Mechanical gear transmission back clearance detection system and detection method

Abstract

The invention discloses a mechanical gear transmission back clearance detection system and detection method, and belongs to the technical field of mechanical transmission and electromechanical detection. The system comprises an excitation module, a signal processing module, a phase analysis module and a back clearance judgment module. The excitation module injects an alternating excitation current with a preset frequency into the motor winding; the signal processing module synchronously collects excitation signals and induction signals generated by motor magnetic field induction; the phase analysis module calculates the phase difference between the two signals in real time; and the back clearance judgment module judges the mechanical back clearance by monitoring a sudden change event of the phase difference and calculates the size of the mechanical back clearance. According to the system, a mechanical sensor does not need to be additionally arranged, real-time, online and high-precision measurement of the back clearance is achieved by detecting microsecond jump of the electromagnetic phase of the motor at the moment of motion reversing, and the problems that a traditional grating or encoder comparison method is high in cost, complex in installation, incapable of detecting the transient back clearance, difficult to integrate and the like are solved. The method is especially suitable for health monitoring and predictive maintenance of precision transmission mechanisms such as robot joints, servo systems and numerical control machine tools.

Description

Technical Field

[0001] This invention relates to the field of mechanical gear transmission backlash detection technology, and in particular to a mechanical gear transmission backlash detection system and detection method. Background Technology

[0002] Gear backlash is a key factor affecting the positioning accuracy, dynamic response, and reliability of precision mechanical systems, and achieving high-precision, real-time online detection of it has significant engineering value. Currently, gear backlash detection mainly relies on adding high-precision physical sensors to the transmission chain for direct or indirect measurement, resulting in several typical technical approaches. Among these, the most common include the grating comparison method and the encoder comparison method.

[0003] The grating comparison method requires the installation of high-precision grating rulers at both the input and output ends, and the back gap is calculated by comparing the positional difference between the two. Although this method can achieve high static accuracy, the grating rulers are expensive, installation is limited by mechanical space, the system is complex, has high requirements for environmental cleanliness, and it is difficult to capture transient back gap changes during operation.

[0004] The encoder comparison method calculates backlash by installing rotary encoders on both the input and output shafts and using the angular difference between the two encoders. However, this method is limited by the encoder's resolution and sampling frequency, and its response speed is insufficient to reflect the transient backlash during high-speed commutation. Furthermore, the installation of dual encoders also faces space constraints, and the introduced mechanical alignment errors directly affect measurement accuracy, significantly increasing system cost.

[0005] In summary, existing technologies generally suffer from inherent defects such as reliance on additional sensors leading to complex structures, high costs, and limited installation, as well as the difficulty in achieving real-time and transient backlash detection during equipment operation due to limitations in sensor response speed or system inertia. Summary of the Invention

[0006] Therefore, it is necessary to provide a mechanical gear transmission backlash detection system and method to address the technical problem of difficulty in achieving real-time and transient backlash detection during equipment operation.

[0007] A mechanical gear transmission backlash detection system, comprising:

[0008] The excitation module is used to inject an alternating excitation current of a preset frequency into the windings of the motor under test;

[0009] The signal processing module is used to acquire the excitation signal corresponding to the excitation current, and the induced signal generated by the excitation current in the motor magnetic field;

[0010] The phase analysis module is used to calculate the phase difference between the excitation signal and the sensing signal in real time.

[0011] The backlash determination module is used to determine that backlash has occurred in the mechanical transmission chain when a sudden event that conforms to preset characteristics is detected in the phase difference, and to determine the amount of backlash based on the amount of change in the phase difference.

[0012] In one embodiment, the backlash determination module is configured to initiate the detection of the phase difference abrupt event when it detects that the absolute value of the actual rotational speed of the motor is lower than a first preset threshold and the actual acceleration direction is reversed; and / or,

[0013] When a reciprocating position micro-motion command with an amplitude less than the second preset threshold is received, the detection of the phase difference abrupt event is initiated.

[0014] In one embodiment, the phase analysis module includes:

[0015] A digital phase-locked loop unit is used to perform real-time phase tracking of the excitation signal and the sensing signal, and output a first phase difference signal; and,

[0016] The Fast Fourier Transform (FFT) analysis unit is used to perform spectral analysis on the induced signal, extract phase information at the preset frequency, and output a second phase difference signal.

[0017] In one embodiment, the back gap determination module is configured to determine a valid back gap phase abrupt change event when the following conditions are met simultaneously:

[0018] (a) The rate of change of the first phase difference signal exceeds a third preset threshold;

[0019] (b) The duration of the phase abrupt change state is less than a fourth preset threshold;

[0020] (c) The signal energy change at the preset frequency output by the Fast Fourier Transform Analysis Unit exceeds the fifth preset threshold.

[0021] In one embodiment, it further includes: a calibration module, used to control the excitation module to perform frequency sweep excitation within a preset frequency band during system initialization, and to select the optimal preset frequency based on the signal response quality at each frequency point.

[0022] In one embodiment, the backlash determination module converts the change in phase difference into a mechanical backlash angle or linear displacement based on a pre-stored phase difference-backlash conversion model.

[0023] In one embodiment, it further includes a health management module for recording historical backlash data, performing trend analysis and fitting, and generating an equipment health index or wear warning signal based on the changing trend of backlash.

[0024] In one embodiment, the system is integrated into a servo drive, robot joint controller, or control unit of a numerical control system.

