VR-assisted debugging and fault locating method for equipment maintenance

Abstract

The present application relates to VR auxiliary debugging and bearing fault positioning technical field, specifically, the present application relates to VR auxiliary debugging and fault positioning method for equipment maintenance, aiming at solving the problem that traditional bearing fault detection adopts visual and static detection, and cannot collect full angle torque dynamic characteristics and large fault positioning error, in the bearing press fitting link, the dynamic torque sensor integrated with the angle encoder is used to collect the instantaneous torque of the full angle, the torque curve of the rotation angle is generated and is associated with the coding storage, when the fault occurs, the measured curve and the standard curve of the same type are called, and are bound to the double virtual bearing model of the VR scene, the virtual driving model is synchronously rotated, the signal module generates the resistance signal according to the torque to control the force feedback data glove, the real rotation feeling is restored by superimposing the viscous damping, the comparison module compares the resistance difference point by point and in the sliding window, if the threshold value is exceeded, the abnormal rotation angle interval is locked, the effective fault area is verified and confirmed through the original point, and the positioning result is output, and the fault positioning precision is improved.

Description

Technical Field

[0001] This invention relates to the field of VR-assisted debugging and bearing fault location technology, and more specifically, to a VR-assisted debugging and fault location method for equipment maintenance. Background Technology

[0002] VR-assisted debugging and bearing fault location is an important technology, specifically applied to the VR-assisted debugging and precise location of equipment bearing faults. By integrating dynamic acquisition of rotational torque and force feedback, it improves fault location accuracy and meets the needs of efficient equipment maintenance. The fault characteristics of equipment bearings are concentrated in the continuous change of rotation angle and torque during full-circumference rotation. Since bearing fault detection uses visual inspection and static parameter detection methods, it is impossible to acquire and analyze the dynamic correlation characteristics of rotational torque during full-circumference rotation. This results in the inability to accurately quantify and identify local wear faults in bearings, leading to large fault location errors. To solve this technical problem, we provide a VR-assisted debugging and fault location method for equipment maintenance. Summary of the Invention

[0003] The purpose of this invention is to provide a VR-assisted debugging and fault location method for equipment maintenance, so as to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, one of the objectives of this invention is to provide a VR-assisted debugging and fault location method for equipment maintenance, comprising the following steps:

[0005] S1. In the bearing press-fitting process of the equipment to be maintained, a dynamic torque sensor with an integrated angle encoder is coupled to the inner hole of the bearing inner ring. The bearing inner ring is driven to rotate one revolution at a constant angular velocity, and the position of each rotation angle is recorded. At the same time, the dynamic torque sensor synchronously collects the instantaneous torque value at each rotation angle position, generates the rotation angle torque curve, and associates the rotation angle torque curve with the bearing's unique identification code and stores it in the maintenance database.

[0006] S2. When the equipment in which the bearing is used malfunctions, the data retrieval module retrieves the corresponding rotational torque curve of the bearing from the maintenance database as the curve to be tested, and at the same time retrieves the standard rotational torque curve of the same type of qualified bearing as the reference curve. The curve loading module binds the curve to be tested and the reference curve to the corresponding first virtual bearing model and second virtual bearing model in the VR scene, respectively.

[0007] S3. The virtual drive module generates a first virtual rotation operation command that rotates at a constant angular velocity and applies the first virtual rotation operation command to the first virtual bearing model. It reads the instantaneous torque value corresponding to the current rotation angle on the curve to be measured in real time. The signal generation module generates a first dynamic resistance control signal according to the magnitude of the instantaneous torque value and a linear mapping relationship. This signal is used to drive the dynamic resistance output of the force feedback data glove in a manner proportional to the magnitude of the dynamic resistance.

[0008] S4. The virtual drive module generates another virtual rotation operation command that rotates at the same constant angular velocity. The second virtual rotation operation command is applied to the second virtual bearing model. The reference instantaneous torque value corresponding to the current rotation angle on the reference curve is read in real time. The signal generation module generates a second dynamic resistance control signal based on the reference instantaneous torque value.

[0009] S5. The comparison module compares the resistance values ​​of the first dynamic resistance control signal and the second dynamic resistance control signal at the same turning position point by point, calculates the resistance difference, and when the resistance difference exceeds the preset threshold, the result output module outputs the positioning result.

[0010] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0011] This invention replaces traditional visual inspection and static parameter detection methods by collecting dynamic correlation features of rotation angle and torque during the full circumference rotation of equipment bearings. It can accurately capture the quantitative features of local wear faults in bearings, effectively making up for the shortcomings of traditional detection methods in identifying subtle faults. Combined with the debugging method of integrating VR scene visualization and force feedback tactile perception, abstract dynamic data is transformed into intuitive visual and tactile perception, improving the accuracy and efficiency of fault location, avoiding errors caused by subjective human judgment, and achieving precise location of fault areas through dynamic feature analysis and quantitative analysis, reducing the risk of equipment operation due to missed fault detection. Attached Figure Description

[0012] Figure 1 This is a flowchart illustrating the overall workflow of the present invention. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] Please see Figure 1 As shown, this embodiment provides a VR-assisted debugging and fault location method for equipment maintenance, including the following steps:

[0015] S1. In the bearing press-fitting process of the equipment to be maintained, the dynamic torque sensor with integrated angle encoder is coupled with the inner hole of the bearing inner ring. The bearing inner ring is driven to rotate one revolution at a constant angular velocity, and the position of each rotation angle is recorded. At the same time, the dynamic torque sensor synchronously collects the instantaneous torque value at each rotation angle position, generates the rotation angle torque curve, and associates the rotation angle torque curve with the bearing's unique identification code and stores it in the maintenance database.

[0016] S2. When the equipment in which the bearing is used malfunctions, the data retrieval module retrieves the corresponding rotational torque curve of the bearing from the maintenance database as the curve to be tested, and at the same time retrieves the standard rotational torque curve of the same type of qualified bearing as the reference curve. The curve loading module binds the curve to be tested and the reference curve to the corresponding first virtual bearing model and second virtual bearing model in the VR scene, respectively.

