Friction test load adjustment and steady state identification method and system based on embedded model

Abstract

The embodiment of the application relates to the field of automatic control, and provides a friction test load adjustment and steady state identification method and system based on an embedded model. Through the cooperative work of an upper computer, an industrial computer and an executing mechanism, a plurality of to-be-tested load conditions are selected in a self-defined test load range to form a test sequence; corresponding contact pressure is applied to a friction pair according to the test sequence and is adjusted to a specific sliding speed; friction factor and wear height data are collected in real time during the test process; an embedded state decision model is used to analyze the characteristics of original measurement data, and whether the fluctuation characteristics of the current friction system satisfy the preset condition of the current friction system reaching a steady state is judged; if the current friction system reaches the steady state, the steady state tribological parameters under the current to-be-tested load condition are recorded, the next to-be-tested load condition in the test sequence is automatically switched to, and the result file is exported when all the to-be-tested load conditions are tested, so that the friction test efficiency and the test result accuracy are improved.

Description

Technical Field

[0001] This application relates to the field of automation control, and in particular to a method and system for adjusting friction test load and identifying steady state based on an embedded model. Background Technology

[0002] Tribological research is fundamental to materials science and mechanical engineering, while tribological testing is a key method for obtaining the frictional loss properties of materials. In existing conventional friction tests, because the break-in period of different friction systems is difficult to predict, and the frictional properties are closely related to the system's load conditions (such as contact pressure and sliding speed), researchers usually use constant loads and conduct experiments within a fixed test time.

[0003] Currently, this constant load and fixed time testing scheme is widely used in the research and development stage of friction materials. The main process relies on manually preset parameters, with data recorded manually by experimenters or processed uniformly after the experiment. Although existing testing methods are widely used, in practice, to ensure that the test data covers sufficient steady-state values, a single experiment usually requires a long testing cycle. Because the experimental duration and load conditions are fixed, a large number of repetitive manual experimental operations must be performed to comprehensively obtain the tribological properties within the system, resulting in significant costs in manpower, electricity, and consumables. Existing testing methods often use a single load condition, which deviates significantly from the complex and variable working conditions in actual production, making it difficult for the measured data to accurately reflect the application effect. Current technical solutions fail to effectively distinguish the frictional performance during the break-in period and the steady-state period, lacking online monitoring and intelligent judgment of the testing process, resulting in the inability to obtain the critical point when the system reaches steady state in real time. Therefore, there is an urgent need to provide a technical solution to address the technical problems of low material research efficiency and poor test result accuracy caused by the difficulty in dynamically adjusting the test duration or automatically switching load conditions in related technologies. Summary of the Invention

[0004] This application provides a method and system for adjusting friction test load and identifying steady state based on an embedded model, aiming to solve the technical problems of low efficiency and poor accuracy of test results in material research caused by the difficulty in dynamically adjusting test duration or automatically switching load conditions in related technologies.

[0005] In a first aspect, embodiments of this application provide a method for adjusting friction test loads and identifying steady-state states based on an embedded model. The method is applied to a friction test system, which includes a host computer, an industrial computer, and an actuator. The method includes: The host computer receives the user's input of a custom test load range and uses the embedded state decision model built into the host computer to select multiple test load conditions within the custom test load range to form a test sequence. The host computer controls the actuator to apply the corresponding contact pressure to the friction pair and adjust it to a specific sliding speed according to the test sequence. During the test, the friction factor and wear height data are collected in real time by the sensor group of the industrial computer, and the collected raw measurement data is transmitted in real time to the host computer with an embedded state decision model. The host computer uses an embedded state decision model to perform feature analysis on the original measurement data to determine whether the fluctuation characteristics of the current friction system meet the preset conditions for the current friction system to reach a steady state. If the current friction system is detected to have reached a steady state, the steady-state tribological parameters under the current load condition are recorded, and the actuator is instructed to automatically switch to the next load condition in the test sequence until all load conditions are tested. After the test, export the result file containing the original measurement data and steady-state tribological parameters under each load condition.

[0006] Secondly, embodiments of this application provide a friction test load adjustment and steady-state identification system based on an embedded model. The system is used to perform friction testing and includes a host computer, an industrial computer, and an actuator. The host computer receives a user-inputted custom test load range and uses an embedded state decision model built into the host computer to select multiple test load conditions within the custom test load range to form a test sequence. The actuator is then controlled to apply a corresponding contact pressure to the friction pair and adjust it to a specific sliding speed according to the test sequence. The actuator is used to apply the corresponding contact pressure to the friction pair and adjust it to a specific sliding speed according to the control instructions of the host computer. An industrial computer is used to collect friction factor and wear height data in real time through a sensor array during the testing process, and transmit the collected raw measurement data to a host computer with an embedded state decision model in real time. The host computer is also used to perform feature analysis on the original measurement data using the embedded state decision model to determine whether the fluctuation characteristics of the current friction system meet the preset conditions for the current friction system to reach a steady state. If the current friction system is found to have reached a steady state, the steady-state tribological parameters under the current load condition are recorded, and the actuator is instructed to automatically switch to the next load condition in the test sequence until all load conditions are tested. After the test, the result file containing the original measurement data and the steady-state tribological parameters under each load condition is exported.

[0007] Thirdly, embodiments of this application also provide an electronic device, the electronic device including a processor and a memory for storing a computer program; the processor is used to execute the computer program and, when executing the computer program, implement the friction test load adjustment and steady-state identification method based on the embedded model as described in the first aspect or any embodiment of this application.

[0008] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer software program, which, when executed by a processor, implements the friction test load adjustment and steady-state identification method based on an embedded model as described in the first aspect or any embodiment of this application.

[0009] This application provides a method and system for adjusting friction test load and identifying steady state based on an embedded model. In this embodiment, a host computer receives a user-inputted custom test load range and selects multiple test load conditions within the custom test load range using an embedded state decision model built into the host computer, forming a test sequence. Then, the host computer controls the actuator to apply corresponding contact pressure to the friction pair and adjust it to a specific sliding speed according to the test sequence. During the test, the sensor group of the industrial computer collects friction factor and wear height data in real time and transmits the collected raw measurement data to the host computer with the embedded state decision model in real time. Next, the host computer uses the embedded state decision model to perform feature analysis on the raw measurement data to determine whether the fluctuation characteristics of the current friction system meet the preset conditions for the current friction system to reach a steady state. Then, if the current friction system is identified as having reached a steady state, the steady-state tribological parameters under the current test load condition are recorded, and the actuator is instructed to automatically switch to the next test load condition in the test sequence until all test load conditions are completed. Finally, after the test is completed, export the result file containing the original measurement data and steady-state tribological parameters under each load condition.

[0010] In this embodiment, an embedded state decision model embedded in the host computer randomly generates the sequence of loads to be tested, no longer limited to traditional fixed load testing. The test conditions are closer to the complex working conditions in actual production, making the acquired tribological data more practical and accurate. Furthermore, the embedded state decision model analyzes the data collected by the sensors in real time and automatically identifies steady states, eliminating the need for fixed test cycles and avoiding the need to extend test time to cover steady states. This significantly shortens the overall test cycle and reduces repetitive manual operations. Simultaneously, this embodiment can automatically switch the load conditions to be tested and complete the data acquisition and analysis process without manual intervention, improving the efficiency of material friction testing and reducing material consumption and time costs. Finally, a result file integrating the original data and steady-state parameters is exported, providing more comprehensive and accurate data analysis results for subsequent materials science research and analysis, further improving the research efficiency in the field of friction testing. Attached Figure Description

[0011] Figure 1 A schematic flowchart illustrating a friction test load adjustment and steady-state identification method based on an embedded model, provided in an embodiment of this application; Figure 2 A schematic diagram illustrating the relationship between test data timing and load-characteristics, provided in an embodiment of this application; Figure 3 This is a schematic diagram of a friction test load adjustment and steady-state identification system based on an embedded model, provided in an embodiment of this application. Detailed Implementation

[0012] This application provides a method and system for adjusting friction test loads and identifying steady-state states based on an embedded model. The method for adjusting friction test loads and identifying steady-state states based on an embedded model can be applied to a terminal device, which can be a mobile terminal, such as a mobile phone, virtual reality device, tablet computer, laptop computer, desktop computer, wearable device, or other electronic device. For example, the terminal device can be a mobile terminal installed in a material testing device. Alternatively, the terminal device can be a testing device installed in conjunction with a material production line or quality inspection equipment. The terminal device can be a server connected to a cloud service system or a server cluster. The connection can be implemented through hardware circuitry or a communication module.

[0013] The following describes some embodiments of this application in detail with reference to the accompanying drawings. Figure 1This is a flowchart illustrating a friction test load adjustment and steady-state identification method based on an embedded model, provided in an embodiment of this application. The method is applied to a friction test system, which includes a host computer, an industrial computer, and an actuator. Specifically, the method achieves intelligent load adjustment and steady-state identification in friction testing through the collaborative work of the host computer, the industrial computer, and the actuator, combined with an embedded state decision model. For example, the host computer is built as a control unit based on the LabVIEW engineering platform, and the built-in embedded state decision model is embedded in the host computer as an executable program module. First, the host computer receives a user-inputted custom test load range. The embedded state decision model selects multiple test load conditions within this range, forming an ordered test sequence. The host computer interacts with the industrial computer via the ADS (Automation Device Specification) communication interface, sending the load settings in the test sequence to the industrial computer. The industrial computer, acting as an EtherCAT (Ethernet for Control Automation Technology) master station, connects to actuators such as servo controllers and pneumatic devices via couplers and I / O terminal modules, and is also connected to a sensor group consisting of force sensors, displacement sensors, and temperature sensors. Upon receiving the command, the pneumatic device applies the corresponding contact pressure to the friction pair (metal shaft and test sample) according to the current load conditions. The servo motor adjusts its speed via a gearbox, causing relative sliding at a specific speed on the friction pair surfaces, thus initiating the friction test under that load condition. During the test, the force sensor collects friction factor data in real time, and the displacement sensor collects system wear height data in real time. These raw measurement data are processed by an amplifier and transmitted to an industrial computer via EtherCAT, then fed back to the host computer. The embedded state decision model in the host computer performs real-time feature analysis on the raw data, determining whether the data fluctuation characteristics meet the preset steady-state conditions. If they do, the test is determined to have reached a steady state under the current load condition, and the host computer then issues a command to control the actuator via the industrial computer to switch to the next load condition in the test sequence, repeating the above test process. After all load conditions have been tested, the system automatically exports a TDMS format file containing the raw measurement data and steady-state tribological parameters under each load condition, completing the entire test process.

[0014] The aforementioned friction testing system includes a host computer, an industrial computer, and actuators. For example, the host computer can serve as the control and data processing center of the friction testing system, such as by developing its own control program based on the LabVIEW engineering platform, embedding an embedded state decision model in the form of executable program modules. The main functions of the host computer include, but are not limited to: receiving user-inputted custom test load ranges; selecting multiple test load conditions within this range to form a test sequence through the embedded state decision model; exchanging data with the industrial computer via the ADS communication interface; issuing setpoints; and receiving measurement data and status information. An asynchronous communication link is established between the host computer and the industrial computer via the ADS communication protocol for non-real-time configuration parameter issuance and result data upload. Simultaneously, the host computer also performs real-time processing and analysis of the received raw data, determines whether the friction system has reached a steady state, and then instructs the switching of load conditions. After the test, it automatically exports a .tdms format file containing the raw measurement data and steady-state tribological parameters under each load.

