Motor rotor dynamic balance detection system and detection method

Abstract

The invention discloses a motor rotor dynamic balance detection system and method, and the method comprises the steps: carrying out the multi-dimensional sensing deployment and all-condition signal redundancy collection and preprocessing, fusing a signal, building a multi-condition phase locking reference, carrying out the environment error compensation and state marking, extracting features based on the reference, separating multi-source interference, and obtaining high-confidence data, the method combines working conditions and the data to calculate the balance weight and intelligently optimize to generate a target instruction, automatically executes the instruction and verifies the instruction, rechecks multiple working conditions, archives data at the cloud end, performs model self-learning and full-process fault tolerance, outputs a detection result and associates motor full-life-cycle data. Complex working condition adaptability and system robustness are improved, automatic and intelligent detection is realized, detection data can support full-life-cycle operation and maintenance of the motor, and the method is suitable for mass production and on-site operation and maintenance detection of the motor.

Description

Technical Field

[0001] This invention relates to the field of motor testing and dynamic balancing technology, and in particular to a motor rotor dynamic balancing detection system and method, which is applicable to mass production testing and on-site operation and maintenance testing of various types of motors, and can cover motor rotor dynamic balancing detection and correction under constant speed, variable speed, variable load and extreme environment. Background Technology

[0002] In the manufacturing and assembly of electric motors, the dynamic balance performance of the rotor directly affects the motor's vibration level, operating efficiency, and service life. Existing rotor dynamic balancing testing methods typically rely on the trial weight method or vibration measurement under a single steady-state condition. This involves attaching trial weights to the rotor surface and gradually adjusting their position and mass to determine the imbalance. While traditional methods can achieve dynamic balance correction under laboratory conditions, the testing process is time-consuming, reliant on manual experience, and difficult to meet the efficiency requirements of large-scale production. As electric motors develop towards higher speeds and higher power, traditional dynamic balancing methods are increasingly showing their shortcomings in terms of testing accuracy and stability.

[0003] Existing motor rotor dynamic balancing testing technologies suffer from narrow adaptability, covering only constant speed or simple variable speed no-load conditions. The limited range of measurement points makes it easy to miss critical vibration signals. Furthermore, model calculation and parameter optimization are inefficient, leading to long testing cycles per unit during mass production. The systems also exhibit weak anti-interference and fault tolerance capabilities, requiring complete retesting due to signal distortion or process interruptions. The automation level of physical correction and testing loops is low, with high reliance on manual labor, making it difficult to guarantee testing consistency. To address these shortcomings, there is an urgent need for a testing system and method that is adaptable to complex operating conditions, highly robust, and efficient, while simultaneously meeting the dual needs of mass production testing in manufacturing workshops and operational maintenance testing in industrial settings. Summary of the Invention

[0004] The purpose of this invention is to provide a motor rotor dynamic balancing detection system and method to solve the technical problems of narrow working condition adaptability, low calculation efficiency, weak anti-interference fault tolerance, and low degree of automation of traditional detection technology.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] A method for dynamic balancing of an electric motor rotor includes the following steps:

[0007] S1: Deploy multi-dimensional sensors on the motor rotor to complete redundant acquisition and preprocessing of multiple types of signals under all operating conditions;

[0008] S2: The multi-dimensional signals acquired through fusion are used to establish a multi-condition phase-locked reference, and environmental parameters are combined to complete the error compensation and status marking of the reference;

[0009] S3: Based on the compensated phase-locked reference, feature extraction is performed on the signal, and multi-source interference is separated to obtain high-confidence feature data;

[0010] S4: Combine working conditions and high-confidence feature data to complete the counterweight calculation, and optimize the counterweight solution through an intelligent optimization algorithm to generate the target counterweight command;

[0011] S5: Automatically executes target counterweight commands and verifies correction results, performs dynamic balance re-inspection of motor rotor under multiple working conditions, and iteratively corrects if the target is not met;

[0012] S6: Upload the entire detection process data to the cloud for archiving and model self-learning, while implementing end-to-end fault tolerance processing throughout the entire detection process;

[0013] S7: Outputs the motor rotor dynamic balance test results and associates the test data with the motor's full life cycle data.