[0017] S3. The virtual drive module generates a first virtual rotation operation command that rotates at a constant angular velocity and applies the first virtual rotation operation command to the first virtual bearing model. It reads the instantaneous torque value corresponding to the current rotation angle on the curve to be measured in real time. The signal generation module generates a first dynamic resistance control signal according to the magnitude of the instantaneous torque value and a linear mapping relationship. This signal is used to drive the dynamic resistance output of the force feedback data glove in a manner proportional to the magnitude of the dynamic resistance.

[0018] S4. The virtual drive module generates another virtual rotation operation command that rotates at the same constant angular velocity. The second virtual rotation operation command is applied to the second virtual bearing model. The reference instantaneous torque value corresponding to the current rotation angle on the reference curve is read in real time. The signal generation module generates a second dynamic resistance control signal based on the reference instantaneous torque value.

[0019] S5. The comparison module compares the resistance values ​​of the first dynamic resistance control signal and the second dynamic resistance control signal at the same turning position point by point, calculates the resistance difference, and when the resistance difference exceeds the preset threshold, the result output module outputs the positioning result.

[0020] In the bearing press-fit assembly process of the equipment to be maintained, after the coaxial positioning and press-fitting of the bearing and the equipment base are completed, the output shaft of the industrial-grade dynamic torque sensor integrating a photoelectric angle encoder is rigidly coaxially coupled to the central inner hole of the bearing inner ring. The coupling interface adopts a double mechanical fixing structure with conical locking and flat key anti-slip, which physically ensures that there is no relative rotation between the sensor drive shaft and the bearing inner ring, ensuring the synchronization of rotational motion and the accuracy of torque acquisition. Subsequently, the servo drive unit controlled by the closed loop drives the output shaft of the dynamic torque sensor to rotate at a preset constant angular velocity, thereby driving the bearing inner ring to complete a full 360° circumference rotation around its central axis. During this constant angular velocity rotation, the photoelectric angle encoder integrated inside the dynamic torque sensor starts the angle detection and hardware pulse triggering mechanism. This angle encoder has a built-in 2048-line grating code disk, and the angle detection resolution can reach [missing information]. The system presets the equal angle interval to a fixed parameter (default setting is 0). It can be adjusted according to the curve resolution requirements. to (Adjustment within the interval) Whenever the inner ring of the bearing rotates to a position at a preset equal angular interval, the grating photoelectric detection unit inside the angle encoder generates a hardware trigger pulse. The hardware trigger pulse is an edge-triggered digital signal, and each hardware trigger pulse corresponds one-to-one with the absolute angular position of the inner ring of the bearing. The hardware trigger pulse is transmitted in real time to the system's synchronous trigger module. The synchronous trigger module, as the synchronous control unit for angle and torque acquisition, adopts a hardware-level parallel output architecture, synchronously distributing each received hardware trigger pulse to the 32-bit binary angle counter built into the angle encoder. The angle counter latches and outputs the current absolute angular position value of the inner ring of the bearing the instant it receives the trigger pulse. Another synchronous transmission is sent to the torque sampling circuit inside the dynamic torque sensor. The analog bandwidth of the torque sampling circuit is not less than 1kHz, and the torque sampling resolution reaches 0.001N·m. The instant it receives the trigger pulse, it starts a single instantaneous torque sampling, acquiring the instantaneous torque value generated by the bearing rotation at the current angular position, ensuring that each angular position value is bound one-to-one with the instantaneous torque value at the same moment, forming a set containing the absolute angular position. and instantaneous torque value The original data points, all original data points are arranged according to the corner position from to The data points are arranged in ascending order to form a sequence of original discrete data covering the full circumference of the bearing. The data processing module reads all the original discrete data points from the synchronization trigger module and sensor unit through the communication interface, and initiates the continuous angular torque curve reconstruction process based on the linear interpolation algorithm. The linear interpolation algorithm is an algorithm that converts discrete original data points into a high-resolution continuous curve, which can improve the continuity and angular resolution of the curve. The linear interpolation algorithm traverses the original data sequence and defines two adjacent original data points as the starting data points. With endpoint data points ,in ; To preset equal angular intervals, the default is... Then, calculate according to the linear interpolation formula. Within the angular interval between two adjacent original data points, interpolation is performed according to a preset density (system default is per...). Nine intermediate data points are inserted at the original interval, meaning the overall angular resolution after interpolation reaches [value missing]. Multiple intermediate data points are evenly inserted, each containing the interpolated angle position and the corresponding calculated instantaneous torque value. This interpolation algorithm is then used to... to After traversing all adjacent original data points within the full circumference angle, the original discrete data points and all interpolated intermediate data points are continuously sorted according to their corner positions, ultimately generating a covering line. to The angular torque curve, consisting of continuous angular positions and their corresponding instantaneous torque values, is reconstructed. The data processing module then performs multi-dimensional standardized preprocessing on the curve. After preprocessing, the system assigns a unique identifier code to the currently detected bearing. This code includes traceability information such as bearing model, production batch, pressing time, equipment number, and serial number. The data processing module strongly associates the preprocessed standardized angular torque curve with this unique identifier code. This association is achieved through database primary key constraints, ensuring a one-to-one correspondence. Finally, the associated data is stored in a dedicated maintenance database for equipment maintenance via the industrial-grade MODBUS-TCP database communication protocol. This database employs a distributed storage architecture.