[0015] The industrial computer, acting as an EtherCAT master, establishes a connection with field devices via couplers and I / O terminal modules. The industrial computer receives setting commands from the host computer and transmits them to relevant actuators and sensors. Simultaneously, it collects sensor signals after amplification and then feeds the data back to the host computer via EtherCAT communication. For example, as an EtherCAT bus master, the industrial computer connects to the actuators and sensor acquisition modules acting as slaves via real-time Ethernet for millisecond-level real-time control command transmission and synchronous sensor signal acquisition. Furthermore, the industrial computer also communicates with servo controllers acting as EtherCAT slaves to indirectly control motor speed, serving as a crucial data transmission and command forwarding hub connecting the host computer and field devices.

[0016] Alternatively, the host computer and the industrial computer exchange data via an ADS communication interface. The industrial computer acts as the master station and connects to the slave devices in the actuator via an EtherCAT bus.

[0017] The actuator mainly consists of a single-axis servo motor, a gearbox, a pneumatic device, a servo controller, and a friction pair (metal shaft and test sample). The servo motor operates under the control of the servo controller, and its speed is adjusted by the gearbox, driving the metal shaft in the friction pair to rotate, causing relative sliding at a specific speed between the metal shaft and the surface of the test sample. The pneumatic device receives control signals from the I / O terminals to increase or decrease the contact pressure, thereby applying pressure to the friction pair corresponding to the load conditions to be tested, providing the necessary load and relative sliding conditions for tribological testing.

[0018] Optionally, the actuator includes a servo motor, a pneumatic device, and a mechanical transmission component. The industrial computer controls the rotational speed of the servo motor via a servo controller and sends control signals to the pneumatic device via I / O terminals to adjust the contact pressure of the friction pair.

[0019] The following reference Figure 1 This paper introduces a method for adjusting friction test loads and identifying steady-state conditions based on an embedded model, and details the following steps involved in this method: Step S101: Receive the user-inputted custom test load range through the host computer, and select multiple test load conditions within the custom test load range using the embedded state decision model built into the host computer to form a test sequence.

[0020] In this embodiment, the custom test load range is a range of load values ​​input by the user on the host computer based on specific tribological research needs and the load conditions that the friction pair may bear in actual application scenarios. This provides a load selection boundary that fits the actual working conditions, avoiding the problem of traditional fixed loads being out of touch with actual usage scenarios. For example, when studying the performance of friction materials in mechanical transmission components, if the user knows that the load borne by the friction pair during actual operation is between 1MPa and 6MPa, then 1MPa-6MPa can be input as the custom test load range into the host computer.

[0021] The embedded state decision model can be implemented as an executable program module built into the host computer's LabVIEW program. It has the core functions of randomly selecting loads and subsequent data processing and analysis, which is the key to realizing intelligent testing. Its design logic is closely aligned with the needs of friction testing, and it can scientifically select loads and determine the test state within a user-defined range. For example, within a user-defined range of 1MPa-6MPa, the embedded state decision model will use a preset random algorithm to select multiple representative load values ​​without repetition, and can then determine whether the test has reached a steady state based on the data collected by the sensors.

[0022] It should be noted that the underlying mathematical formulas or algorithms on which the embedded state decision model relies can be determined according to the testing requirements of the actual testing scenario, or selected based on the material properties of the material under test, or based on the historical analysis method used for the material under test; there are no restrictions here.

[0023] The model parameters of the embedded state decision model can be updated online based on the response characteristics of different friction pairs or historical test results. Online updates are a dynamic adaptation method that does not require interruption of the testing process, suitable for fine-tuning of similar friction pairs or long-term testing scenarios for a single friction pair. Specifically, based on the real-time response characteristics of the current friction pair (such as the intensity of friction factor fluctuations and wear rate changes), and combined with the statistical deviation from historical test data, parameter optimization is completed through gradient correction or weighted iteration. During testing, the model parameters continuously capture real-time response data. When there is a deviation between the data and the characteristics corresponding to the initial preset parameters, the iterative correction mechanism is automatically initiated to ensure that load selection and steady-state determination always adapt to the current friction pair characteristics.

[0024] Alternatively, model parameters can be migrated and configured based on the response characteristics of different friction pairs or historical test results. This approach is a highly efficient adaptation solution for reusing characteristics across friction pairs, suitable for batch testing or rapid adaptation of new friction pairs. Its core is to achieve efficient reuse of mature parameters through feature matching and coefficient correction. For example, first, the characteristic difference coefficients between the source and target friction pairs in dimensions such as material hardness and surface roughness are calculated. Then, through scaling, offsetting, or weighted correction, the mature parameters of the source friction pair are specifically adjusted to generate the adaptation parameters for the target friction pair. Extensive pre-testing of new friction pairs is unnecessary; only minor adjustments through subsequent verification tests are required to meet usage needs, significantly reducing pre-testing costs and improving testing efficiency.

[0025] The underlying mathematical formulas or algorithms upon which the embedded state decision model relies can be any one or a combination of the following: uniform random distribution algorithm, Latin hypercube sampling algorithm, trend judgment algorithm (linear regression analysis), and moving standard deviation judgment algorithm. Among these, the probability formula based on uniform distribution randomly generates multiple non-repeating load values ​​within a user-defined load range (e.g., a MPa-b MPa), ensuring that the load selection covers the entire set interval and avoids concentration in a localized area, thus reflecting the randomness of loads in actual working conditions. Based on the uniform distribution probability density formula, it ensures consistent selection probabilities within each load interval, making it suitable for test scenarios requiring comprehensive coverage of the working conditions. The Latin hypercube sampling algorithm, through stratified sampling logic, divides the custom load range into multiple equally probable intervals, randomly selecting a load value within each interval. This ensures both load randomness and effectively reduces sampling bias, ensuring uniform coverage of the load range by the test sequence, making it particularly suitable for scenarios where a limited number of loads are needed to reflect the overall tribological characteristics. The trend-based judgment algorithm uses a linear regression formula to fit friction factor or wear height data over a period of time to determine whether there is a significant upward or downward trend in the data (data during the break-in period often shows a trend, while steady-state data shows a gentler trend). The moving standard deviation judgment algorithm calculates the standard deviation of multiple consecutive data points in real time. If the standard deviation of M consecutive data points is within a preset stable range without significant abrupt changes, the system is considered to have reached a steady state. This algorithm combines the sliding window concept with the standard deviation formula, enabling real-time response to data changes and avoiding misjudgments of steady state due to single fluctuations, thus meeting the needs of real-time analysis and automatic load switching during testing.

[0026] It should be noted that the above algorithm is only an example. This application does not limit the specific form of the load screening algorithm or the steady-state determination algorithm. As long as it can generate a representative test sequence within the load range set by the user and realize the online identification of the steady state of the friction system, it falls within the protection scope of this application.

[0027] The load condition to be tested is a specific load value selected by the embedded state decision model from the custom test load range. Each value corresponds to an independent test condition. During the test, the corresponding contact pressure needs to be applied to the friction pair for each condition to obtain tribological data under that load. For example, specific values ​​such as 2.3MPa, 3.7MPa, and 5.1MPa selected by the embedded state decision model from the range of 1MPa to 6MPa become a load condition to be tested, and each condition must complete a complete friction test procedure.

[0028] Understandably, in step S101, an intelligent screening strategy based on parameter space coverage optimization can be combined, and an embedded state decision model can be used to select multiple load conditions to be tested within a custom test load range to form a test sequence. Specifically, a two-dimensional load parameter space with contact pressure as the horizontal axis and sliding speed as the vertical axis can be constructed first. Then, initial sampling can be performed according to the set sampling density. Subsequently, by calculating the Euclidean distance and force vector sum between candidate load points, points with excessively high local aggregation can be eliminated or shifted until the load point distribution meets the preset uniformity index. The final generated test sequence presents a statistically significant uniform distribution in the two-dimensional parameter space, ensuring that each load condition can represent a type of real working condition, achieving the most comprehensive coverage of tribological characteristics with the fewest number of tests.

[0029] The test sequence is an ordered set of multiple load conditions to be tested, arranged in a random order according to the embedded state decision model. This provides a clear sequence for the host computer to control the actuators to switch loads, ensuring that the tests are orderly and comprehensively cover the user-defined load range. For example, after the embedded state decision model selects five load conditions to be tested—2.3MPa, 3.7MPa, 5.1MPa, 1.8MPa, and 4.5MPa—the test sequence is formed in the order of 3.7MPa, 1.8MPa, 4.5MPa, 2.3MPa, and 5.1MPa. The host computer will then control the pneumatic device to switch the contact pressure sequentially according to this sequence, gradually completing the test under all load conditions.

[0030] In step S101, the host computer first receives the user-inputted custom test load range. Specifically, the host computer uses a LabVIEW engineering platform to build a visual interactive interface with a dedicated load range parameter setting module, providing users with an intuitive input entry. Users can input the minimum and maximum load values ​​in the host computer based on specific tribological research needs and the load range that the material under test may experience in actual application scenarios. They can also set the accuracy requirements for the load values, such as to one or two decimal places, to ensure that the selected test load conditions meet the accuracy requirements of the test data. After receiving the user-input parameters, the host computer automatically verifies the data validity, checking whether the minimum value is less than the maximum value and whether the value is a reasonable positive number. If there are logical errors or invalid values, the interface will display a prompt message in real time to guide the user to correct them. After successful verification, the host computer stores the custom test load range, providing a clear boundary basis for the subsequent embedded executable program module to randomly select multiple test load conditions and form an ordered test sequence, ensuring that the test load always conforms to the user's preset research scenario.

[0031] Optionally, candidate test load ranges can be automatically configured based on the test target, and users can customize settings within these ranges. Specifically, the LabVIEW program on the host computer has a built-in database of load ranges corresponding to different test targets. The numerical ranges in this database are pre-calibrated and stored based on the actual working conditions of various friction test scenarios, historical test data, and material application characteristics. Users select a specific test target in the host computer's visual interface, such as friction material wear resistance testing or transmission component friction stability testing. The host computer will then call the corresponding database through an embedded executable program module (such as a Python script) to automatically configure and display candidate test load ranges matching the test target. This candidate range fully matches the actual load requirements of the target scenario, preventing users from setting unreasonable ranges due to unfamiliarity with the working conditions. Users can further adjust the minimum, maximum, and numerical accuracy of the load within this candidate range based on their research focus, or they can directly use the candidate range configured by the system. The host computer will perform a second verification on the range finally confirmed by the user to ensure that it falls completely within the automatically configured candidate range and that the numerical logic is compliant. After the verification is passed, the custom range is stored to provide a scientific boundary for the random selection of subsequent test load conditions. This not only reduces the user's operating threshold but also ensures the rationality and practicality of the test range.