[0014] In a further embodiment, in step S1, the multi-dimensional sensing deployment involves constructing a network of core measuring points, auxiliary measuring points, and redundant measuring points. Each measuring point is equipped with multi-axis sensing elements, and corresponding sensing elements are configured at the motor shaft, load end, and environmental end. The multiple types of signals include vibration signals, angular position signals, speed signals, torque load signals, environmental signals, and current signals. The full operating conditions include constant speed segment, variable speed frequency sweep segment, variable load segment, and start-stop segment. The preprocessing includes electromagnetic shielding, filtering and noise reduction, and signal validity marking.

[0015] In a further embodiment, in step S2, the multi-dimensional signals are fused to establish a multi-condition phase-locked reference. This involves extracting the speed and load signals and using their rate of change as phase-locked correction factors, establishing a multi-order mechanical frequency parallel tracking channel to complete phase-locking, and forming a basic phase reference after jitter removal processing. The error compensation involves calling a preset environmental error compensation model and combining temperature, humidity, and dust environment parameters to correct the basic phase reference. The status markers include a quality score for the reference and status indicators for locking, tracking, and loss.

[0016] In a further scheme, in step S3, the feature extraction based on the phase-locked loop reference first constructs an equal-angle adaptive sampling sequence to resample the signal in the angular domain, and then uses a lightweight decomposition algorithm to decompose the resampled signal and extract the feature complex amplitude values; the multi-source interference separation uses current, load, and environmental related signals as references to eliminate the influence of electromagnetic, load, and environmental interference on the feature data; the high-confidence feature data is obtained by marking after cross-measurement point consistency verification.

[0017] In a further step, in step S4, the operating conditions are integrated with the speed, load, environmental parameters, and measurement point location information to form a multi-dimensional vector; the counterweight calculation is achieved through a multi-objective intelligent optimization model, which loads the physical constraints of manufacturing and assembly and completes manufacturability processing to generate an initial counterweight solution; the intelligent optimization algorithm verifies the counterweight effect through virtual correction and pre-simulation, and if the threshold is not reached, iteratively optimizes the model parameters, while reusing the model parameters of the same batch or model of motors in the cloud to improve the calculation efficiency.

[0018] In a further embodiment, in step S5, the automated execution of the target counterweight command is achieved by linking the CNC equipment with the rotor positioning system to convert the counterweight command into CNC operation parameters, thereby implementing automatic weight removal or weight addition operations. The correction result verification uses quality detection sensors to accurately verify the correction position and quality value. If the verification fails, the correction is re-executed. The multi-condition re-inspection includes the target condition, the extreme condition, and the variable load condition. If the re-inspection fails to meet the standards, the re-inspection data is fed back to the counterweight calculation stage for rapid iterative correction.

[0019] In a further embodiment, in step S6, the cloud archiving involves archiving the detected operating condition labels, target counterweight instructions, re-inspection results, and sensor data to the motor dynamic balancing big data database; the model self-learning involves performing cluster analysis on the cloud-based data of motors of the same batch and model, and optimizing the model parameters for counterweight calculation through big data training; the end-to-end fault-tolerant processing includes redundant signal switching when the signal is distorted, rapid recovery when the phase reference is lost, breakpoint continuation measurement when the process is interrupted, and parameter compensation when correcting deviations.

[0020] In a further step, in step S7, the test results are output in the form of a test report, which includes the distribution of test points, operating condition coverage, vibration index, counterweight instructions, and re-inspection results. The full life cycle data of the motor includes factory test data, on-site operation data, and maintenance data. A digital twin of the motor is constructed through data association to provide data support for monitoring the dynamic balance status of the motor and for fault early warning.

[0021] This invention also discloses a motor rotor dynamic balancing detection system for realizing a motor rotor dynamic balancing detection method, including a multi-dimensional sensing acquisition module, a multi-condition phase-locked reference module, an intelligent feature calculation module, an adaptive counterweight optimization module, an automated correction execution module, a cloud data collaboration module, and a full-link fault-tolerant module. Each module is networked through an industrial bus or wireless communication method, and the software adopts a collaborative mode of real-time edge calculation and cloud collaborative optimization.

[0022] In a further embodiment, the multi-dimensional sensing acquisition module serves as the system's foundational layer, enabling redundant acquisition and preprocessing of all measurement points and multiple signals.

[0023] The multi-condition phase-locked reference module receives signals from the sensor acquisition module, generates a load-speed dual-dimensional anti-jitter phase reference, and completes error compensation.