[0021] In a dedicated maintenance database for equipment maintenance, when a standard angle torque curve for a specific bearing model needs to be generated, a same-model retrieval mechanism based on the bearing's unique identifier code is initiated. This unique identifier code employs a hierarchical structured coding rule. The data retrieval module extracts the bearing model field of the standard curve to be generated using a built-in coding parsing algorithm. Using the bearing model field as the search keyword, a structured query is executed to retrieve all pre-processed angle torque curves corresponding to the same bearing model from the full historical storage data of the maintenance database. Unqualified curve data is removed to ensure that all samples participating in the standard curve generation are qualified products of the same model. After the retrieval is completed, the system automatically counts the total number of qualified sample curves. The system presets a minimum effective sample size threshold of 10. If the number of qualified curves retrieved is lower than this threshold, the system automatically triggers a sample replenishment prompt until the minimum sample size requirement is met. Subsequently, the curve loading module uses a high-speed parallel database reading interface to batch load all retrieved qualified angle torque curves of the same model into the system's dedicated curve calculation cache, verifying that each curve fully covers the data. to Full circumference, and ensure all curves use the same interpolation resolution (default). To eliminate calculation biases caused by resolution differences, invalid data in the curves are removed to ensure that all curves involved in the calculation are high-quality and valid data. After loading and verification, the curve loading module performs rotational alignment based on the feature corner positions. The specific alignment process is as follows:

[0022] The curve loading module performs an automatic full-circumference feature point extraction algorithm on each qualified angular torque curve of the same model participating in the calculation to identify the characteristic angular position of each curve. It prioritizes selecting the reference angular position with the minimum instantaneous torque value in each curve (i.e., the reference angle point with the minimum bearing rotational resistance, no jamming, and no eccentricity). This feature point is unique in each qualified bearing curve and has a clear physical meaning, unaffected by assembly phase offset. After extraction, the system uses the characteristic angular position of the first sample curve as the reference zero point (aligned to by default). Perform a rotation phase translation operation on all other sample curves, and calculate the angular offset between the characteristic rotation position of each curve and the reference zero point. The offset is then added to all the corner position data of the curve. The rotation alignment is completed, ensuring that the characteristic corner positions of all sample curves completely coincide to a unified reference angle, eliminating phase deviations caused by assembly and installation. After rotation alignment, the curve loading module performs global superposition and point-by-point arithmetic mean calculation on all aligned corner torque curves. The system uses the aligned unified angle axis as a reference and calculates the result at a fixed angle resolution. )Will to The full circumference is divided into 3601 equally spaced standard angle points. For each standard angle point, the system iterates through all aligned sample curves, extracts the instantaneous torque values ​​of all corresponding angle points, and performs an arithmetic mean calculation on these instantaneous torque values. The calculation formula is as follows: ,in This represents the instantaneous torque value at that angle point. This represents the total number of qualified sample curves of the same model. For the first The sample curves at the angle The instantaneous torque value at the specified angle point is used to retain the common torque characteristics of qualified bearings of the same model through point-by-point averaging. After the average torque value of all standard angle points is calculated, the system continuously arranges all angle points and their corresponding standard instantaneous torque values ​​in ascending order of angle, reconstructing and generating a standard angular torque curve representing the common mechanical characteristics of qualified bearings of the same model. Subsequently, the system performs double validity checks on the generated standard curve. After the checks pass, the system strongly associates and binds the standard angular torque curve with the corresponding bearing model code, and marks traceability data such as curve generation time, number of samples involved in the calculation, and version information. Finally, through the database transaction commit mechanism, the standard curve is permanently stored in the standard curve dedicated storage partition of the maintenance database, physically isolated from the test curve partition of the faulty bearing, supporting the retrieval during fault location.