[0032] As an optional embodiment, in step S101, multiple load conditions to be tested are selected within a custom test load range using an embedded state decision model built into the host computer to form a test sequence. This includes: obtaining the custom test load range, which includes a contact pressure range and a sliding speed range; calling an executable program module built into the host computer program to randomly sample within the contact pressure range and the sliding speed range to generate multiple combined coordinate points representing different combinations of pressure and speed; and arranging the multiple combined coordinate points according to a preset rule or random order to generate a test sequence.

[0033] For example, the host computer utilizes the LabVIEW engineering platform to build a clearly defined visual interactive interface. This interface includes independent contact pressure setting and sliding speed setting modules, providing users with precise and convenient input channels. Users can input the minimum and maximum pressure values ​​in the contact pressure setting module, and set the pressure accuracy as needed (e.g., 0.1 MPa, 0.01 MPa, etc.), based on the actual application scenario, material properties, and research objectives of the friction pair under test (e.g., metal shaft and test sample). Simultaneously, the minimum and maximum speed values, as well as the speed accuracy (e.g., 0.1 m / s, 0.05 m / s, etc.), are input in the sliding speed setting module, thus fully defining the required contact pressure and sliding speed ranges for the test.

[0034] Furthermore, after receiving the various parameters input by the user, the host computer automatically initiates dual validity checks. On one hand, it checks the numerical logic within a single range, determining whether the minimum value is less than the maximum value and whether the input value is a positive number that meets the actual testing requirements, avoiding logical contradictions. On the other hand, it checks the rationality of the pressure and speed parameters, combining the preset thresholds of the system hardware (pressure adjustment range of the pneumatic device, speed adjustment capability of the servo motor and gearbox) to determine whether the user input is within the achievable range of the device, preventing exceeding hardware limits. If there are numerical logic errors or the device exceeds its capabilities, the interface will display clear prompts in real time, clearly indicating the error type and correction direction, guiding the user to adjust the parameters.

[0035] Finally, after the user corrects and confirms the input, the host computer verifies and stores the custom test load range (including the contact pressure range and the sliding speed range), synchronizing it to the executable program module built into the LabVIEW program. This provides a clear and compliant boundary basis for subsequent scripts to randomly sample and generate combined coordinate points of pressure and speed within the two ranges, forming a test sequence. This ensures that the subsequently generated test load conditions not only meet the user's research needs but can also be implemented by the system hardware. For example, after verifying and storing the custom test load range, the host computer first organizes the contact pressure range (including minimum, maximum, and precision parameters) and the sliding speed range (including minimum, maximum, and precision parameters) into a standardized data structure (such as key-value pairs or arrays) to ensure that the parameter format matches the reading requirements of the embedded executable program module. Since the executable program module is integrated into the LabVIEW program, the host computer directly passes the organized load range parameters as input data to the executable program module through the LabVIEW built-in Python call interface. After receiving the parameters, the script automatically parses and stores them, forming the basis for subsequent random sampling to generate combined coordinate points. The entire synchronization process is completed by the internal data interaction mechanism between LabVIEW and the embedded script, without the need for additional external communication links.

[0036] In practical applications, optionally, the host computer verifies the custom test load range based on the verification logic preset in the LabVIEW program and the system hardware parameter thresholds, completing the process step by step in a dimensional manner. The specific process is as follows: First, numerical logic verification is initiated, verifying the rationality of the user-input contact pressure range (minimum and maximum values) and sliding speed range (minimum and maximum values) one by one. It is determined whether the minimum value in each range is less than the maximum value to avoid logical contradictions. At the same time, it is verified whether the input values ​​are positive numbers that meet the test requirements, excluding invalid values ​​such as zero or negative numbers, ensuring that the load and speed conform to the physical scenario of actual friction testing. Subsequently, hardware adaptation verification is performed. The host computer program pre-stores the performance limit parameters of the system hardware, including the pressure adjustment range of the pneumatic device and the speed adjustment limit of the combination of the servo motor and the gearbox. During verification, the user-input pressure range is compared with the rated adjustment range of the pneumatic device, and the sliding speed range is compared with the actual controllable speed range of the servo motor, ensuring that the range set by the user does not exceed the actual operating capability of the hardware device, avoiding equipment failure or test failure due to parameter exceeding limits. If the user sets numerical precision parameters, a precision rationality check will also be performed to determine whether the precision set by the user (such as the pressure precision can be set to 0.001MPa, and the speed precision can be set to 0.0001m / s) is within the measurement and control precision range of the friction test system hardware. This is to avoid the situation where the precision setting is too high, which may cause the subsequent test data to fail to meet the preset precision requirements or result in data distortion.

[0037] During the verification process, if any dimension fails, the host computer's visual interface will display a clear prompt in real time, clearly informing the user of the error type (such as numerical logic contradiction, exceeding hardware limits, unreasonable precision settings, etc.) and the corresponding correction direction, guiding the user to adjust the input parameters. If all dimensions pass the verification, the host computer will store the custom test load range and synchronize it to the embedded executable program module, providing a compliance basis for the generation of subsequent test sequences.

[0038] Optionally, it is assumed that the load conditions to be tested include at least contact pressure and sliding speed. Based on this, in step S101, random sampling is performed within the contact pressure range and sliding speed range to generate multiple combined coordinate points representing different combinations of pressure and speed. This includes: determining the sampling density of the coordinate points to be tested according to the custom test load range, and constructing a two-dimensional load parameter space with contact pressure as the horizontal axis and sliding speed as the vertical axis; randomly sampling the two-dimensional load parameter space according to the sampling density to obtain preliminary candidate coordinate points; calculating the distance metric operator between each candidate coordinate point; combining the distance metric operator to calculate the distribution density evaluation index of each candidate coordinate point in the two-dimensional load parameter space, and redistributing candidate coordinate points with a clustering degree higher than a set clustering degree threshold in a local area according to the distribution density evaluation index, until the point distribution of the candidate coordinate points meets a preset uniformity index, and outputting multiple combined coordinate points so that the finally generated test sequence presents a statistically significant uniform distribution in the two-dimensional load parameter space. Therefore, the above embodiments can ensure the diversity and comprehensive coverage of test load combinations that conform to actual working conditions by constructing a two-dimensional parameter space, random sampling, and uniformity optimization.

[0039] Specifically, in the above embodiments, firstly, the sampling density of the coordinate points to be tested is determined according to the custom test load range. The setting of the sampling density needs to be combined with the research requirements of actual friction testing, equipment testing efficiency, and operating condition coverage requirements. If it is necessary to accurately capture the tribological characteristics under different load combinations, a higher sampling density (i.e., a larger total number of candidate points) can be set. If the focus is on quickly obtaining core operating condition data, the sampling density can be appropriately reduced. The sampling density ultimately corresponds to the preset total number of candidate points. At the same time, a two-dimensional test load parameter space is constructed with contact pressure (P) as the horizontal axis and sliding velocity (V) as the vertical axis. The boundary range of this space is related to the contact pressure range (P) input by the user. min -P max ), sliding speed range (V) min -V max The coordinates are completely consistent, and the distribution boundaries of all candidate coordinate points are clearly defined to ensure that they do not exceed the controllable range of the equipment and the actual working conditions.

[0040] Next, according to the set sampling density, random sampling is performed on the two-dimensional load parameter space. For the transverse contact pressure dimension, at P... min -P max Random values ​​equal to the total number of candidate points are generated within the interval, and all values ​​are uniformly distributed within the interval without significant bias. The same method is used for the vertical axis sliding velocity dimension in V. min -V maxAn equal number of random values ​​are generated within the interval. Then, the random values ​​of the two dimensions are paired one by one, and each pair of (P, V) values ​​is a preliminary candidate coordinate point. All preliminary candidate coordinate points fall within the boundary of the two-dimensional parameter space, forming a preliminary candidate coordinate point set.

[0041] Next, the distance metric operator between each candidate coordinate point is calculated. The distance metric operator is used to characterize the spatial distribution characteristics of the parameters. The initial set of candidate coordinate points is traversed, and for each target candidate point, the straight-line distance (e.g., Euclidean distance) between it and all other candidate coordinate points in the set is calculated one by one. The distance values ​​and spatial relationships of each associated point relative to the target point are fully recorded, providing basic data for subsequent clustering determination. Based on this, the force vector sum of each candidate coordinate point in the two-dimensional parameter space is calculated using the distance metric operator. A local coordinate system is established with each target candidate point as the origin, and the spatial association effects of other candidate points on the target point are transformed into vectors. The vector direction is from other candidate points to the target candidate point (reflecting the clustering direction of surrounding points on the target point), and the vector magnitude is inversely proportional to the distance metric operator between them (the closer the distance, the larger the vector magnitude, representing a stronger clustering effect). The vectors of all associated points acting on the target point are superimposed to obtain the total force vector sum of that point, and its magnitude directly quantifies the degree of clustering in the local area where that point is located.

[0042] In addition to the sum of force vectors of each candidate coordinate point in the two-dimensional parameter space, the distribution density evaluation index can also be one or more of the following parameters: local point density, mean nearest neighbor distance, global coordinate dispersion coefficient, grid point density variance, and PV value interval distribution deviation rate. For example, first, the local clustering degree of a single candidate point is determined by the sum of force vectors and magnitudes and the local point density, marking the high-clustering points that need adjustment. Then, the global uniformity after adjustment is verified by the mean nearest neighbor distance, global coordinate dispersion coefficient, and grid point density variance, ensuring that local adjustments do not affect the overall distribution. Finally, the PV value interval distribution deviation rate is used for engineering condition adaptation, ensuring that the adjusted candidate points meet the actual needs of friction testing and avoiding situations where the points are mathematically uniform but deviate from actual engineering conditions.

[0043] Subsequently, the force vector and modulus of each candidate coordinate point are compared with the preset clustering threshold (set based on the friction test load coverage requirements, reflecting the maximum allowable point density in a local area), and optimization processing is performed on candidate coordinate points with excessive clustering. If the total number of initial candidate points exceeds the preset number of test load conditions, high-clustering points with the largest force vector and modulus are directly removed (prioritizing the removal of points that have the most significant impact on uniformity). If a preset number of candidate points need to be retained, the point is translated to a region with sparse load conditions around it according to the total force vector and the opposite direction of the high-clustering point. The translation distance is set in combination with the two-dimensional parameter space scale and the friction test load gradient requirements, while ensuring that the point is still within the effective range of contact pressure and sliding speed after translation, and does not violate the equipment's rated output and material PV value limits.

[0044] After completing one round of optimization, all remaining candidate coordinate points are re-traversed, and the above process is repeated starting from calculating the distance metric operator. After each iteration, the global uniformity index (such as the global point distance dispersion coefficient, the upper limit multiple of the local area point density, etc.) is checked. When the force vector and magnitude of all candidate coordinate points are lower than the aggregation threshold after iteration, and the global point distribution meets the preset uniformity index, the optimization stops, and the final multiple combined coordinate points are output. These combined coordinate points exhibit a statistically significant uniform distribution in the two-dimensional test load parameter space, corresponding to multiple sets of test load conditions (contact pressure P, sliding velocity V), ensuring that the test covers different load combination scenarios in actual working conditions, improving the comprehensiveness and practicality of the test data, so that the friction test system can automatically approximate the actual working condition test conditions.