[0025] The aforementioned mechanical gear transmission backlash detection system is based on electromagnetic phase analysis and exhibits several significant advantages over traditional detection systems. First, its greatest advantage lies in its "sensorless" design. The system utilizes the motor itself as the detection element, eliminating the need for additional high-precision linear encoders, other physical sensors, etc. This greatly simplifies the mechanical structure, significantly reduces system hardware costs and installation complexity, and avoids the additional errors or space constraints introduced by sensor installation. Second, this solution achieves unprecedented high response speed and real-time performance. Because the detection object is the phase of the electromagnetic signal, its response speed can reach the microsecond level, accurately capturing transient backlash that occurs during changes in motion direction, which is difficult to detect using traditional methods. This achieves true online real-time monitoring without affecting any normal equipment operation. Third, this method possesses extremely high detection accuracy and sensitivity. Utilizing high-resolution phase tracking technology, extremely minute phase jumps can be identified. Corresponding to mechanical backlash, its resolution reaches the 0.005° to 0.02° arcsecond level. For linear transmission mechanisms, the detection sensitivity reaches the 1 to 5 micrometer level, fully meeting the detection requirements of high-precision transmission systems. Finally, the system boasts extremely high versatility and integration. Its principle is not dependent on specific gear types and can be widely applied to various transmission forms such as harmonic reducers, RV reducers, planetary gearboxes, spur gears, and helical gears. The entire system can be directly integrated into existing servo drives, robot controllers, or CNC systems via software algorithms and additional circuitry. It features low power consumption, ease of deployment, and provides robust underlying technical support for predictive maintenance and health management of equipment.

[0026] This invention also provides a method for detecting backlash in mechanical gear transmissions, comprising the following steps:

[0027] Inject an alternating excitation current of a preset frequency into the windings of the motor under test;

[0028] The excitation signal corresponding to the excitation current and the induced signal generated by the excitation current in the motor magnetic field are collected.

[0029] The phase difference between the excitation signal and the sensing signal is calculated in real time.

[0030] Monitor whether the phase difference undergoes a sudden change event that meets preset characteristics;

[0031] If so, it is determined that backlash has occurred in the mechanical transmission chain, and the amount of backlash is determined based on the change in the phase difference.

[0032] In one embodiment, the monitoring step is performed only when the absolute value of the actual rotational speed of the motor is lower than a first preset threshold and the actual acceleration direction is reversed, and / or when the system receives a reciprocating position micro-motion command with an amplitude less than a second preset threshold, the high-sensitivity monitoring of the phase difference abrupt event is performed.

[0033] The aforementioned method for detecting backlash in mechanical gear transmissions firstly achieves intelligent context awareness during the detection process. By strongly correlating the detection action with specific mechanical motion states such as "zero speed crossover" and "small-amplitude reciprocating commands," detection resources are concentrated on the critical moments when backlash is most likely to be exposed. This significantly improves the targeting and signal-to-noise ratio of the detection, avoids meaningless continuous calculations, and saves processor resources. Secondly, the method constructs a multi-level, cross-validated reliable judgment process. From basic phase-locked loop real-time tracking to frequency domain FFT verification, and then to composite criteria combining time, energy, and other dimensions, a rigorous "diagnostic logic chain" is formed. This ensures that an alarm is triggered only when mechanical backlash actually occurs, effectively resisting false signals caused by power grid interference, driver switching noise, mechanical vibration, etc., resulting in extremely high reliability of the detection results. Finally, the method inherently possesses data accumulation and trend analysis capabilities. Since the entire process can be automated, backlash data detected in each reciprocating motion can be automatically recorded and stored. Over long-term operation, this data provides a direct and accurate foundation for plotting gear wear curves, predicting remaining service life, and implementing condition-based predictive maintenance, elevating simple inspection to the level of intelligent health management. Attached Figure Description

[0034] Figure 1 This is a schematic block diagram of a mechanical gear transmission backlash detection system in one embodiment.

[0035] Figure 2 This is a schematic diagram of the flow structure of a mechanical gear transmission backlash detection method in one embodiment. Detailed Implementation

[0036] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below. In the description of the present invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the present invention.

[0037] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0038] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0039] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0040] It should be noted that when an element is referred to as being "fixed to" or "set on" another element, it can be directly on the other element or there may be an intervening element. When an element is considered to be "connected to" another element, it can be directly connected to the other element or there may be an intervening element. The terms "vertical," "horizontal," "upper," "lower," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0041] like Figure 1 As shown, in a specific embodiment of the present invention, a specific implementation of a mechanical gear transmission backlash detection system is provided. This system aims to detect backlash in a gear transmission chain in real time online without the need for external physical sensors. The mechanical gear transmission backlash detection system mainly includes four functional modules: an excitation module, a signal processing module, a phase analysis module, and a backlash determination module.

[0042] The core of the excitation module is a programmable alternating current generator, whose output is connected to the windings of the motor under test (e.g., a servo motor). This generator produces an AC excitation current with precisely controllable frequency and amplitude. Depending on the implementation scenario, it can be configured for single-frequency excitation, multi-frequency excitation, or frequency sweep mode. For example, in a preferred embodiment, the excitation module is configured to inject an alternating current with a frequency of 600Hz and an amplitude of 0.5A into the motor windings. This current amplitude is much smaller than the operating current that generates effective torque in the motor, therefore it will not interfere with the normal operation of the equipment, but only establishes a stable high-frequency electromagnetic field reference in the motor's magnetic circuit.

[0043] The signal processing module includes a high-speed synchronous acquisition circuit. It simultaneously acquires two key signals: the first is the "excitation signal," directly output from the excitation module, representing the original phase of the injected current; the second is the "induced signal," obtained from the motor feedback loop (such as a flux sensor built into the motor, or indirectly obtained by monitoring the back electromotive force), which reflects the changes in the internal magnetic field of the motor under the action of the excitation current. The acquisition of these two signals must be strictly synchronized, with a sampling rate typically between 50kHz and 200kHz to ensure the capture of microsecond-level transient changes.