[0023] During the VR scene initialization and construction phase, the system uses a 3D modeling engine to construct a standardized virtual bearing model based on the 1:1 real physical dimensions of the physical bearing to be maintained. This standardized virtual bearing model is rendered using a triangular patch topology for meshing, restoring the 3D shape, structural features, and rotational reference axis of the physical bearing. To achieve the mapping between the rotational angle and spatial points, the system renders the inner circumference surface of the virtual bearing (i.e., the force-bearing area where the inner ring of the physical bearing rotates and transmits torque) at an angle resolution completely consistent with the rotational torque curve (default). Virtual vertices are evenly distributed at the same angle, from... to A total of 3601 virtual vertices are arranged across the entire circumference. Each virtual vertex has a unique and non-repeatable Cartesian coordinate system in three-dimensional space. The coordinate values ​​are automatically generated by the modeling engine using the rotation axis of the virtual bearing center as the reference and the circumferential angle as the unique variable. The circumferential angle value of the virtual vertex is bound to its spatial coordinates, meaning that each virtual vertex corresponds to only one fixed circumferential angle position. Simultaneously, the system assigns an independent and unique vertex ID to each virtual vertex, automatically generating a standardized index mapping table of "vertex ID - 3D spatial coordinates - corresponding circumferential angle," which is permanently stored in a dedicated high-speed data cache area for VR scenes, providing an index basis for subsequent curve data binding. Subsequently, the curve loading module retrieves the test angle torque curve (faulty bearing) and standard angle torque curve (qualified bearing of the same model) associated with the bearing's unique identifier code from the maintenance database through an industrial-grade database communication interface. The retrieved curve data undergoes full-dimensional analysis and verification, extracting the curves covered by... to Continuous rotation position sequence of full circumference The sequence of instantaneous torque values ​​at the corresponding angle points During the analysis process, it was confirmed that the curve angle resolution and the virtual vertex layout resolution were completely consistent (both were...). To avoid mapping deviations caused by resolution differences, the verification curve fully covers the entire circumference, with no missing angles or torque null values, and removes abnormal torque extreme values ​​to ensure that the curve data involved in the mapping is high-quality and valid. After parsing and verification, the curve loading module initiates the circumferential mapping association process between the corner position and the virtual vertex. Using the rotation axis of the virtual bearing center as a unified reference, and following the rigid matching rule of "corner value = virtual vertex circumferential angle value", it maps each corner position in the corner torque curve. A unique binding is performed on the virtual vertex corresponding to the circumferential angle of the inner ring surface of the virtual bearing, i.e., in the curve. The corner position is uniquely mapped to the circumference angle of the virtual vertex. Spatial points, The corner position is uniquely mapped to the circumference angle of the virtual vertex. Spatial points, The corner position is uniquely mapped to the circumference angle of the virtual vertex. Spatial points, and so on until... The full-circumference angle mapping is completed. During the mapping process, the system automatically calls the "vertex ID-spatial coordinates-circumference angle" index table for matching, ensuring that each corner position corresponds to only one virtual vertex and each virtual vertex is bound to only one corner position. This achieves unbiased binding between the corner dimension of the physical detection data and the spatial dimension of the virtual model. After the angle mapping is completed, the curve loading module performs the instantaneous torque value dynamic attribute mounting operation, using the instantaneous torque value (including three types of fields: original torque value, normalized torque value, and torque fluctuation identifier) ​​corresponding to each corner position in the corner torque curve as the custom dynamic attribute data of the mapped virtual vertex. The data is written into the virtual vertex's dedicated attribute data block. This attribute is attached in a dynamic, non-destructive binding mode, only extending the virtual vertex's mechanical data attributes without changing the virtual bearing model's 3D geometry and rendering characteristics. It supports real-time refresh and replacement during subsequent curve data updates, ensuring smooth VR scene rendering and flexible data linkage. When the virtual drive module generates a virtual rotation operation command with a constant angular velocity and applies it to the virtual bearing model, the VR scene engine drives the virtual bearing model to rotate around the central axis at a uniform speed completely consistent with the physical bearing in real time. During the rotation, the scene engine outputs the current absolute rotation angle of the virtual bearing model in real time. The rotational synchronization accuracy reaches At this point, the curve loading module initiates a real-time torque data closed-loop retrieval mechanism, based on the current absolute rotation angle. The system indexes the target virtual vertex corresponding to the circumferential angle of the inner ring surface of the virtual bearing, and directly reads the instantaneous torque value attribute data attached to that virtual vertex. During the continuous rotation of the virtual bearing model, each virtual vertex on the inner ring surface of the virtual bearing will sequentially enter the current rotation angle position as the model rotates. Each virtual vertex will obtain the corresponding instantaneous torque value according to its bound mapped rotation angle position, eliminating the need for global traversal calculations, improving data retrieval efficiency, and ensuring complete synchronization between torque data and virtual rotation actions in the VR scene. This provides real-time and stable torque data input for the signal generation module to generate dynamic resistance control signals. To more intuitively present the implementation details of this entire process, an example from a real industrial VR application scenario is provided: For a certain model of BRG-6205-Z industrial bearing, the inner ring circumferential surface of its virtual bearing model is... The resolution is set with 3601 virtual vertices, among which the virtual vertex with vertex ID V0001 has the following 3D spatial coordinates: Corresponding to the circumferential rotation angle The virtual vertex with vertex ID V0011 has the following 3D spatial coordinates: Corresponding to the circumferential rotation angle The virtual vertex with vertex ID V0101 has the following 3D spatial coordinates: Corresponding to the circumferential rotation angle The virtual vertex with vertex ID V0901 has the following 3D spatial coordinates: Corresponding to the circumferential rotation angle The curve loading module retrieves the standard angular torque curve for this type of bearing and analyzes it to obtain... Corresponding instantaneous torque value correspond , correspond , correspond These torque values ​​are then loaded into the attribute data blocks of the corresponding virtual vertices. When the virtual drive module... When the virtual bearing model is driven to rotate by a constant angular velocity, the virtual rotation angle is updated in real time. , , The curve loading module immediately indexes the virtual vertex corresponding to the circumferential angle and reads the instantaneous torque value it is mounted on, such as when rotating to... At that time, retrieve the torque attribute value of vertex V0101. Rotate to At that time, retrieve the torque attribute value of vertex V0901. Each virtual vertex of the virtual bearing model can realistically reflect the instantaneous torque characteristics of the corresponding rotation angle position of the physical bearing, realizing a deep virtual-real fusion between the physical rotation angle torque curve and the VR virtual bearing model.

[0024] The second virtual bearing model, already bound to the standard angular torque curve of a qualified bearing of the same model, is loaded into the center of the VR scene. The rotation reference axis of the virtual bearing model is ensured to be coaxially aligned with the user's force feedback operation reference axis. Simultaneously, a full set of hardware initialization operations is performed on the force feedback data glove. After calibration, the force feedback data glove enters standby mode. The virtual drive module operates based on the constant angular velocity parameters (default) used during physical bearing testing. The system generates a standardized second virtual rotation operation command. This command includes four main parameters: rotation direction, constant angular velocity, rotation period, and full circumference coverage. These parameters are completely consistent with the rotation parameters used in physical bearing press-fit testing, ensuring a high degree of homogeneity between the virtual rotation conditions and the physical testing conditions. The virtual drive module continuously applies the second virtual rotation operation command to the second virtual bearing model, driving it to rotate around the central axis. to The virtual bearing rotates at a constant speed throughout its full circumference. During this rotation, the VR scene engine outputs the current absolute rotation angle position of the virtual bearing in real time. The curve loading module reads the corresponding standard instantaneous torque value from the standard rotation angle torque curve in real time based on this rotation angle position. When the virtual bearing rotates to the system's preset calibration characteristic angle position (selected where the torque value is evenly distributed and representative), , , , When there are four calibration points, the system pauses virtual rotation and locks the current standard instantaneous torque value, initiating the torque-resistance mapping calibration process. The signal generation module has a built-in preset initial linear mapping relationship, which adopts a linear proportional mathematical model commonly used in industrial force feedback, and the formula is: ,in To output the resistance control signal amplitude to the force feedback data glove, DQ is the scaling factor to be calibrated. This is the current standard instantaneous torque value. To fix the offset (used to eliminate actuator no-load friction, the default value is 0), the system presets an initial proportional coefficient. (The default value is 50, which can be adjusted adaptively according to the glove model.) The signal generation module calculates the initial resistance control signal amplitude based on the initial linear mapping relationship. The initial control signal is transmitted without delay to the resistance simulation module of the force feedback data glove via a high-speed Bluetooth 5.0 communication link. The resistance simulation module drives the glove's micro servo actuator to output a corresponding static resistance force. At this time, the professional equipment maintenance personnel wearing the force feedback data glove perceive the current resistance through tactile sensation. The system adopts a quantitative resistance perception scoring mechanism (dividing the resistance intensity into 0-10 levels, where level 0 is no resistance and level 10 is the maximum resistance, and the preset standard resistance intensity is the 5th level standard tactile sensation corresponding to the torque value). The user provides real-time feedback on the perceived resistance level through the interactive interface of the VR scene, and the system immediately collects and records the user's feedback resistance value. At the same time, retrieve the preset standard resistance value corresponding to the current calibration point. Calculate the deviation between the two. The system has a preset convergence threshold. (The default value is 0.5; a deviation less than this value is considered a match.) If the initial deviation value is greater than the convergence threshold, the system starts the bisection iterative optimization algorithm to adjust the scaling factor. Adaptive adjustments are made, and the adjustment rules are as follows:

[0025] when When the resistance is too low, the proportional coefficient is increased by 1.2 times. If the resistance is too high, the proportional coefficient is reduced by 0.8. After each adjustment, the resistance control signal, the resistance output of the drive glove, the user feedback resistance, and the deviation value are recalculated. The deviation is iterated and reduced until the deviation between the user feedback resistance intensity and the preset standard resistance intensity is less than the convergence threshold. At this point, the system locks the current proportional coefficient. The system completes the linear mapping calibration of a single calibration point. To ensure the accuracy of the full-circumference mapping, the system sequentially... , , , Repeat the iterative calibration process described above at the four calibration points, and take the arithmetic mean of the scaling factors obtained from the calibration at the four points to obtain the final global scaling factor. This allows us to determine the final linear mapping relationship covering the entire circumference. This mapping relationship accurately matches the user's tactile perception habits, restoring the realistic resistance feel when a qualified bearing rotates. After calibration, the system writes the proportional coefficient, bias, calibration parameters, user information, and calibration time of the final linear mapping relationship into the system's non-volatile configuration file, permanently storing them and automatically associating them with the current bearing model and force feedback glove model. Subsequent testing of the same bearing model directly calls upon these calibration parameters, eliminating the need for repeated calibration. To more intuitively illustrate the details of this entire process, an example from a real industrial VR calibration scenario is provided:

[0026] Select standard instantaneous torque value of Calibration points, preset standard resistance intensity Level, initial proportional coefficient of the system The initial control signal is calculated as follows User feedback on resistance Level, deviation If the level is greater than the threshold by 0.5, the system adjusts the proportional coefficient using a binary search method. Increased output resistance, user feedback Level, deviation The standard was still not met, and it was adjusted again to User feedback Level, deviation The calibration at this point meets the convergence requirement, and the scaling factor is 75. After calibrating four points sequentially, the average final scaling factor is obtained. The final linear mapping relationship is determined as follows: Under this mapping relationship, the resistance corresponding to the standard torque value is completely matched with the preset standard body feeling. When the user wears gloves and rotates the virtual bearing, he can truly feel the rotational resistance of the qualified bearing, providing a benchmark for subsequent comparison of abnormal resistance of faulty bearings.

[0027] The resistance simulation module establishes a two-way real-time communication link with the signal generation module, virtual drive module, and force feedback data glove, with the communication refresh rate locked at 1000Hz. Simultaneously, it performs fully automatic zero-point calibration, mechanical backlash compensation, no-load friction elimination, and torque output linearity calibration on the distributed micro torque actuator built into the force feedback data glove (using a fingertip integrated DC brushless torque motor, with a single finger rated torque output range of 0-5N, response time ≤3ms, and torque control resolution of 0.001N). The built-in micro torque sensor collects the actuator's no-load torque and automatically zeros it, eliminating output errors and calibration. After the completion of the process, the actuator enters a real-time standby state, laying the hardware foundation for subsequent resistance output. Once the system enters the fault location operation state, the resistance simulation module receives the dynamic resistance control signal output by the signal generation module in real time at a high sampling frequency of 1000Hz. This signal is derived from the instantaneous torque value through a calibrated linear mapping relationship, directly corresponding to the mechanical friction resistance of the faulty bearing / standard bearing at the current rotation angle. The resistance simulation module incorporates a 32-bit high-speed DSP digital signal processor to eliminate invalid and distorted signals, converting the parsed digital instructions into force feedback data for the target dynamic resistance force of the glove actuator. This is the rigid component of the bearing rotational resistance. After analysis, the resistance simulation module stores the target dynamic resistance force in real time into the on-chip cache, awaiting the synthesis calculation of the viscous damping component. At the same time, the resistance simulation module obtains the real-time rotation angle sequence of the virtual bearing output by the virtual drive module through the synchronous trigger interface. The angle sequence is completely synchronized with the rotation angles of the first virtual bearing model (under test) and the second virtual bearing model (reference) in the VR scene, achieving an angular resolution of [missing information]. ,cover Uninterrupted output across the entire circumference; the drag simulation module calculates the rate of change of rotation angle (i.e., rotational angular velocity) based on real-time angle sequences using a central numerical differential algorithm. The calculation formula is: ,in To avoid distortion caused by high-frequency noise introduced during differential calculations, which would result in angular velocity fluctuations, the module employs a 15-point moving average filtering algorithm to smooth the calculated angular velocity values, obtaining the real-time rotational angular velocity. This angular velocity is the sole variable simulating the viscous resistance of the lubricating medium. The resistance simulation module incorporates a database of viscous damping coefficients for lubricating media of various bearing models. This database is calibrated based on massive amounts of physical bearing measurement data and covers the viscous damping coefficients of commonly used lubricating media in industrial bearings. (unit: The coefficient value is strongly correlated with the bearing model, lubricating medium type, and operating temperature. The system automatically retrieves the matching calibrated viscous damping coefficient based on the bearing model and lubrication parameters currently being tested. Following the physical rule that the viscous damping component is proportional to the rate of change of the rotation angle, the formula... Calculate the real-time viscous damping component This component is specifically designed to simulate the fluid viscous resistance of the lubricating medium during bearing rotation, perfectly replicating the smooth, viscous feel of a physical bearing rotating from a tactile perspective. After real-time calculation of the target dynamic resistance force and viscous damping component, the resistance simulation module performs a total resistance vector summation calculation. Since both the dynamic resistance force and the viscous damping component are resistance vectors acting in the same direction that oppose bearing rotation (acting in the opposite direction to the virtual bearing's rotation), the total resistance ultimately perceived by the user is... The algebraic sum of the two is given by the formula: The synthesized total resistance torque is the final tactile output command directly applied to the user's hand. The resistance simulation module converts the synthesized total resistance into a PWM pulse width modulation signal or analog drive current that the glove actuator can recognize. An incremental PID closed-loop torque adjustment algorithm is used to control the actuator in real time. The actuator's built-in miniature torque sensor provides real-time feedback of the actual output torque value. The resistance simulation module compares the feedback value with the target total resistance torque in milliseconds, dynamically correcting the drive signal parameters, eliminating actuator output errors, and ensuring that the actual output resistance of the actuator matches the calculated total resistance. The torque deviation is less than or equal to 0.005N, achieving torque control. When equipment maintenance personnel wearing force feedback data gloves perform rotation operations on a virtual bearing model in a VR scene, their fingertips directly perceive the total resistance composed of dynamic resistance and viscous damping components, completely replicating the rotational feel of a physical bearing under real working conditions. This allows users to intuitively distinguish the resistance difference between standard and faulty bearings through touch, providing tactile evidence for fault angle location. To more intuitively present the implementation details of this entire process, it is combined with industrial equipment deep groove ball bearings (model 6205, lithium-based grease, calibrated viscous damping coefficient). Example of VR fault location scenario:

[0028] When the virtual drive module is at a standard constant angular velocity (Right now When the second virtual bearing model (standard) is driven to rotate, the dynamic resistance control signal output by the signal generation module is analyzed to obtain the target dynamic resistance force. After the resistance simulation module calculates the real-time angular velocity, it calculates the viscous damping component. Total combined resistance The force feedback data glove actuator outputs a resistance force of 0.914N after PID closed-loop control. When the user wears the glove and rotates, they feel the rotational resistance of a real bearing. When the user actively increases the rotational speed of the virtual bearing to... ( At this point, the rate of change of rotation angle doubles, and the viscous damping component increases synchronously to [a certain value]. The dynamic resistance remains constant at 0.6N, while the total resistance increases to Users clearly felt that the rotational resistance increased with increasing rotational speed, perfectly simulating the viscous resistance effect of lithium-based grease as rotational speed increased; when the user slowly rotated the virtual bearing to ( When ), the viscous damping component decreases to Total resistance decreased When detecting a faulty bearing, the dynamic resistance control signal at a certain rotation angle is analyzed to obtain... At the standard angular velocity, the viscous damping component remains at 0.314N, but the total resistance suddenly increases to 1.514N. The user immediately senses the abnormally high resistance at this angle using a force feedback glove, quickly locating the fault area. Furthermore, if a solid lubricant bearing is replaced (viscous damping coefficient...),... The system automatically retrieves the corresponding coefficients, and the viscous damping component is halved synchronously. The tactile perception of the low viscosity characteristics of the solid lubricant further verifies the module's multi-media adaptive capability.

[0029] Before the system enters the fault comparison analysis phase, the comparison module first completes the initialization configuration and loading of the same source data. It then synchronously acquires the first dynamic resistance control signal sequence output by the signal generation module (corresponding to the faulty bearing under test, full circumference) through the real-time data interface. angular resolution (Total data points: 3601) and the second dynamic resistance control signal sequence (corresponding to the same model standard bearing, completely homologous to the first signal in terms of source, resolution, and angle axis), while simultaneously loading the system's preset configuration parameters, including sliding window parameters (the window width adopts an equal angle design, and the default setting is set to...). ,correspond At this resolution, 100 consecutive corner data points are used. This width has been calibrated through extensive bearing fault testing. It effectively filters out transient noise without masking the true fault trend. It can be adjusted according to bearing specifications. (Interval adaptive adjustment), sliding step size (set to) This includes 10 data points to achieve continuous and smooth window sliding with no data omissions or overlapping redundancies; a preset threshold for trend resistance difference (calibrated based on the resistance fluctuation range of qualified bearings of the same model, defaulted to 20% of the standard resistance amplitude, adjustable according to detection accuracy requirements); and a minimum continuous length for abnormal areas (default 3 sliding windows, corresponding to...). (To avoid misjudgment due to single-point window anomalies in the corner interval), after loading, the comparison module performs triple data validity checks on the two resistance control signals. Once the checks pass, the two signals are stored in the high-speed computing cache. Then, the comparison module initiates the sliding window averaging calculation process, independently performing equal-angle sliding window averaging calculations on the first and second dynamic resistance control signals respectively. Taking the first dynamic resistance control signal as an example, from... Starting from the initial corner, select a center point with a width of [missing information]. A sliding window is formed by continuous data points, containing resistance control signal values ​​corresponding to 100 consecutive turning angles. The mean of all resistance values ​​within the window is calculated using an arithmetic mean algorithm, and the calculation formula is as follows: ,in This represents the trend resistance value for the current window. The number of data points in the window (100). For the first in the window Each resistance control signal value is calculated, and the window is then pressed. Step length Slide the direction, repeat the mean calculation, and continue until all iterations are completed. Similarly, at all corner positions across the entire circumference, the same sliding window averaging calculation is performed on the second dynamic resistance control signal. After smoothing the two signals through the sliding window, the resistance trend curves of the bearing under test and the reference bearing resistance trend curve are generated respectively. The two trend curves maintain the same angle axis and resolution as the original signal, achieving distortion-free extraction of trend features. After the trend curves are generated, the comparison module initiates the same-source point-by-point trend resistance comparison process. Using a unified corner position as the benchmark, the trend resistance values ​​of the two resistance trend curves at exactly the same corner position are compared point by point, and the trend resistance difference is calculated. To eliminate interference from both positive and negative directions and highlight abnormal amplitudes, the difference is calculated using absolute values, as shown in the formula: ,in For the curve to be measured at the turning angle The trend resistance value at that point, The reference curve at the turning angle The trend resistance value at each point is compared and covered point by point. The comparison module performs a full-circle measurement, ensuring no corner is missed, and simultaneously records the trend resistance difference at each corner position and the corresponding corner in real time. Next, the comparison module enters the process of determining abnormal trend resistance change areas, using the trend resistance difference calculated point by point. Compare with a preset threshold and mark all. For abnormal corner points exceeding a preset threshold, to avoid misjudgment caused by a single window anomaly, the module adopts a continuous anomaly clustering judgment rule, only determining anomalies when the abnormal corner points are continuously distributed and the length of the covered corner interval is greater than or equal to the minimum continuous length of the abnormal region. Only when this condition is met is the continuous interval identified as an area of ​​abnormal trend resistance change, and the starting turning point of this area is locked. With the termination angle The system locates the continuous angular range of the fault. If the anomaly is only a single point or short interval, it is judged as transient interference and not marked. When the comparison module determines that there is a valid abnormal trend resistance change area, it immediately sends an area positioning command to the result output module. The command includes the starting angle, ending angle, abnormal interval length, maximum trend resistance difference, and anomaly level of the abnormal area. After receiving the command, the result output module outputs standardized positioning results. The positioning results clearly mark the specific angular range of the abnormal trend resistance change area and visualize it in the VR scene through color highlighting, angle marking, and area selection. At the same time, it is synchronized to the force feedback data glove to provide users with dual fault prompts through touch and vision.

