Gear transmission system intelligent state sensing device and monitoring method

Abstract

The application relates to an intelligent state sensing device and a monitoring method of a gear transmission system, and belongs to the technical field of mechanical transmission monitoring. The defects of poor real-time performance, low integration and insufficient universality of state monitoring and fault diagnosis of the gear transmission system in the prior art are solved. The device comprises an acceleration sensor, a core processor, a display and output module, and a storage and communication module. The acceleration fault diagnosis model in the core processor is used for preprocessing the vibration signals of gears and bearings in the gear transmission system into root mean squares, obtaining a moving average line in a moving average convergence divergence (MACD) line corresponding to the root mean squares, and monitoring the fault conditions of the gears or bearings of the gear transmission system based on the moving average line and a preset fault threshold. The application realizes efficient, real-time and lightweight online state monitoring and fault diagnosis of the gear transmission system, and is suitable for online state monitoring and fault diagnosis of gears, bearings and other key components.

Description

Technical Field

[0001] This invention belongs to the field of mechanical transmission monitoring technology, specifically relating to an intelligent state sensing device and monitoring method for a gear transmission system. Background Technology

[0002] In recent years, some embedded intelligent systems have begun to be used for condition monitoring of gear transmission systems, but existing solutions mostly focus on simple data acquisition and transmission functions. For example, sensor data is sent to a host computer for processing via an embedded module, and diagnostic algorithms typically run on an external computer or in the cloud. This approach is limited by communication latency and dispersed computing resources, resulting in low system integration and an inability to achieve fast and independent fault diagnosis locally.

[0003] The most common technology currently is to collect vibration signals of the gear transmission system during operation by installing vibration sensors (such as accelerometers), and use signal processing technology (such as Fourier transform and time-frequency analysis) to extract feature parameters (such as amplitude and frequency) to determine whether there are faults such as wear, cracks or spalling in the gears or bearings. The shortcomings of the existing technology mainly include: (1) relying on external data acquisition equipment and independent analysis system, the diagnosis process is usually completed offline, the real-time performance is poor, it cannot meet the needs of real-time monitoring in industrial sites, and the adaptability to complex working conditions is limited; (2) sensors, data processing and diagnosis modules are often separated, the wiring is complicated, which leads to increased system complexity and cost; (3) lacking localized intelligent diagnosis capabilities, it is difficult to classify fault types accurately in real time.

[0004] In addition, some solutions involve installing temperature sensors in key parts of the gearbox to monitor temperature changes during operation and infer the health status of the gears. For example, when gears experience increased friction or lubrication failure, the temperature will rise abnormally. However, this method has low sensitivity to early-stage faults, can only reflect relatively obvious anomalies, and cannot accurately pinpoint the type of fault. Summary of the Invention

[0005] In view of the above problems, the present invention provides an intelligent state sensing device and monitoring method for gear transmission systems, which solves the defects of poor real-time performance, low integration and insufficient versatility in the state monitoring and fault diagnosis of gear transmission systems in the prior art. The embedded intelligent sensing device and method provided by the present invention realizes efficient, real-time and lightweight universal online state monitoring and fault diagnosis of gear transmission systems, and is applicable to online state monitoring and fault diagnosis of gears, bearings and other key components.

[0006] This invention provides an intelligent state sensing device for a gear transmission system, including an acceleration sensor, a core processor, a display and output module, and a storage and communication module;

[0007] An accelerometer is installed in the gearbox near the gears and bearings to collect vibration signals from the gears and bearings in the gear transmission system.

[0008] The core processor integrates an accelerated fault diagnosis model. This model is used to preprocess the vibration signals of gears and bearings in the gear transmission system into root mean square (RMS), obtain the moving average line in the MACD line corresponding to the MMS convergence and divergence index, and monitor the fault status of gears or bearings in the gear transmission system based on the moving average line and a preset fault threshold.

[0009] The display and output module is used to display the fault type, root mean square value, and confidence level in real time.

[0010] The storage and communication module connects to the accelerometer sensor and reads and stores the sensor data collected by the accelerometer sensor.

[0011] In another aspect, the present invention provides a method for intelligent state perception and monitoring of a gear transmission system, which employs the aforementioned intelligent state perception device for the gear transmission system to monitor its intelligent state. The specific steps are as follows:

[0012] Step 1. After preprocessing the vibration signals of the gears and bearings collected by the accelerometer, obtain the current moment. t The root mean square of;

[0013] Step 2. Obtain the standard deviation of the root mean square series based on historical root mean square data. Based on standard deviation Obtain the fault threshold If at the current moment t root mean square Exceeding the fault threshold Proceed to step 3; if the current time t root mean square The fault threshold was not exceeded. Let t = t + 1, then return to step 1;

[0014] Step 3. Obtain information about the current time. t The moving average line in the MACD line is used to monitor whether gears or bearings in a transmission system are faulty, based on the historical root mean square data sequence.