[0024] The intelligent feature calculation module completes signal feature extraction and multi-source interference suppression separation based on the phase reference, and outputs high-confidence feature data;

[0025] The adaptive counterweight optimization module completes counterweight calculation and rapid optimization through a multi-objective intelligent optimization model, and generates target counterweight instructions.

[0026] The automated correction execution module executes the counterweight command and completes the correction result verification and multi-condition re-inspection. If the standard is not met, iterative correction is triggered.

[0027] The cloud-based data collaboration module enables cloud-based archiving of detection data, model self-learning, and parameter reuse.

[0028] The full-link fault-tolerant module provides each module with functions such as signal anomaly handling, process breakpoint continuation testing, rapid benchmark recovery, and deviation compensation, ensuring stable operation of the entire system process.

[0029] The present invention has the following beneficial effects:

[0030] This invention achieves comprehensive signal perception across all operating conditions and dimensions through multi-dimensional sensor deployment and redundant signal acquisition, effectively avoiding signal omission and distortion. The establishment of multi-condition phase-locked references and environmental error compensation ensure the accuracy of the phase references and adaptability to complex operating conditions. Feature extraction and multi-source interference suppression separation provide a high-confidence data foundation for subsequent counterweight calculation. Intelligent optimization algorithms combined with cloud-based parameter reuse significantly improve the efficiency of counterweight calculation and optimization. Automated correction execution and multi-condition re-inspection form a closed-loop control, effectively guaranteeing the correction effect of rotor dynamic balance. End-to-end fault-tolerant processing ensures the continuity of the detection process, improving system stability. The correlation between detection data and motor lifecycle data enables deep data reuse, providing reliable data support for motor condition monitoring and fault early warning. Overall, this invention achieves intelligent, automated, and highly robust motor rotor dynamic balance detection, adaptable to the diverse needs of mass production testing in production workshops and industrial field operation and maintenance testing. Attached Figure Description

[0031] Figure 1 This is a flowchart illustrating the steps of the method of the present invention;

[0032] Figure 2 This is a schematic diagram of the overall architecture and module linkage of the detection system of the present invention;

[0033] Figure 3 This is a schematic diagram of the multi-condition phase-locked loop reference establishment and compensation process. Detailed Implementation

[0034] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0035] refer to Figures 1-3 The method for dynamic balancing of a motor rotor, as shown, executes seven steps in sequence: multi-dimensional sensor deployment and signal acquisition preprocessing, multi-condition phase-locked reference establishment and compensation marking, signal feature extraction and multi-source interference suppression and separation, counterweight calculation and intelligent optimization, automated correction and multi-condition re-inspection, cloud data collaboration and end-to-end fault-tolerant processing, and test result output and full lifecycle data association. The specific implementation of each step is as follows:

[0036] Multi-dimensional sensor deployment and redundant acquisition and preprocessing of multiple signal types under all operating conditions

[0037] First, a three-dimensional measurement point network was constructed, consisting of core measurement points, auxiliary measurement points, and redundant measurement points. Core measurement points were placed at critical rotor vibration locations such as the motor bearing housing, rotor axis, and stator end cover. Auxiliary measurement points were placed at locations related to rotor vibration, such as the motor frame and load connection points. Redundant measurement points were set up one-to-one with the core measurement points to achieve cross-validation of the signals. Multi-axis acceleration sensors were installed at each measurement point. Additionally, angular position and speed sensors were configured at the motor shaft, torque load sensors at the load end, temperature, humidity, and dust sensors at the environmental end, and current sensors on the motor drive side, completing the comprehensive deployment of sensors.

[0038] The motor is controlled to enter a full-condition operating state including constant speed, variable speed frequency sweep, variable load, and start-stop segments. Vibration signals, angular position signals, speed signals, torque load signals, environmental signals, and current signals are synchronously collected through various sensors, achieving redundant acquisition of multiple signal types. The acquired raw signals are then subjected to electromagnetic shielding and filtering / denoising processes. The filtering / denoising is implemented using a convolutional filtering algorithm, the mathematical expression of which is:

[0039]

[0040] In the formula, The original discrete vibration signal before filtering. For discrete sampling point numbers, Let be the unit impulse response of the filter. is the filtered discrete vibration signal, and * is the convolution operator.