[0045] For example, suppose a user is testing a friction material used in mechanical transmission, and the input custom test load range is: contact pressure range of 1MPa-50MPa, and sliding speed range of 0.1m / s-4m / s. Based on the test accuracy requirements, the total number of candidate points corresponding to the sampling density is set to 40, the preset aggregation threshold is 5.0 (vector modulus unit), and the global point distance dispersion coefficient must be less than 0.3. The first step is to construct a two-dimensional test load parameter space with contact pressure as the horizontal axis (1-50MPa) and sliding speed as the vertical axis (0.1-4m / s). The second step involves dividing the contact pressure range into 17 discrete values ​​based on measurement accuracy requirements and randomly and repeatedly selecting 40 values ​​(such as 1MPa, 8MPa, 8MPa, 45MPa, etc.). In the sliding speed range, 8 uniformly distributed values ​​are generated, and 40 values ​​are repeatedly selected from them (such as 0.5m / s, 1m / s, 1.5m / s, 3m / s, etc.). After pairing them one by one, 40 preliminary candidate coordinate points are obtained, such as (1MPa, 0.5m / s), (8MPa, 1m / s), (8MPa, 1.5m / s), (45MPa, 3m / s), etc. The third step involves calculating the Euclidean distance (i.e., the distance metric operator) between each point. It was found that the distances between (2.0MPa, 1.0m / s) and (2.1MPa, 1.1m / s) and (1.9MPa, 0.95m / s) are extremely close. The force vectors and moduli of these three points in this region are 6.2, 5.8, and 5.5, respectively, all exceeding the aggregation threshold of 5.0. The fourth step involves removing (2.0MPa, 1.0m / s) from the list of candidate points, since the total number of preliminary candidate points (40) is close to the number of preset test load conditions (38). The point with the largest force vector and moduli (1.9MPa, 0.95m / s) is then shifted by 0.3MPa pressure and 0.2m / s velocity along the opposite direction of the force (towards the sparse region with a pressure of 3.5MPa and a velocity of 1.5m / s), resulting in the new coordinates (2.2MPa, 1.15m / s). Fifth, after three iterations, the force vector and modulus of all points are recalculated and are all below 5.0. The global point distance dispersion coefficient is 0.25 (less than the preset 0.3), meeting the uniformity index. Sixth, 38 combined coordinate points are output, some of which are as follows: (0.7MPa, 0.3m / s), (1.2MPa, 0.9m / s), (2.1MPa, 1.1m / s), (2.2MPa, 1.15m / s), (3.5MPa, 0.5m / s), (4.8MPa, 1.8m / s), etc. These coordinate points uniformly cover the pressure range of 0.5-5MPa and the velocity range of 0.1-2m / s, corresponding to different load combinations in actual working conditions. Test sequences will be generated based on these points to perform friction tests.

[0046] Optionally, after outputting multiple combined coordinate points, a data smoothing processing model using moving average filtering or Kalman filtering is employed to denoise the raw measurement data containing mechanical vibration noise interference received in real time, extract the friction factor characteristic trend term reflecting the true contact state of the friction pair, and add the friction factor characteristic trend term to the multiple combined coordinate points.

[0047] For example, the data source and processing medium must first be clearly defined. During the test, the force sensor collects raw friction factor data in real time, amplifies it using a corresponding amplifier, connects it to the I / O terminals, and then transmits it to the industrial computer via the EtherCAT bus. The industrial computer then feeds back the raw data to the host computer in real time through the ADS communication interface. The host computer, as the core of data processing, has its LabVIEW program's embedded executable module pre-integrated with two selectable models: moving average filtering and Kalman filtering. Users can choose the appropriate processing model based on the noise characteristics of the mechanical vibration in the test scenario.

[0048] If a moving average filtering model is selected, the size of the sliding window needs to be preset based on the mechanical vibration frequency and testing accuracy requirements of the test equipment. After the host computer receives the raw friction factor data in real time, it sequentially inputs the data into the sliding window according to the time series. The arithmetic mean of all continuous data points within each window is calculated, and this mean is used as the filtered data for the current moment. By continuously sliding the window and repeating the calculation, smoothed friction factor time-series data is obtained, effectively eliminating instantaneous pulse noise caused by mechanical vibration. Subsequently, a trend extraction algorithm is used to separate characteristic trend terms reflecting the true contact state of the friction pair from the smoothed time-series data. These trend terms clearly show the changing pattern of the friction factor over test time, such as rising, falling, or remaining stable.

[0049] If a Kalman filter model is chosen, a data model must first be established based on the physical process of the friction test. Preset initial state values, process noise covariance matrix, and observation noise covariance matrix are required; these parameters can be calibrated based on historical test data or the rated characteristics of the equipment. The host computer inputs the real-time received raw data as observation values ​​into the model. Through a prediction step, it calculates the predicted state value and prediction error covariance at the current moment. Then, through an update step, it combines the observation values ​​to correct the prediction results, obtaining the optimal estimated filtered data, thus achieving adaptive suppression of random vibration noise. Afterward, the friction factor characteristic trend term is extracted from the optimal estimated data to accurately capture the changing patterns of the actual contact state of the friction pair.

[0050] Finally, the characteristic trend terms were bound to the combined coordinate points. Each combined coordinate point originally contained two sets of load parameters: contact pressure and sliding velocity. After binding, these were expanded into a three-dimensional data structure of (contact pressure P, sliding velocity V, and friction factor characteristic trend term). This structured data will provide a reliable basis for the host computer to use the embedded state decision model to determine whether the friction system has reached a steady state, effectively avoiding steady-state misjudgments caused by noise interference, and further improving the practicality and accuracy of the test results.

[0051] For example, assuming a given set of coordinates is (2.5 MPa, 1.0 m / s), during friction testing under this load condition, the raw friction factor data (unit: dimensionless) collected by the force sensor is subject to noise interference due to mechanical vibration. Some continuous raw data are 0.28, 0.36, 0.31, 0.39, 0.33, 0.29, and 0.35. If a moving average filtering model is selected, the preset moving window size is 5. The raw data is input into windows in time series order. The average value of the first window (0.28, 0.36, 0.31, 0.39, 0.33) is 0.334, the average value of the second window (0.36, 0.31, 0.39, 0.33, 0.29) is 0.336, and the average value of the third window (0.31, 0.39, 0.33, 0.29, 0.35) is 0.334, resulting in smoothed time series data of 0.334, 0.336, and 0.334. A trend extraction algorithm identifies a stationary fluctuation trend with a mean of 0.335. This trend is added to the combined coordinate points to obtain three-dimensional data (2.5 MPa, 1.0 m / s, stationary fluctuation trend with a mean of 0.335). If a Kalman filter model is selected, the preset process noise covariance matrix is ​​0.001, and the observation noise covariance matrix is ​​0.005. After inputting the raw data into the model, the optimal estimated data (0.328, 0.334, 0.331, 0.332, 0.330, 0.331, 0.332) is obtained through prediction and update steps. The extracted feature trend term is no obvious upward or downward trend with a fluctuation range of ±0.003. The combined coordinate points after association are (2.5MPa, 1.0m / s, no obvious upward or downward trend with a fluctuation range of ±0.003). After the above two filtering processes, the friction factor feature trend term has eliminated mechanical vibration noise interference and is used to reflect the contact state of the friction pair under this load condition, providing analysis for subsequent judgment on whether the working condition has reached a steady state.

[0052] Step S102: The host computer controls the actuator to apply the corresponding contact pressure to the friction pair and adjust it to a specific sliding speed according to the test sequence.

[0053] Understandably, in step S102, the host computer first reads the current load combination to be executed in the test sequence, which corresponds to the target values ​​of contact pressure and sliding speed. Relying on its own control program written on the LabVIEW engineering platform, it converts these target values ​​into standardized control instructions recognizable by the actuator, including explicit pressure control parameters and motor speed control parameters. Subsequently, through a preset ADS communication interface, the host computer transmits these control instructions to the industrial computer in real time, while simultaneously synchronizing relevant status information of the current test condition to ensure that the industrial computer can accurately receive and understand the test task to be executed.

[0054] As the EtherCAT master station, the industrial computer receives control commands from the host computer and breaks them down according to the system's preset device communication protocol. On one hand, the control signal corresponding to the contact pressure is transmitted to the I / O terminal module via a coupler, providing command support for the pneumatic device's operation. On the other hand, the motor speed control signal corresponding to the sliding speed is precisely sent to the servo controller, which acts as the EtherCAT slave station. This achieves precise distribution and efficient forwarding of control commands to different actuators, building a crucial bridge from command reception to execution.

[0055] After receiving pressure control signals from the industrial computer via the I / O terminal module, the pneumatic device automatically adjusts its operating state according to the target contact pressure value contained in the signal. By increasing or decreasing the internal pressure, the pneumatic device transmits the corresponding pressure to the friction pair (i.e., the metal shaft and the test sample), creating a contact pressure that meets the requirements of the current test sequence. Throughout this process, the pneumatic device responds to control signals and adjusts in real time to ensure that the pressure on the friction pair remains stable within the target range, laying the foundation for test accuracy.

[0056] After receiving the speed control command from the industrial computer, the servo controller drives the single-axis servo motor to operate at the speed set in the command. The speed output by the servo motor is adapted by the gearbox, and according to the target sliding speed value required for the test, the motor speed is converted into a specific relative sliding speed required by the friction pair. Subsequently, the metal shaft is driven to rotate through the mechanical transmission component, and the metal shaft and the surface of the test sample generate a relative sliding that meets the requirements of the current load combination (i.e., the current test load conditions), thus completing the precise adjustment of the sliding speed.

[0057] During the test corresponding to the current load combination, the industrial computer continuously receives status data from various sensors, monitoring in real time whether the contact pressure and sliding speed are within the set range. If a slight deviation occurs, the friction test system will automatically fine-tune the working state of the actuator to ensure stable test conditions. When the host computer recognizes that the current working condition has reached a steady state through the embedded state decision model, it will immediately issue a switching command. The industrial computer then retrieves the next load combination in the test sequence (i.e., the next load condition to be tested) and repeats the above command transmission, execution, and adjustment process until all load combinations in the test sequence have been executed.

[0058] Further optionally, in step S102, during the process of the actuator applying pressure to the friction pair, the pressure value is fed back in real time by the force sensor and compared with the set value in a closed loop, and the output force of the positive pneumatic device in the actuator is adjusted in real time by the industrial computer.