[0015] Optionally, the method also includes step 4, which uses a lightweight fault diagnosis model to classify and diagnose the fault types of gears or bearings in the gear transmission system.

[0016] Optionally, the current time t root mean square The expression is:

[0017]

[0018] in, The first vibration signal of the gear and bearing collected at the current moment. i One data point; This represents the data length of the vibration signal collected by the accelerometer in a single instance at the current moment.

[0019] Optionally, the specific steps of step 3 are as follows:

[0020] Step 31. Obtain information about the current time. t Short-term simple moving average of historical root mean square data series and long-term simple moving average ;

[0021] Step 32. Using a short-term simple moving average and long-term simple moving average Get information about the current moment t MACD line of historical root mean square (RMS) data series ;

[0022] Step 33. Based on information about the current time t MACD values ​​of historical root mean square data series Get the current time t MACD line signal line ;

[0023] Step 34. If the current time t MACD line crossing signal line The cause was determined to be a gear or bearing failure in the transmission system.

[0024] Alternatively, a short-term simple moving average The expression is:

[0025]

[0026] in, Indicates the first k Each historical square root.

[0027] Alternatively, long-term simple moving average The expression is:

[0028] .

[0029] Compared with the prior art, the present invention has at least the following beneficial effects:

[0030] (1) The intelligent state sensing device and monitoring method of the present invention can be localized and highly integrated. By integrating the NPU in the core processor, caching data in the local storage and communication module, and providing real-time feedback in the display and output module, the entire process of edge-side closed-loop processing from signal acquisition to preprocessing, then to RMS-MACD anomaly detection, and then to lightweight model diagnosis is realized, which ensures the basis of real-time performance.

[0031] (2) The root mean square value-moving average convergence divergence index RMS-MACD algorithm of the intelligent state sensing device and monitoring method of the present invention serves as the first line of defense for state monitoring. It only calculates the difference between the root mean square sequence standard deviation and part of the periodic SMA, quickly determines the occurrence of faults, and triggers subsequent diagnosis.

[0032] (3) The lightweight fault diagnosis model of the intelligent state sensing device and monitoring method of the present invention performs accurate classification after the fault occurs, and achieves efficient diagnosis by combining lightweight design and neural network processing unit (NPU) acceleration.

[0033] (4) The intelligent state sensing device and monitoring method of the present invention can monitor the operating status of the gear transmission system in real time and determine whether there is an abnormality or potential fault. Attached Figure Description

[0034] The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of the invention.

[0035] Figure 1 This is a flowchart of the intelligent state perception and monitoring method for gear transmission systems according to the present invention;

[0036] Figure 2 This is a flowchart of the training process for the lightweight fault diagnosis model of this invention. Detailed Implementation

[0037] To better understand the above-described objectives, features, and advantages of the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other. Furthermore, the present invention can be implemented in other ways different from those described herein; therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0038] A specific embodiment of the present invention, such as Figures 1-2 The present invention discloses an intelligent state sensing device for a gear transmission system, comprising a core processor, an acceleration sensor, a display and output module, and a storage and communication module;

[0039] An accelerometer with a sampling frequency of 25.6 kHz is installed in the gearbox near the gears and bearings to collect vibration signals from the gears and bearings in the gear transmission system.

[0040] The core processor uses a dual-core ARM processor with a main frequency of 1.5GHz. It integrates a neural network processing unit (NPU) and an accelerated fault diagnosis model. The NPU is used to accelerate the inference of the fault diagnosis model. The accelerated fault diagnosis model is used to preprocess the vibration signals of gears and bearings in the gear transmission system into root mean square (RMS) and obtain the moving average line in the MACD line corresponding to the MMS convergence and divergence index. Based on the preset fault threshold, the fault status of gears or bearings in the gear transmission system is monitored based on the moving average line.

[0041] The display and output module can display the fault type, root mean square value, and confidence level in real time via an OLED display or a host computer; it also supports displaying alarm signals via the GPIO interface.

[0042] The storage and communication module has a built-in SD card for storing data and lightweight fault diagnosis models, a 1MB data cache module, and connects to the accelerometer via a network cable interface, using the TCP / IP protocol to read sensor data.