[0041] After filtering and denoising, the redundant measurement point signals are compared and verified with the core measurement point signals. The Pearson correlation coefficient method is used to determine the signal validity and mark the signals. The mathematical expression of the algorithm is as follows:

[0042]

[0043] In the formula, Here, N represents the correlation coefficient, and N is the number of sampling points in a single batch. The signal value of the i-th sampling point of the core measurement point. The average value of a single batch of signals at the core measurement points. Let i be the signal value of the i-th sampling point of the redundant measurement points. The average signal of a single batch at redundant measurement points; when When the value is ≥0.8, the signal group is marked as a valid signal. After removing invalid signals, the signal preprocessing is completed, providing a reliable signal basis for the establishment of subsequent multi-condition phase-locked loop references.

[0044] Multi-dimensional signal fusion for building multi-condition phase-locked reference and environmental error compensation and status marking

[0045] Speed ​​and load signals are extracted from the preprocessed multi-type signals. The speed change rate and load change rate are calculated separately, and these two parameters are used as phase-locked loop (PLL) correction factors to replace the traditional PLL logic that relies solely on speed. Parallel tracking channels for first-, second-, and third-order mechanical frequencies are constructed. Based on the aforementioned PLL correction factors, the PLL parameters of each channel are dynamically adjusted. The mathematical expression for the phase reference correction algorithm is as follows:

[0046]

[0047] In the formula, This is the corrected real-time phase value. For time, The original real-time phase value obtained by phase-locked loop. This is the correction factor for the rate of change of rotational speed. This represents the real-time rate of change of rotational speed. This is the load change rate correction factor. This represents the real-time torque load change rate.

[0048] After dynamic correction of the phase-locked loop (PLL) parameters, the PLL results are de-jittered to eliminate phase deviation caused by signal jitter and establish a basic phase reference. A preset environmental error compensation model is then invoked, inputting environmental parameters such as temperature, humidity, and dust collected by environmental sensors. The mathematical expression for the environmental error compensation algorithm is as follows:

[0049]

[0050] In the formula, The real-time phase value is the final phase reference. This is the temperature error compensation coefficient. This is the difference between the actual temperature and the standard temperature. This is the humidity error compensation coefficient. This represents the difference between the actual humidity and the standard humidity. This is the dust concentration error compensation coefficient. This represents the difference between the actual dust concentration and the standard dust concentration.

[0051] After correcting the basic phase reference using the compensation value calculated by the above algorithm, the corrected phase reference is scored for quality. At the same time, according to the real-time operating status of the phase reference, its three states of locking, tracking, and loss are marked respectively. Thus, the establishment of a multi-condition phase-locked loop reference is completed. This reference can adapt to the complex operating conditions of motor with variable speed and variable load.

[0052] High-confidence feature data were obtained based on signal feature extraction and multi-source interference suppression separation of the compensated phase-locked loop reference.

[0053] Based on the phase-locked loop reference that completes error compensation and status marking, and combined with the real-time speed of the motor, an equal-angle adaptive sampling sequence is constructed. The mathematical expression for calculating the number of sampling points per revolution is as follows:

[0054]

[0055] In the formula, The number of adaptive sampling points per revolution. This is the floor function operator. The number of reference sampling points at the reference speed. This refers to the real-time speed of the motor. This is the reference speed of the motor.

[0056] Based on the constructed equiangular adaptive sampling sequence, angular domain resampling is performed on the preprocessed vibration and current signals to convert the time-domain signals into angular domain signals, improving the matching degree between the signals and the rotor rotation state. A lightweight sparse decomposition algorithm is then used to decompose the resampled angular domain signals and extract the feature quantities from the signals. The mathematical expression of the algorithm is as follows:

[0057]

[0058] In the formula, For sparsity coefficients Norm, It is a sparse coefficient vector. Let Φ be the vibration signal vector after corner domain resampling, and Φ be the corner domain atom dictionary matrix constructed based on sector ring topology.

[0059] The first-order and second-order feature complex amplitudes in the signal are extracted using the above algorithm. The mathematical expression for the first-order complex amplitude is:

[0060]

[0061] The mathematical expression for the second-order complex amplitude is:

[0062]

[0063] In the formula, It is a first-order complex amplitude. For the real part of the first-order complex amplitude, The imaginary part of the first-order complex amplitude. The imaginary unit, It is a second-order complex amplitude. For the real part of the second-order complex amplitude, It represents the imaginary part of the second-order complex amplitude.