[0059] Specifically, during the process of the pneumatic device in the actuator applying pressure to the friction pair, the force sensor continuously collects the actual value of the contact pressure of the friction pair in real time. The collected raw pressure signal is first amplified by a corresponding amplifier and then connected to the system's I / O terminal module. Through the EtherCAT bus communication link, the amplified pressure feedback data is transmitted in real time to the industrial computer, which acts as the EtherCAT master station, ensuring that the industrial computer can obtain the actual pressure value in a timely manner. After receiving the pressure feedback data, the industrial computer performs a closed-loop comparison between the actual pressure value and the pressure setpoint corresponding to the current test sequence. The pressure setpoint is converted from the contact pressure target value issued by the host computer. Through preset comparison logic, the industrial computer accurately calculates the magnitude and direction of the deviation between the actual pressure value and the setpoint, determining whether the current pneumatic device output force meets the test requirements, providing a basis for the generation of adjustment commands. Based on the deviation result obtained from the closed-loop comparison, the industrial computer automatically generates the corresponding adjustment command. This adjustment command is transmitted to the I / O terminal module through a coupler, and then the I / O terminal sends the command to the pneumatic device. If the actual pressure value is lower than the set value, the pneumatic device will respond to the command by increasing the internal pressure output, thereby increasing the pressure applied to the friction pair. If the actual pressure value is higher than the set value, the pneumatic device will reduce the pressure output, thereby reducing the applied pressure. This correction process is repeated in real time until the deviation between the actual pressure value fed back by the force sensor and the set value is within a preset allowable range. This ensures that the pressure on the friction pair remains stable at the contact pressure required by the current test sequence, further improving the accuracy of the load conditions during the test and ensuring the reliability of the test data.

[0060] Step S103: During the test, the friction factor and wear height data are collected in real time by the sensor group of the industrial computer, and the collected raw measurement data is transmitted in real time to the host computer with an embedded state decision model.

[0061] For example, in the test data acquisition step, a force sensor is used to measure the friction factor online, a displacement sensor is used to measure the system wear height online, and a temperature sensor is used to monitor the system temperature status.

[0062] Specifically, during the testing process, a sensor array connected to the industrial computer continuously acquires data in real time. This sensor array primarily includes force sensors, displacement sensors, and temperature sensors, each performing its specific function of monitoring certain parameters. The force sensors directly monitor the mechanical signals generated during the contact friction between the friction pair (e.g., a metal shaft and the test sample), calculating the friction factor data online through preset signal conversion logic to capture the frictional characteristics of the friction pair. The displacement sensors track the relative displacement changes of the friction pair in real time, combining this with initial test parameters to measure the system wear height data online, reflecting the material wear during the test. The temperature sensors continuously monitor the temperature of the testing system, capturing temperature changes in real time to provide data support for determining whether the test environment is stable and whether overheating is affecting the test results.

[0063] The raw signals (mechanical signals, displacement signals, and temperature signals) collected by each sensor must undergo signal preprocessing. First, the weak raw signals are amplified by appropriate amplifiers to ensure signal strength meets the requirements of subsequent transmission and processing, avoiding data distortion due to weak signals. The processed signals are then connected to the system's I / O terminal module and transmitted in real-time to the industrial computer, which acts as the EtherCAT master station, via the EtherCAT bus communication link. After receiving the data, the industrial computer performs preliminary formatting to ensure the data conforms to the host computer's receiving standards. Then, through a preset ADS communication interface, the formatted raw measurement data (including friction factor, wear height, and system temperature) is transmitted in real-time to the host computer, which has an embedded state decision model. The entire transmission process relies on efficient bus communication and interface protocols to ensure the real-time performance and integrity of data transmission, providing reliable data input for subsequent data processing and analysis by the host computer.

[0064] Step S104: The host computer uses an embedded state decision model to perform feature analysis on the original measurement data to determine whether the fluctuation characteristics of the current friction system meet the preset conditions for the current friction system to reach a steady state.

[0065] As an optional embodiment, it is assumed that the fluctuation characteristics include the fluctuation variance of the friction factor data and the slope value of the wear height data changing over time. Based on the above assumption, in step S104, the host computer uses an embedded state decision model to perform feature analysis on the original measurement data to determine whether the fluctuation characteristics of the current friction system meet the preset conditions for the current friction system to reach a steady state, including: First, in the above embodiment of step S104, an embedded state decision model embedded in the host computer is invoked, and the size parameters of the sliding window are adaptively configured according to the dynamic response characteristics of the current friction system. The embedded state decision model is pre-fitted based on historical test data of similar materials and the dynamic characteristics of the friction system. For example, an embedded state decision model in the form of an executable program module (such as a Python script) embedded in the LabVIEW program of the host computer is invoked. This model is pre-fitted based on historical test data of similar materials and the dynamic characteristics of the friction system, and adaptively adjusts the size of the sliding window by identifying the dynamic response characteristics of the current friction system. The embedded state decision model analyzes the raw measurement data transmitted from the industrial computer in real time to determine the current dynamic index status of the system. If the data shows high fluctuation intensity and high rate of change, such as when the friction system is in the break-in period or load switching transition period, the embedded state decision model will automatically configure a small sliding window of 5 to 20 data points to ensure that the instantaneous changes in the system state can be quickly captured, avoiding lag in friction feature capture due to an excessively large window. If the detected data exhibits low fluctuation intensity and low rate of change, such as when the friction system tends towards a steady state, the embedded state decision model automatically configures a large sliding window of 20 to 50 data points. This reduces random interference such as high-frequency mechanical vibrations through statistical averaging of more data points, improving the stability of subsequent fluctuation variance and wear slope calculations. If the detected dynamic index is in the transition range between the two situations, the embedded state decision model dynamically adjusts the window size according to linear interpolation rules, achieving a smooth transition in window size and avoiding distortion in feature calculations due to abrupt window changes. This provides an effective data window adapted to the current state of the friction test system for subsequent feature analysis. The number of data points in the sliding window can be set according to the sampling frequency and the dynamic response characteristics of the friction test system; this application does not limit this.

[0066] Then, in the above embodiment of step S104, the friction factor data and wear height data within the sliding window are extracted from the raw measurement data transmitted in real time. For example, the friction factor data is obtained online by a force sensor, and the wear height data is obtained online by a displacement sensor.

[0067] For example, after configuring the sliding window size, the embedded state decision model extracts the friction factor data and wear height data within the current sliding window from the raw measurement data transmitted in real time to the host computer. The friction factor data is acquired online by a force sensor, and the wear height data is acquired online by a displacement sensor. This raw data is processed by an amplifier and then transmitted to an industrial computer via an I / O terminal module and an EtherCAT bus. It is then fed back to the host computer via an ADS communication interface, ensuring that the extracted data within the window is continuous, complete, and reflects the true operating state of the friction system within the time period corresponding to the current sliding window, providing a reliable data foundation for characteristic parameter calculation.

[0068] Furthermore, in the above embodiment of step S104, the fluctuation variance of the friction factor data within the sliding window is calculated, and the slope value of the wear height data changing with time is calculated. For example, the original friction factor data within the sliding window is denoised to filter out interference signals caused by high-frequency mechanical vibrations during the test, resulting in a smoothed friction factor sequence. Within the sliding window, the average value of all friction factors in the smoothed friction factor sequence is calculated, and the deviation value between each friction factor and the average value is calculated one by one. The average value is then obtained by squaring all deviation values ​​to obtain the fluctuation variance within the sliding window that reflects the overall deviation of the friction factors. Linear regression analysis is performed on the wear height data within the sliding window and the corresponding timestamps to fit a local wear linear equation. This local wear linear equation represents the local correlation law generated by the change of wear height data with time, and the slope of the local wear linear equation is extracted as the slope value of the wear height data changing with time.

[0069] Specifically, firstly, a moving average filtering algorithm or a low-pass filtering algorithm is used to denoise the original friction factor data within the sliding window. For example, by iteratively optimizing the mean of consecutive adjacent data points, or by suppressing high-frequency fluctuations in the data, interference signals caused by high-frequency mechanical vibrations during the test are filtered out, resulting in a friction factor sequence with stable numerical changes. Firstly, relying on the executable program module embedded in the LabVIEW program on the host computer, a preset moving average filtering algorithm or low-pass filtering algorithm is called to perform targeted denoising on the original friction factor data within the sliding window. This original friction factor data is collected online by force sensors, transmitted to an industrial computer via the EtherCAT bus, and then fed back to the host computer via the ADS communication interface, inevitably carrying interference signals generated by high-frequency mechanical vibrations during the test. If a moving average filtering algorithm is selected, the script will continuously take a preset number of adjacent data points according to the time series of the data within the sliding window to calculate the arithmetic mean, gradually smoothing the instantaneous fluctuations in the original data through iterative calculations. If a low-pass filtering algorithm is selected, a preset frequency threshold is used to suppress high-frequency fluctuations in the data above that threshold, retaining only the low-frequency effective signal reflecting the true changes in the friction factor. Both algorithms can effectively filter out interference from high-frequency mechanical vibrations, and finally output a smooth friction factor sequence with stable numerical changes that closely matches the actual friction state of the friction pair, providing an accurate data foundation for subsequent fluctuation variance calculation.

[0070] After obtaining the smoothed friction factor sequence, the executable program module embedded in the host computer automatically initiates the fluctuation variance calculation process. Specifically, it iterates through all smoothed friction factor data within the sliding window, calculates the overall average of the sequence using the arithmetic mean formula, and uses this as a benchmark value to measure the central tendency of the friction factors. Each friction factor data point in the sequence is extracted one by one, and its deviation from the overall average is calculated to accurately capture the degree of deviation of individual data points relative to the benchmark. All calculated deviation values ​​are squared to eliminate the mutual cancellation effect of positive and negative deviations, while amplifying the weight of larger deviations. Finally, all squared deviation values ​​are summed and divided by the total number of friction factor data points within the sliding window to obtain the fluctuation variance of the friction factors within that sliding window. The entire calculation process is automated by the executable program module without manual intervention, ensuring the accuracy and efficiency of the calculation results. This fluctuation variance directly quantifies the overall deviation of the friction factors and is one of the core characteristic parameters for judging the stability of a friction system.

[0071] For the wear height data within the sliding window, the executable program module embedded in the host computer extracts the slope value through linear regression analysis. First, the wear height data within the sliding window is paired with the corresponding acquisition timestamps to form multiple sets of (time, wear height) data samples. This wear height data is acquired online by displacement sensors, amplified, transmitted via EtherCAT bus, and formatted before being fed back to the host computer, accurately reflecting the wear changes of the friction pair during the test. Then, the script calls the linear regression algorithm to fit these data samples, constructing a local wear linear equation (in the form y=kx+b, where y is the wear height, x is the time, k is the slope, and b is the intercept). This linear equation accurately describes the local correlation law of wear height change with time within the sliding window, eliminating the influence of local data fluctuations caused by random factors. Finally, the slope k is extracted from the fitted local wear linear equation. This slope value is the slope value of the wear height data change with time within the current sliding window, intuitively reflecting the wear rate at this stage, providing crucial wear state information for subsequent judgment of whether the friction system has entered a steady state.