[0044] The phase analysis module receives two synchronization signals from the signal processing module. Its core task is to calculate the instantaneous phase difference Δφ(t) between the excitation signal and the induced signal in real time with extremely high resolution and speed. In one specific implementation, this module uses digital phase-locked loop (DPLL) technology to continuously track and output this phase difference. DPLL has a phase resolution of up to 0.01° to 0.1° and a tracking delay of less than 20 microseconds, enabling it to quickly respond to minute changes in phase.

[0045] The backlash detection module continuously monitors the real-time phase difference Δφ(t) curve output by the phase analysis module. It incorporates an intelligent algorithm to identify characteristic phase abrupt changes caused by mechanical backlash. When the algorithm detects a phase difference jump conforming to a preset pattern within a very short time (e.g., a sharp increase in the phase change rate, with the jump duration in milliseconds), it determines that a backlash event has occurred. Subsequently, the module converts the phase abrupt change Δφ into the mechanical angular backlash Δθ (calculated using the formula Δθ = k × Δφ) based on a preset calibration coefficient k, ultimately outputting the quantified backlash value. This module is also responsible for generating alarm signals and triggering data recording.

[0046] The aforementioned mechanical gear transmission backlash detection system is based on electromagnetic phase analysis and exhibits several significant advantages over traditional detection methods. First, its greatest advantage lies in its "sensorless" design. The system utilizes the motor itself as the detection element, eliminating the need for additional high-precision linear encoders, other physical sensors, etc. This greatly simplifies the mechanical structure, significantly reduces system hardware costs and installation complexity, and avoids the additional errors or space constraints introduced by sensor installation. Second, this solution achieves unprecedented high response speed and real-time performance. Since the detection object is the phase of the electromagnetic signal, its response speed can reach the microsecond level, accurately capturing transient backlash that occurs during changes in motion direction, which is difficult to detect using traditional methods. This achieves true online real-time monitoring without affecting any normal equipment operation. Third, this method possesses extremely high detection accuracy and sensitivity. Utilizing high-resolution phase tracking technology, extremely minute phase jumps can be identified. Corresponding to mechanical backlash, its resolution reaches the 0.005° to 0.02° arcsecond level. For linear transmission mechanisms, the detection sensitivity reaches the 1 to 5 micrometer level, fully meeting the detection requirements of high-precision transmission systems. Finally, the system boasts extremely high versatility and integration. Its principle is not dependent on specific gear types and can be widely applied to various transmission forms such as harmonic reducers, RV reducers, planetary gearboxes, spur gears, and helical gears. The entire system can be directly integrated into existing servo drives, robot controllers, or CNC systems via software algorithms and additional circuitry. It features low power consumption, ease of deployment, and provides robust underlying technical support for predictive maintenance and health management of equipment.

[0047] It is worth mentioning that, corresponding to the above systems, such as Figure 2 As shown, the present invention also provides a method 10 for detecting backlash in mechanical gear transmissions, the method 10 comprising the following steps:

[0048] S101. Inject an alternating excitation current of a preset frequency into the windings of the motor under test.

[0049] S102. Collect the excitation signal corresponding to the excitation current, and the induced signal generated by the excitation current in the motor magnetic field.

[0050] S103. Calculate the phase difference between the excitation signal and the sensing signal in real time.

[0051] S104. Monitor whether the phase difference undergoes a sudden change event that meets preset characteristics.

[0052] S105. If so, it is determined that backlash has occurred in the mechanical transmission chain, and the amount of backlash is determined based on the change in the phase difference.

[0053] The above steps demonstrate the specific implementation flow of the mechanical gear transmission backlash detection method 10. This method is executed in a control unit (such as a DSP or FPGA) via software logic, and specifically includes the following sequential steps:

[0054] The first step is to inject a specific excitation current. When the device starts up or enters detection mode, the control unit commands the internal excitation circuit to inject a set of preset parameters of alternating excitation current into the three-phase windings or designated windings of the servo motor. For example, a three-frequency combination current of 523Hz, 937Hz, and 1500Hz is used, with the amplitude of each frequency current controlled at around 0.2A, to superimpose a stable high-frequency magnetic flux reference containing multiple frequency components in the air gap magnetic field of the motor.

[0055] The second step involves synchronously acquiring excitation and induction signals. Using a high-precision ADC channel, the voltage / current signal at the injection point is synchronously acquired as the "excitation reference signal," while the signal output from the motor encoder signal processing chip, or calculated by an observer, related to the motor rotor flux linkage, is acquired as the "induction feedback signal." The acquisition process is performed at a constant high sampling rate (e.g., 100kHz) to ensure data synchronization.

[0056] The third step is to calculate the dynamic phase difference in real time. The core of processing the two acquired digital signals is to calculate the phase difference φ_in(t) - φ_out(t) between them in real time. In implementation, an optimized digital phase-locked loop algorithm is used to phase-lock the two signals, directly outputting the instantaneous phase difference Δφ(t). Simultaneously, for verification, a parallel computing thread is established to perform a Fast Fourier Transform on the induced feedback signal every 256 points, extracting the phase values ​​at the excitation frequency points (523Hz, 937Hz, 1500Hz) as auxiliary verification values ​​for the phase difference.