[0043] Furthermore, the core processor embeds an embedded Linux system based on the aarch64 architecture, supporting the PyTorch 2.0.0 runtime environment to ensure compatibility with general deep learning models.

[0044] Furthermore, the model file (.pth format) of the lightweight fault diagnosis model is optimized with TorchScript to adapt to the aarch64 architecture and improve inference efficiency.

[0045] Another specific embodiment of the present invention discloses an intelligent state perception and monitoring method for a gear transmission system, which uses the aforementioned intelligent state perception device for a gear transmission system to monitor the intelligent state of the gear transmission system. The specific steps are as follows:

[0046] Step 1. After preprocessing the vibration signals of the gears and bearings collected by the accelerometer, obtain the current moment. t root mean square The expression is:

[0047]

[0048] in, For the current moment t The first vibration signal of the collected gears and bearings i One data point; This represents the data length of the vibration signal collected by the accelerometer in a single instance at the current moment.

[0049] Step 2. Obtain the standard deviation of the root mean square RMS series based on historical root mean square RMS data. Fault thresholds are obtained based on standard deviation. If at the current moment t root mean square Exceeding the fault threshold Proceed to step 3; if the current time t root mean square The fault threshold was not exceeded. Let t = t + 1, then return to step 1;

[0050] Step 3. Calculate the value of the current time using a Simple Moving Average (SMA) model. t The moving average line in the moving average convergence and divergence indicator MACD (Moving Average Convergence and Divergence) of historical root mean square (RMS) data series is used to monitor whether gears or bearings in a transmission system are faulty, thereby simplifying calculations and adapting to the resource constraints of embedded systems.

[0051] Step 31. Obtain information about the current time. t Short-term simple moving average of historical root mean square data series and long-term simple moving average ;

[0052] Furthermore, short-term simple moving average The expression is:

[0053]

[0054] in, Indicates the first k The historical root of square represents the current moment. t The 11th time from the previous time to the current time k Each historical square root.

[0055] Furthermore, long-term simple moving average The expression is:

[0056]

[0057] in, Indicates the first n The historical root of square represents the current moment. t The 25th time from the previous time to the current timen Each historical square root.

[0058] Step 32. Using a short-term simple moving average and long-term simple moving average Get information about the current moment t MACD value of historical root mean square RMS data series The expression is:

[0059] .

[0060] Step 33. Based on information about the current time t MACD value of the root mean square data sequence Get the current time t MACD line signal line The expression is:

[0061] .

[0062] in, Indicates the current time t The first 8 moments to the current moment j One MACD value.

[0063] Step 34. If the current time t MACD line crossing signal line ( And at the current moment t root mean square Exceeding the fault threshold If this occurs, it is determined that a gear or bearing failure has occurred in the transmission system.

[0064] Step 4. Call the lightweight fault diagnosis model in offline mode to classify and diagnose the fault types of gears or bearings in the gear transmission system.

[0065] Optionally, the MobileNetV3 model can be used as a lightweight fault diagnosis model.

[0066] Specifically, the lightweight fault diagnosis model takes vibration signals as input and outputs the fault type of the gear transmission system ("normal", "tooth surface scuffing", "gear tooth breakage", "bearing inner ring fault", "bearing outer ring fault", etc.) and confidence level. The diagnostic results can be displayed on an OLED screen (e.g., "bearing inner ring fault, confidence level 95%), or transmitted to a host computer. Severe faults will trigger GPIO alarms.

[0067] The trained lightweight fault diagnosis model is saved as a .pth file and converted into a format that can run on embedded systems using TorchScript. The lightweight fault diagnosis model reduces the number of parameters by more than 90% compared to conventional fault diagnosis models, while maintaining a diagnostic accuracy of over 95% to adapt to the resource constraints of embedded systems.

[0068] Furthermore, the intelligent state perception device for gear transmission systems is compatible with any gear transmission system fault diagnosis model developed based on the PyTorch deep learning framework. It can be continuously updated and iterated upon, and can be used to test different fault diagnosis algorithms. The NPU ensures rapid inference, meeting the requirements of real-time fault diagnosis.

