The algorithm steps are as follows:

[0029] 1) Use Gaussian smoothing to smooth the vibration data and eliminate the low-energy environmental noise;

[0030] 2) Divide the vibration data into n arrays of length m as neural nodes and mark them with corresponding labels as samples;

[0031] 3) Initialize FD-ELM and input the sample set consisting of vibration signals to train the neural network:

[0032] The formula of FD-ELM can be summarized as follows:

[0033]

[0034] Among them, x is the input vibration signal, α is the input weight matrix, β is the output weight matrix, g is the activation function, and L is the number of input samples.

[0035] Use formula (2)(3) to initialize the input weight matrix α of the fixed dictionary learning machine.

[0036]

[0037]

[0038] Among them, C is the encoding matrix, o, p and q are independent variables. In this paper, their default values ​​are 3, 2 and 1.53 respectively. In different data sets, the values ​​of the three variables should be adjusted according to the characteristics of the data set.

[0039] Input n labeled samples into the fixed dictionary learning machine for label learning:

[0040] The objective function is solved using the generalized inverse matrix method to obtain the output weight matrix β:

[0041] minimize||fT||=||sigmoid(x m×n code n·m )·β m×j -T|| (4)

[0042] Where T is the label matrix.

[0043] 4) Formula (1) and the calculated α matrix and β matrix constitute a trained neural network-based fault diagnosis model.

[0044] Step 5: When the six-degree-of-freedom motion platform to be diagnosed is turned on and off, the diagnostic action is automatically performed and the vibration data of the electric cylinder is obtained by the acceleration sensor.

[0045] Step 6: Use the trained fault diagnosis model to analyze the vibration data to determine whether the electric cylinder is in a normal or faulty state. If it is in a faulty state, analyze the energy characteristics of the vibration data of the three sensors to locate the specific faulty electric cylinder. Figure 5 As shown:

[0046] 1) If the vibration data of the electric cylinder is recognized as normal: end the diagnosis and complete the startup and shutdown actions.

[0047] 2) If the vibration data of the electric cylinder is identified as one of the three states of joint hinge wear, lead screw wear, and telescopic rod crack: Calculate the energy of the vibration data obtained by the three acceleration sensors respectively through formula (5). Select the two electric cylinders corresponding to the acceleration sensors with the highest energy as the electric cylinders recommended for repair, and then determine the specific electric cylinder with fault according to the energy of the vibration data obtained by the remaining two sensors. Then report the determined electric cylinder and fault mode to the maintenance personnel for repair.

[0048] E=∑|x| (5)

[0049] Among them, E is the energy size and x is the vibration signal.

[0050] according to Figure 1 The installation positions of the sensors on the six-DOF motion platform and the fault location methods are summarized in Table 1.

[0051] Table 1

[0052]

[0053] by Figure 1 Take the two electric cylinders at the location of the acceleration sensor 2 as an example:

[0054] (1) If the faulty electric cylinder is the one connected to the location of acceleration sensor 2, the energy of the vibration signals obtained by the three sensors is ranked as follows: E2>E1, E3, and E1≈E3.

[0055] (2) If the faulty electric cylinder is the electric cylinder corresponding to the position of acceleration sensor 2 in the counterclockwise direction, the energy magnitudes of the vibration signals obtained by the three sensors are ranked as follows: E2>E1>E3.

[0056] The beneficial effects of the present invention are:

[0057] 1) The layout of the three acceleration sensors combined with the vibration signal energy analysis method in step 6 can reduce the number of acceleration sensors as much as possible and save costs while ensuring that the electric cylinder can quickly locate the fault;

[0058] 2) The setting of diagnostic actions can obtain vibration data with strong periodicity, so that the neural network algorithm can be used in the field of fault diagnosis of the six-degree-of-freedom motion platform of the flight simulator;

[0059] 3) The diagnostic action of rotating the platform on the six-degree-of-freedom motion platform clockwise around the vertical axis can make each electric cylinder move regularly, which is conducive to discovering fault characteristics;