[0057] The fourth step is intelligent monitoring of phase change events. The system doesn't blindly monitor phase differences at all times; instead, it intelligently determines when to enter a "high-sensitivity detection window." For example, when the servo system's speed loop feedback shows the motor speed drops to within ±1 rpm, and the accelerometer or current loop information indicates the torque direction is about to reverse, the system determines that the device has entered the "reciprocating motion zero zone," at which point a high-frequency, high-sensitivity phase change monitoring algorithm is automatically activated.

[0058] The fifth step is to determine and calculate the backlash. Within the sensitive window, if the calculated phase difference Δφ(t) meets the composite judgment conditions (e.g., the rate of change exceeds 2° / ms, the duration of the abrupt change is less than 5ms, and the corresponding frequency FFT energy rises by more than 25%), a backlash event is confirmed. Subsequently, based on the conversion coefficient k obtained from pre-shipment calibration or self-learning, the phase abrupt change Δφ is converted into the mechanical backlash angle Δθ using the formula Δθ = k × Δφ. This result can be displayed in real time or sent to the host computer via a communication interface.

[0059] The aforementioned method for detecting backlash in mechanical gear transmissions firstly achieves intelligent context awareness during the detection process. By strongly correlating the detection action with specific mechanical motion states such as "zero speed crossover" and "small-amplitude reciprocating commands," detection resources are concentrated on the critical moments when backlash is most likely to be exposed. This significantly improves the targeting and signal-to-noise ratio of the detection, avoids meaningless continuous calculations, and saves processor resources. Secondly, the method constructs a multi-level, cross-validated reliable judgment process. From basic phase-locked loop real-time tracking to frequency domain FFT verification, and then to composite criteria combining time, energy, and other dimensions, a rigorous "diagnostic logic chain" is formed. This ensures that an alarm is triggered only when mechanical backlash actually occurs, effectively resisting false signals caused by power grid interference, driver switching noise, mechanical vibration, etc., resulting in extremely high reliability of the detection results. Finally, the method inherently possesses data accumulation and trend analysis capabilities. Since the entire process can be automated, backlash data detected in each reciprocating motion can be automatically recorded and stored. Over long-term operation, this data provides a direct and accurate foundation for plotting gear wear curves, predicting remaining service life, and implementing condition-based predictive maintenance, elevating simple inspection to the level of intelligent health management.

[0060] To make detection more efficient and accurate, this embodiment employs refined intelligent control over the timing of detection activation, known as a "trigger detection window mechanism." The system or method does not continuously operate at maximum sensitivity; instead, it dynamically activates and deactivates the high-precision backlight detection mode based on the actual operating status of the equipment.

[0061] One specific trigger condition is "speed zero-crossing detection." The system monitors the motor's rotational speed ω in real time. When the absolute value of the rotational speed drops to an extremely low level (e.g., |ω| < 1 rpm), this usually means that the motor is at a critical point of reversing its direction of motion. Simultaneously, the system determines whether the motor's acceleration direction is reversing by observing current commands or acceleration estimates. When both conditions are met, and this state lasts for a very short time (e.g., less than 20 milliseconds), the system determines that the device has entered the "reciprocating commutation zone," which is the critical moment when backlash is most likely to be "exposed" and measured. At this point, the system immediately switches the phase analysis module and backlash determination module to the highest performance mode, ready to capture any possible instantaneous phase jumps.

[0062] Another practical trigger condition is "small-amplitude micro-motion command detection." In many precision alignment, vibration suppression, or force control scenarios, the equipment receives reciprocating position adjustment commands with extremely small amplitudes. This system monitors the position commands sent by the upper-level controller. When it detects that the position command is oscillating at high frequency within a very small angular range (e.g., ±0.1° to ±2°), even if the actual rotational speed has not reached strict zero speed, the system considers this a typical behavior pattern that easily triggers backlash impact. Once this pattern is identified, the system automatically activates the high-sensitivity backlash detection window.

[0063] In this way, by introducing these intelligent triggering mechanisms, the system can take the initiative to measure at the most appropriate time, just like an experienced engineer, thereby minimizing the average computational load and false alarm probability of the system while ensuring an extremely high capture rate.

[0064] To acquire phase difference information with high reliability and accuracy, this embodiment employs a dual-channel parallel processing phase analysis architecture. The core of this architecture consists of two independent yet mutually verifying phase extraction mechanisms: a main channel for real-time phase tracking based on a digital phase-locked loop (DPLL), and an auxiliary verification channel for frequency domain phase extraction based on a fast Fourier transform (FFT).

[0065] The main channel is the primary channel for phase analysis. The DPLL is a closed-loop feedback control system that synchronizes the phase of its internal oscillator with the phase of the input signal. In this application, the excitation signal is used as the reference signal input to the DPLL's reference terminal, and the induced signal is used as the signal to be tracked input to the DPLL's feedback terminal. The DPLL dynamically adjusts its internal state, outputting the instantaneous phase difference Δφ_pll(t) between the two signals in real time. The advantages of this method are its extremely fast response speed (delay <20μs) and extremely high phase resolution (up to 0.01°), perfectly capturing transient changes. Its sampling rate is set between 50kHz and 200kHz to accommodate different frequency requirements and processor capabilities. The DPLL also integrates anti-aliasing filtering and noise suppression algorithms to initially clean the signal.