[0072] Finally, in the above embodiment of step S104, it is determined whether the fluctuation variance is less than a preset stability threshold and whether the slope value enters a preset linear wear reference range. The stability threshold is preset based on the inherent friction characteristics of the friction pair material, historical test data of similar materials, and the allowable range of friction fluctuations in actual application scenarios. The linear wear reference range is preset based on the rated wear characteristics of the friction pair material, the variation law of wear height over time in multiple sets of predictive tests, and the normal wear range under actual working conditions. If the fluctuation variance is less than the stability threshold and the slope value falls within the preset linear wear reference range, and satisfies the preset convergence criterion within a continuous preset sliding window period, then it is determined that the current friction system meets the preset conditions for reaching a steady state.

[0073] For example, the embedded state decision model compares the calculated variance of the friction factor fluctuation with a preset stability threshold, and simultaneously matches the slope of the wear height change over time with a preset linear wear reference interval. If the variance of the friction factor fluctuation is less than the stability threshold, and the slope of the wear height remains constant within the linear wear reference interval, it indicates that the friction characteristics of the current friction system tend to stabilize, and the wear state conforms to the normal law under steady state. At this time, it is determined that the current friction system meets the preset conditions for reaching steady state. If any condition is not met, it is determined that the system has not reached steady state, and the model will continue to repeat the above feature analysis process based on subsequent real-time data until the preset steady-state conditions are met. It is worth noting that "remaining constant" here can refer to changes within a set fluctuation range, not a static value. This set fluctuation range can be set according to the actual application scenario.

[0074] It is worth noting that the stability threshold is set based on the actual needs of friction testing and the reliability of the data. First, based on the inherent frictional characteristics of the friction pair materials, the friction pair consists of a metal shaft and a test sample. The inherent properties of different materials, such as hardness, surface roughness, and intermolecular forces, directly determine their basic level of frictional stability. For example, metal materials with higher hardness and smoother surfaces experience more uniform force during friction, resulting in naturally smaller fluctuations in the friction factor. Conversely, test samples with rough surfaces or uneven materials are prone to larger fluctuations. Therefore, it is necessary to determine a reasonable range for the fluctuation of the friction factor based on the inherent properties of the two materials, providing a core physical basis for threshold setting. Second, relying on historical test data of similar materials, by integrating past friction factor monitoring data of similar friction pair materials under different test conditions, statistical analysis methods are used to extract the fluctuation characteristics of the data and calculate the common range and critical value of the friction factor variance in historical tests. For example, the variance distribution of the friction factor of similar materials under near-actual operating loads is statistically analyzed, and the upper limit of the 95% confidence interval in the distribution is taken as a reference benchmark to ensure that the stability threshold has sufficient historical data support and avoids subjective setting. Finally, considering the permissible range of friction fluctuations in practical application scenarios, the acceptable limits of friction fluctuations are determined based on the actual usage scenarios of the friction pair (such as mechanical transmission components, static friction structures, etc.). If the friction pair is used in a precision transmission system with extremely high requirements for operational stability, even minor friction fluctuations in real-world scenarios may lead to a decrease in equipment accuracy; therefore, a relatively strict stability threshold needs to be set. If it is used in ordinary mechanical structures, the tolerance for fluctuations is relatively high, and the threshold can be appropriately relaxed to ensure that the threshold is highly matched with the actual application requirements, making the steady-state determination results more practically valuable.

[0075] The calibration of the linear wear reference range can be based on material properties, predictive test data, and actual working conditions to ensure accurate identification of normal wear states under steady-state conditions. Firstly, based on the rated wear characteristics of the friction pair materials, each friction material has its rated wear resistance performance indicators, including core parameters such as rated wear rate and service life. These parameters are determined by the material's composition and manufacturing process. For example, high-strength wear-resistant alloys have lower rated wear rates, while ordinary metal materials have relatively higher rated wear rates. The linear wear reference range needs to be defined with the material's rated wear rate as the core benchmark, delineating a wear slope range that conforms to the material's inherent wear resistance characteristics, ensuring that the range does not exceed the material's normal wear capacity boundary. Secondly, referring to the wear height variation over time in multiple sets of predictive tests, before formal testing, multiple sets of predictive tests under different loads and speeds are conducted on the target friction pair. Wear height and corresponding timestamp data are collected in real time, and the wear slope of each set of predictive tests is obtained through linear regression analysis. The wear slopes from multiple sets of predictive tests were statistically analyzed to determine their distribution characteristics. The slope range under normal wear conditions was extracted (excluding steep slopes caused by abnormal wear such as overload). This data served as the core support for the linear wear reference range, ensuring that the range covers common wear rates under steady-state conditions. Thirdly, combining the normal wear range under actual operating conditions, the working environment of the friction pair in real-world applications (such as load fluctuations, temperature changes, and operating time) was analyzed in depth to investigate the normal wear rate range of the friction pair under actual operating conditions. For example, in actual operating conditions, the friction pair operates under medium loads for extended periods, and its normal wear rate is similar to the wear rate under the corresponding load in the predictive tests. Therefore, the range of normal wear rates statistically derived from actual operating conditions was used as the boundary to correct the range obtained from the predictive tests. This ensures that the linear wear reference range not only conforms to the material's inherent characteristics but also accurately matches the normal wear state in actual use, effectively distinguishing between normal and abnormal wear during steady-state determination and improving the accuracy of the determination results.

[0076] Figure 2 This is a schematic diagram illustrating the relationship between the time series of test data and the load-characteristic relationship. Figure 2 In the diagram, the scatter plot on the left uses sliding speed (m / s) as the horizontal axis and friction factor as the vertical axis. Each scatter point corresponds to a set of steady-state data (sliding speed, friction factor), reflecting the correlation of friction characteristics under different load conditions. The time series curve on the right uses test time (hours) as the horizontal axis and wear height as the vertical axis, showing the trend of wear height changes in different test stages (break-in period, steady-state period), intuitively reflecting the evolution of wear over time. Therefore, the scatter plot on the left can be used to verify the rationality of the uniform load distribution in step S101, show the variation law of friction factor under different working conditions, and provide an intuitive reference for subsequent data backtracking and analysis of material tribological properties. The time series curve on the right can be used to verify the necessity of steady-state identification in step S104, distinguishing between rapid wear during the break-in period and linear wear during the steady-state period.

[0077] In another optional embodiment, to further reduce the impact of equipment operating condition fluctuations (such as sudden temperature changes, load shifts, and speed jitters) on steady-state judgment and avoid steady-state misjudgments, the fluctuation characteristics may further include: auxiliary state parameters, which include any one or a combination of system temperature, normal load closed-loop error, and sliding speed fluctuation amplitude. Specifically, the system temperature is collected in real time by a temperature sensor of the contact area of ​​the friction pair and the ambient temperature of the test system, directly reflecting the balance between frictional heat generation and system heat dissipation. The normal load closed-loop error is calculated in real time by an industrial computer of the deviation between the actual normal pressure value fed back by the force sensor and the set pressure value of the current test sequence, quantifying the stability of the pneumatic device load closed-loop adjustment. The sliding speed fluctuation amplitude is calculated by an industrial computer based on the servo motor speed feedback value, calculating the difference between the maximum and minimum sliding speed per unit time, characterizing the stability of the relative sliding speed of the friction pair.

[0078] Based on this, in step S104, the host computer uses an embedded state decision model to perform feature analysis on the original measurement data to determine whether the fluctuation characteristics of the current friction system meet the preset conditions for the current friction system to reach a steady state. This also includes: using the embedded state decision model to determine whether the current friction system has reached a steady state based on the fusion convergence state of the friction factor data, the wear height data, and the auxiliary state parameters. For example, the determination conditions for the core parameters remain unchanged (friction factor fluctuation variance < stability threshold, wear height slope constant within the linear wear reference interval), and the convergence conditions for each auxiliary state parameter are pre-calibrated based on the equipment hardware characteristics, friction pair material properties, and actual working condition requirements, as detailed below: Firstly, the convergence condition for system temperature can be that the rate of change of the average system temperature within the sliding window is less than a preset temperature change rate threshold, and the temperature fluctuation range is within a preset temperature stability range. In steady state, the friction system achieves dynamic equilibrium between frictional heat generation and system heat dissipation, with no significant upward or downward trend in temperature and minimal instantaneous fluctuations. This threshold or range needs to be set in conjunction with the thermal expansion coefficient, heat resistance characteristics, and allowable temperature range of the friction pair materials under actual operating conditions (e.g., the temperature stability range of high-temperature wear-resistant materials can be appropriately widened).

[0079] Secondly, the convergence condition for the normal load closed-loop error can be that the root mean square value of the normal load closed-loop error within the sliding window is less than a preset error threshold. In steady state, the closed-loop regulation of the pneumatic device has reached a stable state, and the deviation between the actual pressure and the set pressure will remain within a very small range. Using the root mean square value instead of the single deviation value for judgment can effectively eliminate the influence of instantaneous random deviations and more accurately quantify the overall stability of the load regulation.

[0080] Third, the convergence condition for the sliding speed fluctuation amplitude can be that the sliding speed fluctuation amplitude within the sliding window is continuously less than the preset speed fluctuation threshold. In steady state, the servo motor speed regulation is stable, and the relative sliding speed of the friction pair has no obvious jitter. This threshold is set based on the control accuracy of the servo system and the requirements of actual working conditions for sliding speed stability (such as the friction test of transmission components, where the speed fluctuation threshold needs to be strictly limited).

[0081] The fusion convergence state of the aforementioned friction factor data, wear height data, and auxiliary state parameters can be set according to the actual scenario. For example, the determination of the fusion convergence state has a high degree of scenario adaptability, and the combination selection of auxiliary state parameters and the calibration of convergence thresholds can be tailored to the actual application conditions of the friction pair. For test scenarios with high temperature and high frictional heat generation, system temperature can be selected as the core auxiliary parameter, and the calibration threshold of the temperature stability range can be appropriately relaxed. For friction testing of precision mechanical transmission components, where the stability requirements of load and speed are stringent, normal load closed-loop error and sliding speed fluctuation amplitude can be selected as auxiliary parameters, and the convergence thresholds of these two parameters can be tightened. For complex and variable actual working conditions in industrial sites, requiring a comprehensive characterization of the system's operating state, three full auxiliary parameters can be selected to achieve a more comprehensive fusion convergence determination, ensuring that the steady-state data obtained from the test matches the actual application scenario.

[0082] Furthermore, satisfying the preset convergence criteria within a continuous preset sliding window period means that after pre-calibrating the number of continuous sliding window periods based on the dynamic response characteristics of the friction system, historical test data of similar materials, and actual test accuracy requirements, the core fluctuation characteristic parameters of the friction system must continuously and stably meet the steady-state judgment conditions within this continuous period, rather than being a random coincidence within a single window. This avoids steady-state misjudgments caused by local data fluctuations. Specifically, within the calibrated continuous sliding window period, the variance of the friction factor data must always remain below the preset stability threshold, and the slope of the wear height data over time must also continuously fall within the preset linear wear reference range. Neither of these parameters should exceed the threshold or range, nor should they exhibit obvious upward or downward trends. If auxiliary state parameters such as system temperature, normal load closed-loop error, and sliding speed fluctuation amplitude are introduced during the judgment process, these parameters must also continuously meet their respective convergence conditions within this continuous period, remaining in a stable state overall. Meanwhile, the convergence criterion also includes restrictions on the magnitude of parameter changes between adjacent sliding windows. It requires that the difference between the core parameter and the auxiliary parameter within consecutive windows be within a preset, small range, ensuring that the friction system's state continuously and stably tends towards a steady state, rather than experiencing sudden parameter changes. This guarantees the authenticity and reliability of the steady-state identification results. The above is merely an example, and this application does not limit the convergence criterion used in practical applications.