[0064] Using the current signal as the electromagnetic interference suppression reference, the torque load signal as the load interference suppression reference, and the environmental signal as the environmental interference suppression reference, an interference compensation algorithm is used to separate and eliminate the influence of electromagnetic interference, load interference, and environmental interference on the characteristic complex amplitude. The mathematical expression of the algorithm is as follows:

[0065]

[0066] In the formula, The characteristics of the complex amplitude after descrambling. The original complex amplitude characteristics without scrambling. This is the electromagnetic interference compensation coefficient. The complex amplitude value of electromagnetic interference calculated from the current signal. This is the load interference compensation coefficient. The load disturbance complex amplitude value calculated from the torque signal. This is the environmental disturbance compensation coefficient. The complex amplitude of environmental interference calculated for environmental signals.

[0067] The consistency of the descrambled feature data is verified across measurement points. By comparing the same feature data at different measurement points, the verified feature data is marked and retained, thus obtaining high-confidence feature data, which provides accurate data input for subsequent counterweight calculation.

[0068] Weight calculation and intelligent optimization combining working conditions and high-confidence feature data

[0069] By integrating motor speed, load, environmental parameters, and measuring point location information, a multi-dimensional operating condition vector is constructed, the mathematical expression of which is:

[0070]

[0071] In the formula, This refers to the motor speed. For torque load, For ambient temperature, For ambient humidity, For environmental dust concentration, Let m be the location parameter of the m-th measuring point, where m is the total number of measuring points.

[0072] The constructed multi-dimensional operating condition vector, along with the aforementioned high-confidence feature data, is input into a multi-objective intelligent optimization model. Simultaneously, physical constraints from the motor manufacturing and assembly process are loaded into this model, specifically including disabled sectors, upper and lower limits of single-sector mass, and upper limit of total planar mass. The model's solution results undergo manufacturability processing. Through inverse mapping calculations of the model, an initial counterweight solution is generated. This initial counterweight solution is a dual-plane counterweight parameter vector, mathematically expressed as:

[0073]

[0074] In the formula, The counterweight mass of the first equilibrium plane, The phase angle of the counterweight on the first equilibrium plane. The counterweight mass of the second equilibrium plane, The phase angle of the counterweight on the second equilibrium plane.

[0075] A virtual correction simulation is performed by injecting an angular synchronous electromagnetic torque synchronized with the first-order phase into the motor through a driver. This simulation is used to verify the vibration suppression effect of the initial counterweight solution. If the preset effect threshold is not met, the mapping coefficients and coupling weight parameters of the multi-objective intelligent optimization model are iteratively optimized using a particle swarm optimization algorithm. The mathematical expression for the velocity update of the particle swarm algorithm is as follows:

[0076]

[0077] The mathematical expression for position update is:

[0078]

[0079] In the formula, Let ω be the velocity of the i-th particle in dimension d at generation t+1, and ω be the inertial weight. Let be the velocity of the i-th particle in dimension d at generation t. , As a learning factor, , A random number in the range [0,1]. Let be the optimal position for the i-th particle. Let be the position of the i-th particle in dimension d at generation t. This represents the global optimal position of the particle swarm. Let be the position of the i-th particle in the d-th dimension at generation t+1.

[0080] The mathematical expression for the fitness function in particle swarm optimization is:

[0081]

[0082] In the formula, For fitness value, This is the vibration weighting coefficient. This represents the complex amplitude of rotor vibration. For the total mass of the counterweight, This is the weighting coefficient for the counterweight.

[0083] After the parameter iterative optimization is completed by the above particle swarm intelligent optimization algorithm, the optimized target counterweight instruction is generated. At the same time, the model parameters of the same batch or model of motors stored in the cloud are reused to avoid the counterweight calculation of a single motor starting from scratch, effectively improving the overall efficiency of counterweight calculation.

[0084] Automated execution and correction verification of target counterweight commands and dynamic balancing re-inspection under multiple working conditions

[0085] The generated target counterweight command is transmitted to the CNC de-weighting / weighting equipment. This CNC equipment is linked with the rotor sector precision positioning system to convert parameters such as counterweight mass, phase angle, and balance plane in the target counterweight command into CNC operation parameters. The de-weighting or weighting operation is automatically performed on the corresponding balance plane and sector of the rotor, realizing the automated execution of the counterweight command.