[0066] In the auxiliary verification channel, to combat strong noise interference and provide a verification benchmark, the system synchronously runs a frequency domain analysis channel based on FFT. This channel performs 64-256 point FFT operations on the induced signal at a slightly lower update rate (e.g., once every 1 millisecond). Through FFT, the signal is transformed from the time domain to the frequency domain. The algorithm accurately locates the spectral line of the excitation current frequency in the frequency domain (e.g., if the excitation is 600Hz, it finds the amplitude peak near 600Hz). Then, the phase value at that frequency point is directly calculated using complex spectrum data. The calculation formula is: φ fft =arctan(Im(X[f0]) / Re(X[f0])), where X[f0] is the complex value at the excitation frequency f0 in the FFT result, and Re and Im represent the real and imaginary parts, respectively. Simultaneously, the phase of the excitation signal at the same frequency point is calculated, and the difference between the two yields the phase difference Δφ_fft(t) based on the frequency domain. The advantages of this channel are strong anti-interference capability, effectively filtering out noise unrelated to the excitation frequency (such as driver PWM harmonics and power grid interference), and stable and reliable phase values, but the update speed is relatively slow.

[0067] In this way, by maintaining these two channels, the system achieves an optimal balance between speed and reliability. Under most stable conditions, the DPLL output is used as the standard; when encountering strong interference or requiring final confirmation, the results of the FFT channel are referenced for cross-validation and correction.

[0068] Understandably, simply detecting a phase change is insufficient to determine the occurrence of a backslot; a rigorous algorithm is necessary to distinguish between genuine backslot transitions and phase fluctuations caused by other disturbances. The backslot determination module in this embodiment employs an algorithm based on multi-dimensional feature synthesis, significantly improving the accuracy of the determination.

[0069] The determination occurs within the "back gap sensitive window" opened by the intelligent trigger detection window mechanism. The system intensively monitors the real-time phase difference Δφ_pll(t) of the main channel output, while simultaneously considering information from the FFT auxiliary channel. A valid back gap event must simultaneously or mostly satisfy the following set of composite conditions:

[0070] 1. First-order phase change rate (gradient) exceeding the limit: This is the most direct criterion for detecting sudden changes. The algorithm calculates the differential of the phase difference with respect to time in real time, i.e., dΔφ / dt. When the absolute value of this differential exceeds the preset threshold φ_thresh, a preliminary alarm is triggered. The threshold setting needs to consider the system noise level, with typical values ​​between 0.5° / ms and 3° / ms. Too small a threshold will lead to false alarms, while too large a threshold will miss small back gaps. The formula is expressed as: |dΔφ(t) / dt|>φ thresh This condition ensures that what is detected is a "rapid jump" rather than a slow temperature drift.

[0071] 2. Phase Jump Duration Characteristics: The impact caused by mechanical backlash is transient, and the resulting phase anomaly recovers quickly. Therefore, the algorithm measures the total duration T_dwell from the onset of the phase jump to its stabilization. In real backlash events, T_dwell is typically very short, ranging from 1 to 10 milliseconds. The system sets a time upper limit (e.g., 10 ms). Only when the duration of the phase jump event is less than this upper limit is it considered to meet the transient characteristics of backlash. This effectively filters out longer time-history phase shifts caused by sudden load changes, continuous slippage, etc.

[0072] 3. FFT Energy Ratio Verification: When backlash occurs, minute impacts in the transmission chain can trigger high-frequency vibrations, which can sometimes modulate the inductive signal. The algorithm monitors the change in signal energy (i.e., the square of the modulus of the complex value X[f0]) at the excitation frequency f0 in the FFT channel. If the energy at this frequency point changes significantly compared to before the phase abrupt change (e.g., an increase or decrease of more than 20%) when a phase abrupt change event occurs, this is strong evidence that the spectral structure of the inductive signal has undergone an impact-related change, further supporting the conclusion of backlash determination.

[0073] 4. Multi-channel consistency check: This is the final "decision" step. The system integrates all the above information and compares the results of the DPLL channel with those of the FFT channel. The priority of the decision logic is as follows:

[0074] Highest confidence level: The DPLL channel detects a fast phase jump (satisfying conditions 1 and 2), and the FFT channel also shows a synchronous phase jump at the corresponding time.

[0075] High confidence level: The DPLL channel detected a rapid phase jump (satisfying conditions 1 and 2), and although the FFT channel did not show a significant phase jump, its energy at the excitation frequency point changed significantly (satisfying condition 3).

[0076] Careful verification is required: If only the DPLL channel detects a jump, while the FFT channel shows no abnormalities, the system may mark it as a "suspected event," requiring further evaluation based on multiple subsequent tests or operational information.

[0077] Thus, through this multi-dimensional and multi-channel comprehensive judgment combining "AND" and "OR" logic, the system can separate real back gap events from complex industrial environmental noise with extremely high confidence.

[0078] To ensure that the system maintains optimal detection performance under different motors and operating conditions, this embodiment integrates a powerful adaptive calibration function, mainly reflected in the automatic optimization of the excitation frequency.

[0079] The "frequency sweep calibration mode" can be activated upon initial system power-on, motor replacement, or routine maintenance. In this mode, the control unit instructs the excitation module to sequentially output test excitation currents of different frequencies within a pre-set wide frequency band (e.g., from 100 Hz to 2000 Hz) in fixed steps (e.g., 10 Hz or 50 Hz). For each output frequency f_i, the system performs the following operations:

[0080] 1. Inject an excitation current with a constant amplitude.

[0081] 2. After stabilization, collect the induction signal for a period of time.

[0082] 3. Calculate the signal-to-noise ratio (SNR) of the induced signal at this frequency. The SNR can be calculated based on the FFT results, comparing the amplitude of the spectral line at the excitation frequency with the average amplitude of the adjacent noise band.