[0030] After the comparison module calculates the average using a sliding window, compares trend resistance point by point, and clusters continuous abnormal intervals, it has initially identified a certain range of continuous turning angles as an area of ​​abnormal trend resistance change, and recorded the initial original turning angle position of this abnormal area. With the termination of the original corner position At this point, the comparison module immediately initiates the process of extracting the original corner position points in the abnormal area. The system first defines the original corner position points as the actual sampling points (intermediate data points generated by nonlinear interpolation) generated by the angle encoder through hardware trigger pulses during bearing press-fit testing. These points are the most reliable data basis for detailed fault verification. The comparison module then proceeds according to... and Within the specified range, from the full set of original data corresponding to the first dynamic resistance control signal (the faulty bearing under test) and the second dynamic resistance control signal (the standard bearing of the same model) stored in the maintenance database, all original angle position points contained within the abnormal range are extracted, and the total number of original angle position points within the range is simultaneously counted. During the extraction process, intermediate data points generated by linear interpolation are removed, retaining only the original sampling points triggered by the hardware. Simultaneously, the validity of the original data is verified, ensuring complete coverage of all physical sampling points in areas of abnormal trend-based resistance changes. This provides reliable original data support for subsequent fine-grained difference calculations. After the original point extraction is complete, the comparison module enters the point-by-point calculation process for fine-grained resistance differences. The system pre-loads preset thresholds for fine-grained resistance differences. Confirmation ratio The preset threshold for fine resistance difference is based on the maximum normal fluctuation range of the original resistance signal of qualified bearings of the same model, calculated statistically. The threshold value calibrated in principle, and the confirmed proportion, are the minimum effective percentage calibrated after actual measurement and verification using a massive number of bearing fault samples. For each original corner position point within the area of ​​abnormal trend resistance change, the comparison module reads the original value of the first dynamic resistance control signal corresponding to that point. Compared with the original value of the second dynamic resistance control signal According to the fine resistance difference The formula calculates the absolute difference point by point, reflecting the true degree of resistance anomaly of the bearing under test at the original rotation angle position. After the calculation is completed, the fine resistance difference of each original point is compared with the fine preset threshold. A point-by-point rigidity comparison is performed to mark the original corner positions where the fine resistance difference exceeds a preset threshold, and the effective number of such threshold-exceeding original points is counted in real time. The point-by-point calculation and marking process covers all original corner locations within the abnormal area. After completing the statistics of original points exceeding the threshold, the comparison module initiates the calculation of the proportion of points exceeding the threshold and the fault validity determination process, based on the proportion. The formula calculates the percentage of original corner positions within the ultra-fine threshold of the abnormal trend resistance change area, and then the comparison module compares the calculated actual percentage. Confirmation ratio Perform rigid comparison and execute dual validity determination rules:

[0031] If the actual percentage Greater than or equal to the confirmation ratio This indicates that most of the original corner positions within the abnormal trend resistance change area have real and continuous resistance anomalies, rather than local noise or smoothing artifacts, and the abnormal area is determined to be an effective trend fault area.

[0032] If the actual percentage Less than the confirmation ratio This indicates that only a few points in the abnormal area exceed the threshold, and the resistance is abnormally discontinuous. It is a false anomaly caused by non-fault fluctuations or trend smoothing. The abnormal trend resistance change area is directly determined to be invalid and removed without triggering any fault output command. This judgment rule fundamentally eliminates the problem of misjudgment in the trend screening stage and realizes the confirmation of fault.

[0033] When the comparison module finally determines that a certain abnormal trend resistance change area is a valid trend fault area, it immediately sends a valid fault location trigger command to the result output module. The command includes the original start and end angles of the valid fault area, the end angle, the number of original points exceeding the threshold, the total number of original points, the actual proportion, the maximum fine resistance difference, and the fault severity level. After receiving the trigger command, the result output module immediately generates the location result. The location result not only clearly marks the specific angle range of the valid trend fault area, but also simultaneously includes quantitative data for fault verification. At the same time, the valid fault area is visualized in the VR scene, and the parallel dynamic feedback data glove outputs enhanced tactile prompts when the user rotates to the area, realizing triple fault location feedback of vision, touch, and data. This solves the industry pain point that single trend comparison is prone to misjudgment and insufficient fault location accuracy. It is the ultimate guarantee for realizing bearing fault diagnosis in VR-assisted debugging and fault location methods for equipment maintenance.