[0086] The corrected rotor counterweight position and actual counterweight mass value are precisely verified using quality detection sensors. If the verification result does not meet the requirements of the target counterweight instruction, the automated correction operation is re-executed. If the verification passes, the motor is controlled to operate under target, extreme, and variable load conditions to conduct multi-condition dynamic balancing re-inspection. Vibration signals are collected simultaneously during the re-inspection process, and characteristic amplitude values ​​are extracted and compared with preset acceptance limits. The mathematical expression of the algorithm for determining whether the re-inspection meets the standards is as follows:

[0087]

[0088] In the formula, The vibration reduction rate, The original rotor vibration amplitude value before correction. This is the corrected rotor vibration complex amplitude value; when If the re-inspection rate is ≥85%, the re-inspection is deemed to have passed. If the threshold is not met, the re-inspection data is fed back to the counterweight calculation stage, and rapid iterative correction is performed based on the re-inspection data until the dynamic balance of the rotor meets the acceptance requirements.

[0089] Cloud archiving of the entire testing process, model self-learning, and end-to-end fault tolerance.

[0090] All operational condition labels, target counterweight instructions, correction and verification results, re-inspection data, raw data collected by each sensor element, and extracted feature data generated throughout the entire testing process are uploaded to the cloud and archived in the motor dynamic balancing database, achieving centralized storage and management of testing data. Cluster analysis is performed on the testing data of motors of the same batch and model in the cloud database. The parameters of the multi-objective intelligent optimization model are optimized and adjusted through big data training. The mathematical expression for the iterative optimization of the model parameters is:

[0091]

[0092] In the formula, Let be the model parameters for the (k+1)th iteration. The model parameters for the k-th iteration are... For learning rate, Let L be the loss function with respect to the parameters of the k-th iteration. The partial derivatives of .

[0093] The mathematical expression for the loss function is:

[0094]

[0095] In the formula, K is the total number of samples. This is the predicted value of the complex amplitude of the vibration for the i-th sample. Let be the actual value of the vibration complex amplitude of the i-th sample. The above iterative optimization algorithm enables the model to learn and iterate independently, continuously improving the accuracy and speed of subsequent motor rotor dynamic balance testing.

[0096] Throughout the entire testing process, the operational status of each stage is monitored in real time, and end-to-end fault tolerance is implemented simultaneously: if signal distortion or sensor failure occurs in the sensing acquisition stage, the system automatically switches to the valid signal of redundant measurement points to ensure that the signal acquisition process is not interrupted; if the phase reference stage experiences a phase reference loss, a temporary phase reference is quickly generated using historical valid phase data combined with current and load auxiliary signals to achieve rapid phase reference recovery; if the testing process is paused due to equipment failure, network interruption, or other reasons, the system automatically records the breakpoint location, and resumes the testing process directly from the breakpoint after the fault is cleared, without the need for a full retest; if deviations occur in the virtual correction or automated correction stages, the system automatically triggers the deviation compensation mechanism to quickly adjust model parameters and recalculate and correct, ensuring the continuity and stability of the testing process throughout.

[0097] Test result output and correlation with motor lifecycle data

[0098] Based on various information such as sensor data, feature data, counterweight instructions, correction results, and re-inspection data from the entire testing process, a motor rotor dynamic balance test report is generated. The report clearly includes core information such as the distribution of measuring points, the scope of operating conditions, vibration indicators, details of counterweight instructions, correction results, and re-inspection data, thus completing the formal output of the test results.

[0099] All data generated in this test will be linked and integrated with the motor's factory test data, on-site operation data, and maintenance data to construct a digital twin of the motor. Through this digital twin, the dynamic balance status of the motor will be monitored and analyzed throughout its entire life cycle, providing accurate data support for subsequent dynamic balance fault early warning and operation and maintenance plan formulation, and realizing the deep reuse and value mining of test data.

[0100] This embodiment also provides a motor rotor dynamic balancing detection system for implementing the above detection method. It adopts an architecture of distributed acquisition, edge computing, cloud collaboration, and automated execution. It consists of a multi-dimensional sensing acquisition module, a multi-condition phase-locked reference module, an intelligent feature calculation module, an adaptive counterweight optimization module, an automated correction execution module, a cloud data collaboration module, and a full-link fault-tolerant module. Each module is sequentially connected and has complementary functions. At the hardware level, flexible networking in the workshop or industrial site is achieved through industrial bus or 5G / LoRa wireless communication. At the software level, a collaborative mode of real-time edge computing and cloud collaborative optimization is adopted.