[0083] After scanning the entire frequency band, the system will obtain a frequency-signal-noise ratio (SNR) curve. The algorithm will automatically select one to three frequency points with the highest SNR as the "optimal excitation frequencies" for subsequent long-term operation. For example, if the scan reveals that the SNR is prominent at 523Hz, 937Hz, and 1500Hz and they are distributed across different frequency bands, the system may be configured in a three-frequency excitation mode. The benefits of selecting high SNR frequencies are: improved signal purity, enhanced anti-interference capability, and thus indirectly improved phase detection accuracy and backlash determination reliability.

[0084] In addition, the calibration module may include compensation for temperature drift. The system monitors the motor winding resistance or ambient temperature and, based on a pre-stored thermal model, fine-tunes the amplitude or phase reference of the excitation current to counteract the effects of temperature changes on electromagnetic parameters, ensuring the stability of long-term measurements.

[0085] Understandably, after detecting a phase change Δφ, it needs to be converted into a mechanical backlash quantity (angle Δθ or linear displacement Δs) with practical engineering significance. This embodiment achieves this conversion by establishing and applying a "phase difference-backlash quantity" conversion model.

[0086] The core of the model is a proportionality coefficient k. Its theoretical basis lies in the fact that when there is a backlash in the transmission chain, the motor shaft will experience a tiny "idle" angle at the moment of commutation. This idle angle Δθ causes an instantaneous change in the relative position between the motor rotor and the stator magnetic field, resulting in a sudden change Δφ in the phase of the induced magnetic field. In the ideal linear model, the two are directly proportional: Δθ = k × Δφ, where the coefficient k is a comprehensive parameter that incorporates various factors such as the number of motor pole pairs, the overall reduction ratio of the transmission chain, and the magnetic coupling coefficient.

[0087] There are two methods to obtain the accurate value of k:

[0088] 1. Factory Calibration Method: After the equipment is assembled, use an external high-precision instrument (such as a laser interferometer or ultra-high precision encoder) to measure the actual backlash of a small section of the transmission chain, and simultaneously run this detection system to record the corresponding phase change Δφ. By averaging multiple measurements, the k-value of this specific equipment / transmission chain can be directly calculated and stored in the system memory.

[0089] 2. Self-learning method: For some intelligent systems, in the early stages of their lifespan, a series of small reciprocating motions with known amplitudes can be actively executed, assuming that the initial backlash is zero or known, and the k value can be deduced from the detected phase changes. Alternatively, during operation, the k value can be slowly corrected online through the statistical correlation of long-term data.

[0090] It should be noted that once the value of k is determined, the system can output the quantized backlash in real time. For example, if a phase jump Δφ = 0.5° is detected, and the calibration coefficient k = 0.04 ° / (electric angle °) is known, then the output mechanical backlash angle Δθ = 0.04 * 0.5 = 0.02° can be calculated. If the pitch circle radius of the output gear is known, it can be further converted into linear backlash.

[0091] The detection system in this embodiment can not only provide instantaneous back gap values, but also achieve a leap from "detection" to "diagnosis" and "prediction" through an integrated health management module.

[0092] This module is responsible for storing and managing historical detection data. Whenever a backlash event is confirmed and quantified, its timestamp, backlash amount Δθ, and contextual information such as the operating conditions at that time (e.g., load, speed) are recorded in non-volatile memory.

[0093] Based on this time-series data, the health management module performs the following advanced analyses:

[0094] 1. Moving average and filtering: The original back clearance data is processed by moving average to smooth out the random fluctuations of a single measurement and obtain a smooth curve that reflects the true wear trend.

[0095] 2. Mechanical cycle alignment: For periodically operating equipment, aligning the backlash data with the mechanical cycle of the equipment (such as a fixed motion range of a robot joint) allows for a clearer observation of the backlash growth under specific actions.

[0096] 3. Long-term trend fitting: Curve fitting is performed on the smoothed backlash-time or backlash-operating cycle data. Commonly used models include linear models (predicting uniform wear) and exponential models (predicting accelerated wear). The fitted curve can be extrapolated to predict the time point when the backlash reaches a preset alarm threshold or failure threshold.

[0097] 4. Health Index Calculation: The system calculates a "Transmission Chain Health Index (H)" between 0 and 100, taking into account factors such as the current backlash value, the backlash growth rate, and the equipment's cumulative operating time. For example, H=100 for initial new equipment. As the backlash increases and the growth rate accelerates, the H value gradually decreases. Users can set the system to trigger an alert when the H value falls below a certain level (e.g., 70), prompting the need for maintenance.

[0098] These analysis results can be displayed locally via a human-machine interface or sent to a host computer monitoring system via industrial buses such as CAN, EtherCAT, and RS485, providing direct and quantitative decision-making basis for implementing predictive maintenance (PdM) and optimizing equipment maintenance plans.

[0099] The back gap detection scheme described in this embodiment has high flexibility in physical implementation, but its most advantageous integration method is to embed it as a core functional module into the electrical control system of existing equipment.

[0100] A typical integration approach is as additional firmware functionality to a servo driver. The excitation module can utilize the driver's existing current loop output circuit by superimposing a tiny AC command. The signal processing module can share the driver's ADC sampling resources. The phase analysis module and backlight determination module can be added as new tasks, running in the driver's digital signal processor (DSP) or microcontroller (MCU), executing in parallel with other control tasks (such as position loop, speed loop, and current loop). Finally, backlight data can be uploaded via the driver's built-in communication port (such as EtherCAT or CANopen). This integration method requires no additional external hardware; high-value health monitoring functionality can be added to existing servo products solely through software upgrades.

[0101] Another integration method is as a software function package within a robot joint controller or numerical control (CNC) system. In multi-axis systems, the main controller can centrally manage the backlash detection tasks of each joint or axis, coordinate their calibration and detection timing, and centrally collect, analyze, and display the health status of all drive chains. The system can directly read feedback data from each axis driver and execute phase analysis algorithms, or send instructions to the drivers to complete local detection and report the results.