[0083] Step S105: If the current friction system is identified as having reached a steady state, the steady-state tribological parameters under the current load condition are recorded, and the actuator is instructed to automatically switch to the next load condition in the test sequence until all load conditions are tested.

[0084] As an optional embodiment, in step S105, if the current friction system is identified as having reached a steady state, the steady-state tribological parameters under the current load condition are recorded. Specifically, when the host computer identifies that the current friction system has reached a steady state through the embedded state decision model, it immediately initiates the recording process of steady-state tribological parameters. The recording work is performed by the host computer as the core execution entity, relying on the Python script embedded in the LabVIEW program to automatically complete data processing and storage. The recorded steady-state tribological parameters cover multiple types of key information, including basic monitoring parameters, such as the mean friction factor after smoothing within the sliding window, which intuitively reflects the core friction characteristics of the friction pair under steady state; the steady-state wear height corresponding to the current working condition, reflecting the degree of wear of the material in the stable friction stage; and the average temperature of the system during the steady state, providing a basis for analyzing the influence of temperature on friction performance. Simultaneously, it also includes characteristic analysis parameters, such as the previously calculated friction factor fluctuation variance and the slope value of the wear height change over time, which quantify the stability and wear rate under steady state. All parameters are associated with the current load conditions (contact pressure, sliding speed) and are first stored in the host computer's preset steady-state database to form structured data entries, providing a regular data foundation for subsequent data export and analysis, and ensuring that the steady-state characteristics under each load condition are completely preserved.

[0085] As an optional embodiment, in step S105, the actuator is instructed to automatically switch to the next load condition to be tested in the test sequence. Specifically, a load switching command is sent from the host computer to the industrial computer; after receiving the load switching command, the industrial computer stops data recording under the current load condition and stores the real-time values ​​collected by the current sensor into the steady-state database of the corresponding load node; the pneumatic device is controlled to adjust the output pressure according to the next load condition to be tested in the test sequence; the servo motor is controlled to adjust the speed according to the next load condition to be tested in the test sequence; data acquisition is blocked within a preset buffer time after the switching is completed, and online monitoring under the next load condition to be tested is restarted after the current friction system stabilizes.

[0086] In the above embodiments, after the host computer completes the recording of steady-state parameters, it automatically generates a load switching command. This command includes key parameters such as the target contact pressure and target sliding speed values ​​for the next load condition to be tested in the test sequence. After being formatted into standardized commands recognizable by the industrial computer using a LabVIEW program, it is transmitted to the industrial computer in real time through a preset ADS communication interface. Simultaneously, the host computer sends a control signal to stop the current data recording, ensuring that test data under the current load condition is not added, avoiding confusion between data from different operating conditions, and providing clear instruction guidance for load condition switching. After receiving the load switching command and stop recording signal from the host computer, the industrial computer first stops the real-time data acquisition and recording work under the current load condition. Then, it integrates the last set of real-time values ​​collected by the current sensor with the previously recorded steady-state parameters and stores them together in the steady-state database of the corresponding load node, ensuring that the test data for the current operating condition is completely archived for subsequent traceability and analysis. After completing the data archiving, the industrial computer parses the parameters of the next load condition to be tested according to the command, preparing for the adjustment of the drive actuator and establishing a key connection link for load condition switching. The industrial computer generates a corresponding pressure control command based on the target contact pressure value for the next load condition. This command is transmitted to the I / O terminal module via a coupler, and then sent to the pneumatic device via the I / O terminal. The pneumatic device responds to the command by adjusting its operating state. If the target pressure value for the next condition is higher than the current value, it increases the internal pressure output. If it is lower than the current value, it decreases the pressure output until the deviation between the actual pressure value fed back by the force sensor and the target value is within a preset allowable range, thus completing the precise switching of the contact pressure of the friction pair and providing the required load conditions for the next test condition. For the target sliding speed value of the next load condition, the industrial computer generates a corresponding speed control command and sends it to the servo controller, which acts as an EtherCAT slave. The servo controller drives the single-axis servo motor to operate at the speed set by the command. The motor output speed is adapted by the gearbox to convert the motor speed into the specific relative sliding speed required for the next condition. Through the mechanical transmission components, the metal shaft is driven to rotate, causing the metal shaft to slide relative to the test sample surface in accordance with the requirements of the new load combination, thus achieving precise switching of the sliding speed. After pressure and speed adjustments are completed, the system enters a preset buffer time phase. During this period, the industrial computer disables data acquisition, shielding invalid data generated by system state fluctuations during the switching process to prevent such data from interfering with the accuracy of subsequent steady-state identification. The buffer time is set based on the dynamic response characteristics of the friction system, ensuring that the contact state of the friction pair, system temperature, and other parameters stabilize at the initial state under the new operating conditions.After the buffer period ends, the industrial computer restarts its data acquisition function, the sensor group resumes real-time monitoring of friction factor, wear height and system temperature data, the host computer synchronously resumes data reception and feature analysis, and officially starts the friction test under the next load condition to be tested, until all load conditions in the test sequence have been tested.

[0087] Step S106: After the test, export the result file containing the original measurement data and steady-state tribological parameters under each test load condition. Optionally, the result file is a TDMS (Technical Data Management Streaming) format file, which contains the calculated values ​​of the tribological parameters in steady state under each test load condition. For example, the hierarchical data model can be encapsulated into a TDMS format file and exported to achieve high-speed storage and rapid backtracking of massive amounts of test data. In practical applications, it can also be exported as other file formats.

[0088] As an optional embodiment, in step S106, the original sensor signal sequence collected during the test process is integrated into a time-domain dataset; the average friction factor and wear rate after entering steady state under each load condition are extracted as steady-state tribological parameters; the time-domain dataset and the steady-state tribological parameters are encapsulated into a structured data file and exported.

[0089] For example, after the test, the host computer first initiates the raw data integration process. Utilizing the data analysis capabilities of the LabVIEW program, it summarizes all raw signal sequences collected by the sensor groups throughout the test, including raw friction factor data recorded by the force sensor, raw wear height data collected by the displacement sensor, and raw system temperature data monitored by the temperature sensor. Simultaneously, it associates the timestamps corresponding to each data set to ensure consistency across the time dimension. Following the execution order of each load condition in the test sequence, the raw data under different operating conditions are categorized and organized to form a structured time-domain dataset. This dataset completely preserves all raw monitoring information from the start of the test to the end of all operating conditions, with no data omissions or losses, providing a complete raw data foundation for subsequent data backtracking and secondary analysis.

[0090] Further, optionally, the structured data file includes, but is not limited to, a TDMS format file.

[0091] With the support of the executable program module embedded in the host computer, steady-state tribological parameters under various load conditions are automatically extracted. For each load condition, the script accurately locates the effective data segment after the system enters steady state. Based on the previously smoothed friction factor sequence, the mean friction factor of this data segment is calculated as the core friction characteristic parameter under steady state. Combining the wear height data of the steady-state stage with the corresponding time span, the wear slope obtained through linear regression analysis is further converted into the wear rate per unit time, intuitively reflecting the wear rate of the material under steady state. Each set of extracted steady-state parameters (mean friction factor, wear rate) is bound to the corresponding load condition (contact pressure, sliding speed) to ensure accurate correlation between parameters and working conditions, clearly presenting the steady-state tribological characteristics under different load combinations.

[0092] To achieve orderly data storage and efficient backtracking, the host computer constructs a hierarchical data model from the integrated time-domain dataset and extracted steady-state tribological parameters. This hierarchical data model uses the test project as the root directory, with sub-directories: the first level categorizes data according to the load condition numbers in the test sequence, with each number corresponding to an independent load condition folder; the second level further divides each load condition folder into raw data subfolders and steady-state parameter subfolders. The raw data subfolder stores the time-domain dataset for that load condition, while the steady-state parameter subfolder stores the corresponding core parameters such as the mean friction factor and wear rate. Simultaneously, a global information file is added to the root directory to record the basic test configuration (such as custom load range, sampling density, steady-state judgment threshold, etc.), forming a logically clear and hierarchically distinct data structure, facilitating rapid location and retrieval of target data later.

[0093] The host computer encapsulates the constructed hierarchical data model using the built-in file processing interface of the LabVIEW program. The file processing interface can be designed based on a TDMS format data management scheme, supporting high-speed storage and efficient retrieval of massive amounts of data, perfectly adapting to the multi-dimensional, large-capacity test data generated by this testing system. Alternatively, it can be designed based on other file formats or support adaptive switching between multiple file formats. During the encapsulation process, the program automatically maps the directory structure and data content of the hierarchical data model to the internal storage structure of the TDMS file, ensuring that the data's correlation and integrity are not compromised. After encapsulation, the host computer automatically exports the TDMS format result file according to the user-preset storage path (which can be set through a visual interface). This TDMS format result file contains both complete raw measurement data and calculated steady-state tribological parameters under various operating conditions, achieving centralized management of test data and meeting diverse needs such as subsequent data backtracking, scientific research analysis, and report preparation.

[0094] In this embodiment, an embedded state decision model embedded in the host computer randomly generates the sequence of loads to be tested, no longer limited to traditional fixed load testing. The test conditions are closer to the complex working conditions in actual production, making the acquired tribological data more practical and accurate. Furthermore, the embedded state decision model analyzes the data collected by the sensors in real time and automatically identifies steady states, eliminating the need for fixed test cycles and avoiding the need to extend test time to cover steady states. This significantly shortens the overall test cycle and reduces repetitive manual operations. Simultaneously, this embodiment can automatically switch the load conditions to be tested and complete the data acquisition and analysis process without manual intervention, improving the efficiency of material friction testing and reducing material consumption and time costs. Finally, a result file integrating the original data and steady-state parameters is exported, providing more comprehensive and accurate data analysis results for subsequent materials science research and analysis, further improving the research efficiency in the field of friction testing.