[0101] The system comprises several modules: a multi-dimensional sensing acquisition module (the foundational layer), a multi-dimensional sensing acquisition module (the basic layer), and an adaptive counterweight optimization module. The multi-condition phase-locked loop (PLL) reference module receives the signal data from the multi-dimensional sensing acquisition module, extracts speed and load change rate as correction factors, establishes a multi-order mechanical frequency parallel tracking channel to complete phase locking, generates a basic phase reference, and then combines environmental parameters to perform error compensation. The module also performs quality scoring and status marking on the reference, outputting the compensated PLL reference. An intelligent feature extraction module performs angular domain resampling and lightweight feature extraction based on the PLL reference. It also separates multi-source interference using current, load, and environmental signals as references, performs cross-measurement point consistency verification on the feature data, and outputs high-confidence feature data. The system integrates operating conditions and high-confidence feature data, completes counterweight calculation through a multi-objective intelligent optimization model, iterates and optimizes the counterweight solution using an intelligent optimization algorithm, and reuses cloud parameters to improve calculation efficiency, generating and outputting target counterweight instructions. The automated correction execution module receives the target counterweight instructions and achieves automated correction through the linkage of CNC equipment and rotor positioning system, accurately verifies the correction results, and conducts multi-condition dynamic balance re-inspection. If the standard is not met, iterative correction is triggered. The cloud data collaboration module receives the entire detection process data and completes cloud archiving. It performs cluster analysis on data of the same batch and model to achieve model self-learning and provides parameter reuse support for counterweight calculation. The full-link fault tolerance module provides signal anomaly handling, process breakpoint continuation testing, benchmark rapid recovery, and correction deviation compensation functions for each module of the system, monitors the operating status of each module in real time, and ensures stable operation of the entire system process.

[0102] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting the dynamic balance of a motor rotor, characterized in that, Includes the following steps: S1: Deploy multi-dimensional sensors on the motor rotor to complete redundant acquisition and preprocessing of multiple types of signals under all operating conditions; S2: The multi-dimensional signals acquired through fusion are used to establish a multi-condition phase-locked reference, and environmental parameters are combined to complete the error compensation and status marking of the reference; S3: Based on the compensated phase-locked reference, feature extraction is performed on the signal, and multi-source interference is separated to obtain high-confidence feature data; S4: Combine working conditions and high-confidence feature data to complete the counterweight calculation, and optimize the counterweight solution through an intelligent optimization algorithm to generate the target counterweight command; S5: Automatically executes target counterweight commands and verifies correction results, performs dynamic balance re-inspection of motor rotor under multiple working conditions, and iteratively corrects if the target is not met; S6: Upload the entire detection process data to the cloud for archiving and model self-learning, while implementing end-to-end fault tolerance processing throughout the entire detection process; S7: Outputs the motor rotor dynamic balance test results and associates the test data with the motor's full life cycle data.

2. The method for dynamic balancing of motor rotors according to claim 1, characterized in that, In step S1, the multi-dimensional sensing deployment involves constructing a network of core measuring points, auxiliary measuring points, and redundant measuring points. Each measuring point is equipped with multi-axis sensing elements, and corresponding sensing elements are configured at the motor shaft, load end, and environment end. The multiple types of signals include vibration signals, angular position signals, speed signals, torque load signals, environmental signals, and current signals. The full operating conditions include constant speed segment, variable speed frequency sweep segment, variable load segment, and start-stop segment. The preprocessing includes electromagnetic shielding, filtering and noise reduction, and signal validity marking.

3. The method for dynamic balancing of motor rotors according to claim 1, characterized in that, In step S2, the multi-dimensional signals are fused to establish a multi-condition phase-locked reference. This involves extracting the speed and load signals and using their rate of change as a phase-locked correction factor. A multi-order mechanical frequency parallel tracking channel is established to complete the phase-locking. After jitter removal, a basic phase reference is formed. The error compensation involves calling a preset environmental error compensation model and combining temperature, humidity, and dust environment parameters to correct the basic phase reference. The status markers include a quality score for the reference and status indicators for locking, tracking, and loss.