[0102] This deep integration approach enables the back gap detection technology to be seamlessly integrated into various high-end electromechanical equipment at extremely low marginal cost, truly achieving "ubiquitous intelligent monitoring".

[0103] To ensure the detection system can operate stably and reliably under various complex conditions, this embodiment has refined the excitation module strategy in depth, providing a variety of configurable excitation modes to meet the needs of different application scenarios.

[0104] 1. Single-frequency excitation mode:

[0105] This is the most basic and commonly used mode. The excitation module generates a sinusoidal current of a single frequency. The choice of its frequency f is crucial and needs to be balanced among several constraints:

[0106] 1.1 Avoid interference: Avoid the PWM switching frequency of the motor driver and its main harmonics (such as 5kHz, 10kHz), and keep away from the power grid frequency (50 / 60Hz) and its harmonics.

[0107] 1.2 Balancing Response and Signal-to-Noise Ratio: If the frequency is too low, the system response is slow and susceptible to low-frequency vibration interference; if the frequency is too high, it may be limited by the parasitic inductance of the motor windings and the driver sampling, resulting in severe signal attenuation. Therefore, through theoretical analysis and experimental verification, an optimal single-frequency operating range was determined: 200Hz to 2kHz. In a specific implementation case, 600Hz was selected as the center frequency. At this point, the amplitude of the excitation current was set between 0.1A and 0.8A. This amplitude is sufficient to produce a clear and measurable change in magnetic flux in the motor's magnetic circuit, yet it is far smaller than the current value required for the motor to generate effective torque (typically several amperes to tens of amperes), thus ensuring that the impact on the normal operation of the equipment is negligible.

[0108] 2. Multi-frequency excitation mode:

[0109] To enhance the system's robustness and anti-interference capability in high-noise environments, this embodiment supports simultaneous multi-frequency excitation. The excitation module can generate a superimposed signal of two or three sinusoidal currents at different frequencies and inject it into the motor windings. For example, a preferred three-frequency combination is: f1=523Hz, f2=937Hz, f3=1500Hz. These frequency points are carefully selected, not integer multiples of each other, and distributed within a preferred frequency band. The phase analysis module within the control unit needs to have multi-channel processing capabilities, capable of independently performing phase-locked loop and FFT analysis on each excitation frequency. The advantages of multi-frequency excitation are:

[0110] 2.1 Frequency diversity: If a certain frequency point is severely polluted by sudden noise, the system can still make a reliable judgment based on data from other frequency points.

[0111] 2.2 Cross-validation: The phase jump trend measured at multiple frequency points should be consistent, which provides an additional verification dimension for the determination of back gap events.

[0112] 3. Frequency sweep calibration mode:

[0113] As mentioned earlier, this mode is used for system initialization or periodic calibration. The excitation module outputs a series of test currents at frequencies ranging from 100Hz to 2000Hz, in steps of 10Hz to 50Hz. For each frequency, the system not only evaluates the signal-to-noise ratio but also records the stability of phase tracking.

[0114] Through scanning, the system can:

[0115] 3.1 Draw a complete frequency response spectrum and find the "optimal frequency band" with the highest signal-to-noise ratio and the most stable phase tracking.

[0116] 3.2 Identify and avoid system resonance points to prevent the excitation frequency from coupling with the mechanical structure's resonance frequency, which could lead to signal distortion or amplification.

[0117] 3.3 Compensation for Long-Term Drift: As equipment ages, motor parameters may change, and the optimal operating frequency may shift. Periodic frequency sweeps can reposition the optimal frequency, achieving self-adaptation.

[0118] To clarify the in-depth details of the phase detection and processing algorithm, that is, to elaborate more on the key algorithm details inside the phase analysis module, since these details are the core of achieving high-precision detection, these key algorithm details include the implementation details of the digital phase-locked loop (DPLL) and the implementation details of the fast Fourier transform (FFT) verification channel.

[0119] A digital phase-locked loop (DPLL) typically consists of a phase detector (PD), a loop filter (LF), and a numerically controlled oscillator (NCO) forming a closed loop. In one specific embodiment, an improved digital implementation is employed, using a phase detector, a loop filter, and sampling and delay techniques.

[0120] First, the phase detector uses a multiplier or a phase difference calculation unit based on the CORDIC algorithm to directly calculate the phase error between the excitation signal and the local NCO output signal.

[0121] Then, the loop filter typically employs a proportional-integral (PI) structure. Its parameters (bandwidth, damping coefficient) require careful tuning. Too wide a bandwidth results in poor noise suppression; too narrow a bandwidth leads to slow dynamic response and an inability to track rapid phase transitions. In this embodiment, the DPLL is designed with a wide dynamic bandwidth, enabling it to quickly lock onto the steady-state phase and respond rapidly to and track phase transitions at the instant of backlash occurrence (<1ms). The output of the loop filter directly controls the frequency of the NCO, forcing the NCO's phase to synchronize with the input signal phase.

[0122] Finally, sampling and delay are implemented to ensure the entire DPLL operation runs on a high-speed clock, preferably a clock with a sampling rate of 50-200kHz. From signal input to phase difference output, the total computational delay is strictly controlled to within 20 microseconds. This extremely low latency is a technical guarantee for capturing transient events.

[0123] For the Fast Fourier Transform (FFT) verification channel, which serves as an auxiliary and verification path, its implementation optimization focuses on accuracy and anti-interference capabilities.