[0034] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A VR assisted commissioning and fault localization method for equipment maintenance, characterized in that: Includes the following steps: S1. In the bearing press-fitting process of the equipment to be maintained, a dynamic torque sensor with an integrated angle encoder is coupled to the inner hole of the bearing inner ring. The bearing inner ring is driven to rotate one revolution at a constant angular velocity, and the position of each rotation angle is recorded. At the same time, the dynamic torque sensor synchronously collects the instantaneous torque value at each rotation angle position, generates the rotation angle torque curve, and associates the rotation angle torque curve with the bearing's unique identification code and stores it in the maintenance database. S2. When the equipment in which the bearing is used malfunctions, the data retrieval module retrieves the corresponding rotational torque curve of the bearing from the maintenance database as the curve to be tested, and at the same time retrieves the standard rotational torque curve of the same type of qualified bearing as the reference curve. The curve loading module binds the curve to be tested and the reference curve to the corresponding first virtual bearing model and second virtual bearing model in the VR scene, respectively. S3. The virtual drive module generates a first virtual rotation operation command that rotates at a constant angular velocity and applies the first virtual rotation operation command to the first virtual bearing model. It reads the instantaneous torque value corresponding to the current rotation angle on the curve to be measured in real time. The signal generation module generates a first dynamic resistance control signal according to the magnitude of the instantaneous torque value and a linear mapping relationship. This signal is used to drive the dynamic resistance output of the force feedback data glove in a manner proportional to the magnitude of the dynamic resistance. S4. The virtual drive module generates another virtual rotation operation command that rotates at the same constant angular velocity. The second virtual rotation operation command is applied to the second virtual bearing model. The reference instantaneous torque value corresponding to the current rotation angle on the reference curve is read in real time. The signal generation module generates a second dynamic resistance control signal based on the reference instantaneous torque value. S5. The comparison module compares the resistance values ​​of the first dynamic resistance control signal and the second dynamic resistance control signal at the same turning position point by point, calculates the resistance difference, and when the resistance difference exceeds the preset threshold, the result output module outputs the positioning result.

2. The VR assisted commissioning and fault location method for equipment maintenance of claim 1, wherein: When the bearing inner ring is driven to rotate at a constant angular velocity, the angle encoder generates a hardware trigger pulse at each preset equal angular interval position. The hardware trigger pulse is simultaneously sent by the synchronous trigger module to the angle counter of the angle encoder and the torque sampling circuit of the dynamic torque sensor to form the original data point corresponding to the angle. The data processing module reads the original data points and uses a linear interpolation algorithm to insert multiple intermediate data points between every two adjacent original data points, generating an angle torque curve composed of continuous angle positions and their corresponding instantaneous torque values. The angle torque curve is then preprocessed, and the preprocessed angle torque curve is associated with the bearing's unique identifier code and stored in the maintenance database.

3. The VR assisted commissioning and fault location method for equipment maintenance of claim 2, wherein: In the maintenance database, the rotational torque curves of all bearings of the same model are retrieved based on the unique identification code of the bearing. The curve loading module aligns, superimposes, and averages the rotational torque curves. During the calculation, the curve loading module rotates and aligns each rotational torque curve according to the characteristic rotational position of each rotational torque curve, takes the arithmetic mean of multiple instantaneous torque values ​​at the same rotational position, generates the standard rotational torque curve of the bearing model, and stores it in the maintenance database.

4. The VR-assisted debugging and fault location method for equipment maintenance according to claim 3, characterized in that: The curve loading module maps and associates the corner position in the corner torque curve with the corresponding virtual vertex on the virtual bearing model that has a unique coordinate in three-dimensional space, and uses the instantaneous torque value corresponding to the corner position as the attribute data of the virtual vertex. When the virtual bearing model is driven by a virtual rotation operation command, each virtual vertex on the surface of the virtual bearing model obtains the instantaneous torque value in real time according to its mapped corner position.

5. The VR assisted commissioning and fault location method for equipment maintenance of claim 4, wherein: In the VR scene, the second virtual rotation operation command is applied to the second virtual bearing model bound with a known standard angle torque curve. When the standard instantaneous torque value at a certain angle position is read, the signal generation module outputs an initial control signal to the force feedback data glove. At the same time, the resistance intensity perceived and fed back by the user through the force feedback data glove is recorded. The proportional coefficient of the linear mapping relationship is adjusted iteratively until the resistance intensity fed back by the user matches the preset standard resistance intensity, and then the final linear mapping relationship is determined.

6. The VR assisted commissioning and fault location method for equipment maintenance of claim 5, wherein: It also includes a resistance simulation module, which receives dynamic resistance control signals and controls the actuator of the force feedback data glove to generate corresponding resistance force. At the same time, the resistance simulation module also superimposes a viscous damping component proportional to the rate of change of the rotation angle onto the dynamic resistance based on the rate of change of the current rotation angle. The viscous damping component is used to simulate the viscous resistance caused by the lubricating medium when the bearing rotates. The total resistance perceived by the user through the force feedback data glove is the vector sum of the dynamic resistance and the viscous damping component.

7. The VR assisted commissioning and fault location method for equipment maintenance of claim 6, wherein: When the dynamic resistance value corresponding to the first dynamic resistance control signal calculated based on the test curve continuously exceeds the dynamic resistance value corresponding to the second dynamic resistance control signal obtained based on the reference curve within a certain corner range, and the excess exceeds the preset threshold, the resistance simulation module controls the force feedback data glove to generate a pulse vibration signal in the actuator. The pulse vibration signal is superimposed on the total resistance to prompt the user in the form of tactile pulses.

8. The VR assisted commissioning and fault location method for equipment maintenance of claim 4, wherein: The comparison module performs sliding window averaging calculations on the first dynamic resistance control signal and the second dynamic resistance control signal corresponding to the test curve and the reference curve at multiple consecutive corner positions to obtain two resistance trend curves. Then, the trend resistance values ​​of the two resistance trend curves at the same corner position are compared point by point to calculate the trend resistance difference. When the trend resistance difference exceeds the preset threshold, the result output module also outputs the positioning result, which is the abnormal trend resistance change area exceeding the preset threshold.

9. The VR assisted commissioning and fault location method for equipment maintenance of claim 4, wherein: When the comparison module determines that the trend resistance difference of a certain corner interval exceeds the preset threshold, it retrieves the first dynamic resistance control signal and the second dynamic resistance control signal within the original corner position range corresponding to the abnormal trend resistance change area, and re-compares the first dynamic resistance control signal and the second dynamic resistance control signal at each original corner position within the abnormal trend resistance change area to calculate the fine resistance difference.

10. The VR assisted commissioning and fault location method for equipment maintenance of claim 9, wherein: Only when the proportion of the number of original corner positions in the abnormal trend resistance change area that exceed the preset threshold reaches the confirmation ratio, the abnormal trend resistance change area is determined to be a valid trend fault area, and the result output module is triggered to output the positioning result pointing to the valid trend fault area.

Images