[0060] 4) Different from the traditional electrical detection of electric cylinders, the present invention focuses on the detection of mechanical failures of electric cylinders according to the use scenarios of electric cylinders in flight simulators. This approach is conducive to obtaining repairs before the function of the electric cylinder is degraded or fails, and ensures the safety of equipment and personnel;

[0061] 5) The automated detection method saves a lot of manpower and improves detection efficiency. The analysis of vibration signals is conducive to discovering early faults of the electric cylinder and guiding maintenance and repair;

[0062] 6) The fixed dictionary extreme learning machine algorithm used has low complexity, fast running speed, high recognition accuracy, and is extremely efficient in the use of computing power and memory resources. It is very suitable for the electric cylinder fault diagnosis scenario when the six-degree-of-freedom motion platform of the flight simulator is turned on and off.

Claims

1. A method for diagnosing faults of electric cylinders of a six-degree-of-freedom motion platform, characterized in that The following steps are involved: Step 1: Install several acceleration sensors with axial sampling frequency below the upper platform of the six-degree-of-freedom motion platform; Step 2: Analyze the motion characteristics of the six-degree-of-freedom motion platform and set a set of actions that can obtain periodic vibration data as diagnostic actions; Step 3: Run the diagnostic action on the six-degree-of-freedom motion platform, and obtain the vibration data of the electric cylinder in different states by the acceleration sensor; Step 4: Use FD-ELM to learn the vibration data and generate a neural network-based fault diagnosis model; Step 5: When the six-degree-of-freedom motion platform to be diagnosed is turned on and off, the diagnostic action is automatically performed and the vibration data of the electric cylinder is obtained by the acceleration sensor; Step 6: Use the trained fault diagnosis model to analyze the vibration data to determine whether the electric cylinder is in a normal or faulty state. When there are three acceleration sensors installed under the upper platform of the six-degree-of-freedom motion platform and the diagnosis result is a fault, the energy of the vibration data obtained by the three acceleration sensors is calculated by formula (5). The two electric cylinders corresponding to the sensors with the highest energy are selected as the electric cylinders recommended for maintenance, and then the specific electric cylinder with a fault is determined according to the energy of the vibration data obtained by the remaining two sensors. E=∑|x| (5) Among them, E is the energy size and x is the vibration signal.

2. A method for diagnosing faults of electric cylinders of a six-degree-of-freedom motion platform according to claim 1, characterized in that The diagnostic action is to keep the upper platform of the six-degree-of-freedom motion platform horizontal and move to the outer diameter limit, and then make the upper platform move clockwise at a constant speed with the vertical axis as the center.

3. A method for diagnosing faults of electric cylinders of a six-degree-of-freedom motion platform according to claim 1, characterized in that The status of the electric cylinder in step three includes normal, joint hinge wear, lead screw wear, and telescopic rod crack.

4. A method for diagnosing faults of electric cylinders of a six-degree-of-freedom motion platform according to claim 1, characterized in that The steps of the FD-ELM algorithm are as follows: Gaussian smoothing is used to smooth the vibration data and remove low-energy ambient noise; Divide the vibration data into n arrays of length m and mark them with corresponding labels as samples; Initialize FD-ELM and input a sample set consisting of vibration signals to train the neural network: The formula of FD-ELM is as follows: Among them, x is the input vibration signal, α is the input weight matrix, β is the output weight matrix, g is the activation function, and L is the number of input samples; Use formula (2) (3) to initialize the input weight matrix α of the fixed dictionary learning machine: Among them, C is the encoding matrix, o, p and q are independent variables; The objective function is solved using the generalized inverse matrix method to obtain the output weight matrix β: minimize‖f-T‖=‖sigmoid(x m×n ·code n·m )·β m×j -T‖ (4) Where T is the label matrix; Formula (1) and the calculated α matrix and β matrix constitute a trained neural network-based fault diagnosis model.