4. The method for dynamic balancing of motor rotors according to claim 1, characterized in that, In step S3, the feature extraction based on the phase-locked reference first constructs an equal-angle adaptive sampling sequence to resample the signal in the angular domain, and then uses a lightweight decomposition algorithm to decompose the resampled signal and extract the feature complex amplitude value. The separation of multi-source interference is based on current, load, and environmental related signals to eliminate the influence of electromagnetic, load, and environmental interference on the feature data; the high-confidence feature data is obtained by marking after cross-measurement point consistency verification.

5. The method for dynamic balancing of motor rotors according to claim 1, characterized in that, In step S4, the operating conditions integrate speed, load, environmental parameters, and measurement point location information to form a multi-dimensional vector; the counterweight calculation is achieved through a multi-objective intelligent optimization model, which loads the physical constraints of manufacturing and assembly and completes manufacturability processing to generate an initial counterweight solution; the intelligent optimization algorithm verifies the counterweight effect through virtual correction and pre-running, and iteratively optimizes the model parameters if the threshold is not reached, while reusing the model parameters of the same batch or model of motors in the cloud to improve the calculation efficiency.

6. The method for dynamic balancing of motor rotors according to claim 1, characterized in that, In step S5, the automated execution of the target counterweight command is achieved by linking the CNC equipment with the rotor positioning system to convert the counterweight command into CNC operation parameters and implement automatic weight removal or weight addition operations. The verification of the correction result is carried out by the quality detection sensor to accurately verify the correction position and quality value. If the verification fails, the correction is re-executed. The multi-condition re-inspection includes the target condition, the extreme condition, and the variable load condition. If the re-inspection fails to meet the standard, the re-inspection data is fed back to the counterweight calculation stage for rapid iterative correction.

7. The method for dynamic balancing of motor rotors according to claim 1, characterized in that, In step S6, the cloud archiving refers to archiving the detected working condition labels, target counterweight instructions, re-inspection results, and sensor data to the motor dynamic balancing big data database; the model self-learning refers to performing cluster analysis on the same batch and model of motor data in the cloud, and optimizing the model parameters for counterweight calculation through big data training; the end-to-end fault tolerance processing includes redundant signal switching when the signal is distorted, rapid recovery when the phase reference is lost, breakpoint continuation measurement when the process is interrupted, and parameter compensation when correcting deviations.

8. The method for dynamic balancing of motor rotors according to claim 1, characterized in that, In step S7, the test results are output in the form of a test report, which includes the distribution of test points, operating condition coverage, vibration index, counterweight instructions, and re-inspection results. The full life cycle data of the motor includes factory test data, on-site operation data, and maintenance data. A digital twin of the motor is constructed through data association to provide data support for monitoring the dynamic balance status of the motor and early warning of faults.

9. A motor rotor dynamic balancing detection system for implementing the motor rotor dynamic balancing detection method according to any one of claims 1-8, characterized in that, It includes a multi-dimensional sensing and acquisition module, a multi-condition phase-locked reference module, an intelligent feature calculation module, an adaptive counterweight optimization module, an automated correction execution module, a cloud data collaboration module, and a full-link fault-tolerant module. Each module is networked through an industrial bus or wireless communication. At the software level, it adopts a collaborative mode of real-time edge calculation and cloud collaborative optimization.

10. The motor rotor dynamic balancing detection system according to claim 9, characterized in that, The multi-dimensional sensing acquisition module enables redundant acquisition and preprocessing of all measurement points and multiple signals. The multi-condition phase-locked reference module receives signals from the sensor acquisition module, generates a load-speed dual-dimensional anti-jitter phase reference, and completes error compensation. The intelligent feature calculation module completes signal feature extraction and multi-source interference suppression separation based on the phase reference, and outputs high-confidence feature data; The adaptive counterweight optimization module completes counterweight calculation and rapid optimization through a multi-objective intelligent optimization model, and generates target counterweight instructions. The automated correction execution module executes the counterweight command and completes the correction result verification and multi-condition re-inspection. If the standard is not met, iterative correction is triggered. The cloud-based data collaboration module enables cloud-based archiving of detection data, model self-learning, and parameter reuse. The end-to-end fault-tolerant module provides each module with functions such as signal anomaly handling, process breakpoint continuation testing, rapid baseline recovery, and deviation compensation, ensuring stable operation of the entire system process.

Images