[0095] This application provides a friction test load adjustment and steady-state identification system based on an embedded model. The system is used for friction testing and includes a host computer, an industrial computer, and an actuator. The host computer receives a user-inputted custom test load range and selects multiple test load conditions within the custom test load range using an embedded state decision model built into the host computer, forming a test sequence. It then controls the actuator to apply corresponding contact pressure to the friction pair and adjust it to a specific sliding speed according to the test sequence. The actuator applies corresponding contact pressure to the friction pair and adjusts it to a specific sliding speed according to the control instructions from the host computer. The industrial computer collects friction factor and wear height data in real time through a sensor array during the test and transmits the collected raw measurement data to the host computer with the embedded state decision model in real time. The host computer is also used to perform feature analysis on the raw measurement data using an embedded state decision model to determine whether the fluctuation characteristics of the current friction system meet the preset conditions for the current friction system to reach a steady state. If the current friction system is identified as having reached a steady state, the steady-state tribological parameters under the current load condition are recorded, and the actuator is instructed to automatically switch to the next load condition in the test sequence until all load conditions are tested. After the test, a result file containing the raw measurement data and the steady-state tribological parameters under each load condition is exported. In some embodiments, the friction test load adjustment and steady-state identification system based on the embedded model can be applied to terminal devices. It should be noted that, for the sake of convenience and brevity, the specific working process of the friction test load adjustment and steady-state identification system based on the embedded model described above can be referred to the corresponding process in the aforementioned embodiments of the friction test load adjustment and steady-state identification method based on the embedded model, and will not be repeated here.

[0096] In practical applications, for example, Figure 3This is a schematic diagram of a friction test load adjustment and steady-state identification system based on an embedded model. Figure 3 The hardware composition and interconnections of the intelligent friction testing system based on an embedded model are described below. The core components include a host computer, an industrial computer, actuators, and a sensor array. The actuators consist of a servo motor, gearbox, pneumatic device, metal shaft (one end of the friction pair), and test sample (the other end of the friction pair). The sensor array includes mechanical sensors, displacement sensors, and temperature sensors. The control and communication components include a servo controller, industrial computer, host computer, and I / O terminals. All components are connected via specific communication links. Figure 3 The lines in the diagram represent the connection relationships between the components, and the arrows indicate the direction of data flow processing between the components.

[0097] This application provides a terminal device. The terminal device includes a processor and a memory, which are connected via a bus, such as an I / O bus. 2 C-bus. Specifically, the processor provides computing and control capabilities to support the operation of the entire terminal device. The processor can be a central processing unit, or it can be other general-purpose processors, digital signal processors, application-specific integrated circuits, field-programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Those skilled in the art will understand that the structures shown in the above embodiments are merely block diagrams of some structures related to the embodiments of this application, and do not constitute a limitation on the terminal device to which the embodiments of this application are applied. Specific servers may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. The processor is used to run computer programs stored in memory, and when executing the computer programs, implements any of the embedded model-based friction test load adjustment and steady-state identification methods provided in the embodiments of this application. It should be noted that those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the terminal device described above can be referred to the aforementioned embodiments of the embedded model-based friction test load adjustment and steady-state identification method, and will not be repeated here.

[0098] This application also provides a storage medium for computer-readable storage, wherein the storage medium stores one or more programs that can be executed by one or more processors to implement the steps of any of the friction test load adjustment and steady-state identification methods based on embedded models provided in the specification of this application.

Claims

1. A method for adjusting friction test load and identifying steady-state conditions based on an embedded model, characterized in that, The method is applied to a friction testing system, which includes a host computer, an industrial computer, and an actuator. The method includes: The host computer receives the user's input of a custom test load range and uses the embedded state decision model built into the host computer to select multiple test load conditions within the custom test load range to form a test sequence. The host computer controls the actuator to apply the corresponding contact pressure to the friction pair and adjust it to a specific sliding speed according to the test sequence. During the test, the friction factor and wear height data are collected in real time by the sensor group of the industrial computer, and the collected raw measurement data is transmitted in real time to the host computer with an embedded state decision model. The host computer uses an embedded state decision model to perform feature analysis on the raw measurement data to determine whether the fluctuation characteristics of the current friction system meet the preset conditions for the current friction system to reach a steady state. If the current friction system is detected to have reached a steady state, the steady-state tribological parameters under the current load condition are recorded, and the actuator is instructed to automatically switch to the next load condition in the test sequence until all load conditions are tested. After the test, export the result file containing the original measurement data and steady-state tribological parameters under each load condition.

2. The friction test load adjustment and steady-state identification method based on an embedded model according to claim 1, characterized in that, The process involves using an embedded state decision model built into the host computer to select multiple load conditions within a custom test load range, forming a test sequence, including: Obtain the custom test load range, which includes the contact pressure range and the sliding speed range; The executable program module built into the host computer program is invoked to randomly sample within the contact pressure range and sliding speed range to generate multiple combined coordinate points to represent different combinations of pressure and speed. The test sequence is generated by arranging multiple combined coordinate points according to preset rules or random order.

3. The friction test load adjustment and steady-state identification method based on the embedded model according to claim 2, characterized in that, The load conditions to be measured include at least contact pressure and sliding velocity; the random sampling within the contact pressure range and sliding velocity range generates multiple combined coordinate points to represent different combinations of pressure and velocity, including: The sampling density of the coordinate points to be tested is determined according to the custom test load range, and a two-dimensional test load parameter space is constructed with contact pressure as the horizontal axis and sliding speed as the vertical axis. Based on the sampling density, random sampling is performed on the two-dimensional test load parameter space to obtain preliminary candidate coordinate points; Calculate the distance metric operator between each candidate coordinate point; Combining the distance metric operator, the distribution density evaluation index of each candidate coordinate point in the two-dimensional test load parameter space is calculated. Based on the distribution density evaluation index, the candidate coordinate points with a clustering degree higher than a set clustering degree threshold in the local area are redistributed and adjusted until the point distribution of the candidate coordinate points meets the preset uniformity index. Then, multiple combined coordinate points are output so that the final generated test sequence presents a statistically significant uniform distribution in the two-dimensional test load parameter space.

4. The friction test load adjustment and steady-state identification method based on an embedded model according to claim 1, characterized in that, The fluctuation characteristics include the fluctuation variance of the friction factor data and the slope value of the wear height data over time; the step of using an embedded state decision model on a host computer to perform feature analysis on the original measurement data to determine whether the fluctuation characteristics of the current friction system meet the preset conditions for the current friction system to reach a steady state includes: The embedded state decision model built into the host computer is invoked, and the size parameters of the sliding window are adaptively configured according to the dynamic response characteristics of the current friction system. The embedded state decision model is obtained by pre-fitting historical test data of similar materials with the dynamic characteristic law of the friction system. The friction factor data and wear height data within the sliding window are extracted from the raw measurement data transmitted in real time. Calculate the variance of the friction factor data within the sliding window, and calculate the slope of the wear height data over time; Determine whether the fluctuation variance is less than a preset stability threshold and whether the slope value enters a preset linear wear reference range; wherein, the stability threshold is preset based on the inherent friction characteristics of the friction pair material, historical test data of similar materials and the allowable range of friction fluctuation in actual application scenarios, and the linear wear reference range is preset based on the rated wear characteristics of the friction pair material, the change law of wear height over time in multiple sets of predictive tests and the normal wear range under actual working conditions. If the fluctuation variance is less than the stability threshold, and the slope value falls within the preset linear wear reference range, and satisfies the preset convergence criterion within a continuous preset sliding window period, then the current friction system is determined to meet the preset conditions for reaching a steady state.

5. The friction test load adjustment and steady-state identification method based on an embedded model according to claim 4, characterized in that, The fluctuation characteristics also include auxiliary state parameters, which include any one or a combination of system temperature, normal load closed-loop error, and sliding speed fluctuation amplitude. The step of using an embedded state decision model on a host computer to perform feature analysis on the original measurement data and determine whether the fluctuation characteristics of the current friction system meet the preset conditions for the current friction system to reach a steady state also includes: The embedded state decision model is used to determine whether the current friction system has reached a steady state based on the convergence state of the friction factor data, the wear height data, and the auxiliary state parameters.

6. The friction test load adjustment and steady-state identification method based on the embedded model according to claim 4, characterized in that, The calculation of the fluctuation variance of the friction factor data within the sliding window, and the calculation of the slope value of the wear height data changing over time, include: The original friction factor data within the sliding window is denoised to obtain a smoothed friction factor sequence. Within the sliding window, the average value of all friction factors in the smoothed friction factor sequence is calculated. The deviation value between each friction factor and the average value is calculated one by one. The average value is then taken after squaring all the deviation values ​​to obtain the fluctuation variance within the sliding window that reflects the overall deviation of the friction factors. Linear regression analysis is performed on the wear height data within the sliding window and the corresponding timestamps to obtain a local wear linear equation. The local wear linear equation is used to represent the local correlation law generated by the change of wear height data over time, and the slope of the local wear linear equation is extracted as the slope value of the change of wear height data over time.

7. The friction test load adjustment and steady-state identification method based on an embedded model according to claim 1, characterized in that, The instruction actuator automatically switches to the next load condition to be tested in the test sequence, including: Send load switching commands from the host computer to the industrial computer; After receiving the load switching command, the industrial computer stops recording data under the current load condition and stores the real-time values ​​collected by the current sensors into the steady-state database of the corresponding load node. The control pneumatic device adjusts the output pressure according to the next load condition to be tested in the test sequence; The servo motor is controlled to adjust its speed according to the next load condition in the test sequence. Data acquisition is blocked within a preset buffer period after the switch is completed. Once the current friction system stabilizes, online monitoring will be restarted under the next load condition to be tested.

8. The method for adjusting friction test load and identifying steady-state conditions based on an embedded model according to claim 1, characterized in that, The exported result file, containing the original measurement data and steady-state tribological parameters under each load condition, includes: The raw sensor signal sequences collected during the testing process are integrated into a time-domain dataset; The mean friction factor and wear rate after entering steady state under each load condition are extracted as steady-state tribological parameters. The time-domain dataset and the steady-state tribological parameters are encapsulated into a structured data file and exported.

9. The method for adjusting friction test load and identifying steady-state conditions based on an embedded model according to claim 1, characterized in that, The host computer and the industrial computer exchange data via the ADS communication interface; The industrial computer acts as the master station and connects to the slave devices in the actuator via the EtherCAT bus. During the process of the actuator applying pressure to the friction pair, the pressure value is fed back in real time by the force sensor and compared with the set value in a closed loop. The output force of the positive pneumatic device in the actuator is then corrected in real time by the industrial computer.

10. A friction test load adjustment and steady-state identification system based on an embedded model, characterized in that, The system is used to perform friction testing. The system includes a host computer, an industrial computer, and an actuator. The host computer receives a user-inputted custom test load range and uses an embedded state decision model built into the host computer to select multiple test load conditions within the custom test load range to form a test sequence. According to the test sequence, the actuator is controlled to apply a corresponding contact pressure to the friction pair and adjust it to a specific sliding speed. The actuator is used to apply the corresponding contact pressure to the friction pair and adjust it to a specific sliding speed according to the control instructions of the host computer. An industrial computer is used to collect friction factor and wear height data in real time through a sensor array during the testing process, and transmit the collected raw measurement data to a host computer with an embedded state decision model in real time. The host computer is also used to perform feature analysis on the original measurement data using the embedded state decision model to determine whether the fluctuation characteristics of the current friction system meet the preset conditions for the current friction system to reach a steady state. If the current friction system is found to have reached a steady state, the steady-state tribological parameters under the current load condition are recorded, and the actuator is instructed to automatically switch to the next load condition in the test sequence until all load conditions are tested. After the test, the result file containing the original measurement data and the steady-state tribological parameters under each load condition is exported.

Images