[0069] To illustrate the effectiveness of the method proposed in this invention, the following detailed description of the above technical solution is provided through a specific embodiment. The specific implementation steps are as follows:

[0070] The intelligent state sensing device for gear transmission systems includes a core processor, acceleration and vibration sensors, a display and output module, and a storage and communication module. Installed near the gears and bearings in the gearbox, it is used for online state monitoring and fault diagnosis of the gear transmission system. The proposed real-time state monitoring algorithm and lightweight fault diagnosis model for gear transmission systems are deployed in an embedded controller, achieving high real-time state monitoring and diagnosis of the gear transmission system at the edge. The specific training process of the lightweight fault diagnosis model is as follows:

[0071] Step 1: An acceleration vibration sensor is deployed in the gearbox near the gears and bearings to collect vibration signals of the gear transmission system in real time at a sampling frequency of 25.6KHz, which is used to train the lightweight fault diagnosis model.

[0072] Step 2: Construct training and testing datasets for gear transmission systems under typical faults. The training set is used to train a lightweight fault diagnosis model, and the testing set is used to test the diagnostic success rate of the fault classification algorithm.

[0073] Step 3: Adjust the parameters of the lightweight fault diagnosis model and train it offline to obtain a MobileNetV3 lightweight fault diagnosis model with high diagnostic accuracy.

[0074] Step 4: After the lightweight fault diagnosis model is trained, save it as a .pth format and optimize it with TorchScript to adapt to the aarch64 architecture and improve the inference efficiency on the edge side.

[0075] Step 5: The online state monitoring algorithm does not require offline training and relies on real-time vibration signals for state monitoring.

[0076] Step 6: Deploy the real-time state monitoring algorithm and the trained MobileNetV3 model in the embedded system at the edge.

[0077] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for intelligent state perception and monitoring of a gear transmission system, comprising using an intelligent state perception device for the gear transmission system to intelligently perceive and monitor the state of the gear transmission system, characterized in that, The intelligent state sensing device for the gear transmission system includes an acceleration sensor, a core processor, a display and output module, and a storage and communication module. An accelerometer is installed in the gearbox near the gears and bearings to collect vibration signals from the gears and bearings in the gear transmission system. The core processor integrates an accelerated fault diagnosis model. This model is used to preprocess the vibration signals of gears and bearings in the gear transmission system into root mean square (RMS), obtain the moving average line in the MACD line corresponding to the MMS convergence and divergence index, and monitor the fault status of gears or bearings in the gear transmission system based on the moving average line and a preset fault threshold. The display and output module is used to display the fault type, root mean square value, and confidence level in real time. The storage and communication module connects to the accelerometer sensor and reads and stores the sensor data collected by the accelerometer sensor. The specific steps are as follows: Step 1. After preprocessing the vibration signals of the gears and bearings collected by the accelerometer, obtain the current moment. t The root mean square of; Step 2. Obtain the standard deviation of the root mean square series based on historical root mean square data. Based on standard deviation Obtain the fault threshold If at the current moment t root mean square Exceeding the fault threshold Proceed to step 3; if the current time t root mean square The fault threshold was not exceeded. Let t = t + 1, then return to step 1; Step 3. Obtain information about the current time. t The moving average line in the MACD line is used to monitor whether gears or bearings in a transmission system are faulty, based on the historical root mean square data sequence.

2. The intelligent state perception and monitoring method for gear transmission systems according to claim 1, characterized in that, It also includes step 4, which uses a lightweight fault diagnosis model to classify and diagnose the fault types of gears or bearings in the gear transmission system.

3. The intelligent state perception and monitoring method for gear transmission systems according to claim 1, characterized in that, Current moment t root mean square The expression is: in, The first vibration signal of the gear and bearing collected at the current moment. i One data point; This represents the data length of the vibration signal collected by the accelerometer in a single instance at the current moment.

4. The intelligent state perception and monitoring method for gear transmission systems according to claim 1, characterized in that, The specific steps of step 3 are as follows: Step 31. Obtain information about the current time. t Short-term simple moving average of historical root mean square data series and long-term simple moving average ; Step 32. Using a short-term simple moving average and long-term simple moving average Get information about the current moment t MACD line of historical root mean square (RMS) data series ; Step 33. Based on information about the current time t MACD values ​​of historical root mean square data series Get the current time t MACD line signal line ; Step 34. If the current time t MACD line crossing signal line The cause was determined to be a gear or bearing failure in the transmission system.

5. The intelligent state perception and monitoring method for gear transmission systems according to claim 4, characterized in that, Short-term simple moving average The expression is: in, Indicates the first k Each historical square root.

6. The intelligent state perception and monitoring method for gear transmission systems according to claim 4, characterized in that, Long-term simple moving average The expression is: in, Indicates the first n Each historical square root.

Images