[0124] First, windowing and filtering are applied. Before performing an FFT on the induced signal, it is passed through a bandpass filter with its center frequency locked at the current excitation frequency and a bandwidth of ±20 Hz. This effectively filters out out-of-band noise. Simultaneously, a Hanning window is applied to the time-domain data to reduce spectral leakage.

[0125] Next is the phase calculation, which, as described in paragraph six, is derived from the real and imaginary parts of the complex spectrum. To improve accuracy, peak interpolation algorithms (such as the centroid method or quadratic interpolation) can be used to estimate the frequency and phase more precisely, avoiding errors introduced by insufficient frequency resolution (pick fence effect).

[0126] Finally, energy calculation is performed. At the excitation frequency f0, the formula for calculating the signal energy E is: E[f0]=(Re(X[f0])) 2 +(Im(X[f0])) 2 The system continuously monitors the baseline level of E[f0] and calculates the relative rate of change of energy when a mutation event occurs: rate of change of energy = |E after [f0]-E before [f0]∣ / E before [f0]×100%, when this rate of change exceeds a preset threshold (such as 20%), it is considered an effective energy change.

[0127] To further explain in detail how to establish and calibrate the key conversion factor k, which is the bridge for converting electrical signals into mechanical quantities, this section elaborates on three aspects: theoretical model foundation, factory calibration process, and online self-learning and self-adaptation.

[0128] First, there's the theoretical model foundation: a more accurate conversion model not only considers linear proportional relationships but may also include nonlinear terms and operating condition compensations. A more general model expression is: Δθ = k1⋅Δφ + k2⋅(Δφ) 2 +C(T, Im), where: k1 is the linear principal coefficient, directly related to the transmission ratio and the number of pole pairs; k2 is the quadratic term coefficient used to compensate for possible slight nonlinearities; C(T, Im) m This is a compensation term, which may be related to the motor temperature T and the load current I. m This is relevant. In applications requiring high precision, temperature compensation and load compensation are necessary.

[0129] Secondly, the specific steps of the factory calibration process are as follows:

[0130] Step 1: Set up a high-precision calibration platform: Install the transmission mechanism under test (such as a servo motor + reducer) on the calibration platform, and connect its output shaft to an ultra-high precision absolute encoder (such as 26 bits or more) or a laser interferometer.

[0131] Step 2, System Connection and Preheating: Connect this detection system to the motor driver and power it on to preheat to a stable operating temperature.

[0132] The third step is to perform standard reciprocating micro-motion: control the motor to perform a series of precise reciprocating movements with known angular displacements (e.g., ±0.05°, ±0.1°, ±0.2°) near the zero point. An external high-precision instrument records the actual mechanical displacement curve output for each movement, including the "plateau segment" or "hysteresis" caused by backlash.

[0133] Step 4: Synchronous Data Acquisition: This detection system synchronously records all phase abrupt change events and their corresponding Δφ values ​​that occur during each reciprocating motion.

[0134] Step 5: Data Matching and Fitting: The mechanical backlash Δθ_meas measured by external instruments is matched one-to-one with the phase abrupt change Δφ detected by the system. Multiple sets of (Δθ_meas, Δφ) data points are used to perform curve fitting using the least squares method to determine the coefficients k1, k2, etc., in the model. For a simple linear model, the value of k is the slope of the line fitted between the Δθ_meas and Δφ data points.

[0135] Step 6: Coefficient burning: Burn the optimal model coefficients (k values) obtained from the fitting into the non-volatile memory of the detection system.

[0136] Finally, there's online self-learning and self-adaptation: For equipment already in use, the system can perform online self-learning during the initial operation phase or specific maintenance windows. For example, when confirming the equipment is in good new condition, the system performs several standard micro-motions and records the Δφ_baseline at this point. At this time, it can be assumed that Δθ is very small or a known value, thus deriving an initial k value. In subsequent long-term operation, the system can monitor the long-term statistical stability of the k value calculation results, or combine other sensor information (such as increased vibration) to slowly and constrainedly adjust the k value online to adapt to extremely slow parameter drift.

[0137] To more fully illustrate the implementation of the various embodiments of the present invention, in one embodiment, a complete workflow from installation to operation is described below, combined with a specific application scenario: online monitoring of backlash in the sixth axis (wrist joint) of an industrial robot.

[0138] 1. System Integration and Installation: The robot joints utilize an integrated servo motor-driven harmonic reducer. This detection system is integrated into the joint's drive controller as an add-on circuit board and upgraded firmware. The excitation module's output is connected to the motor driver's current feedback loop via a small coupling circuit. No new sensors need to be installed on the mechanical structure.

[0139] 2. Initial Power-On Initialization and Calibration: After the robot is powered on, the joint controller performs the following steps:

[0140] 2.1 Frequency Sweep Self-Calibration: When the joint is under no-load or near zero speed, the frequency sweep program (100-2000Hz) is automatically run. After system analysis, it is determined that the signal-to-noise ratio of this specific joint motor is optimal at two frequency points: 843Hz and 1210Hz, so it is set to dual-frequency excitation mode.

[0141] 2.2. Reference Phase Establishment: At the selected excitation frequency, run DPLL to lock and record the reference phase difference φ0 in steady state.

[0142] 2.3 Model parameter loading: Read the backlash conversion factor k (e.g., k=0.038 ° / °) calibrated at the factory for this model of joint from the memory.

[0143] 3. Normal Operation and Real-time Monitoring: The robot is deployed on the production line to perform handling tasks. During each work cycle, when the wrist joint needs to reverse direction for precise positioning:

[0144] 3.1 Intelligent Trigger: The control system sends a minute position adjustment command (±0.5°), while the motor speed drops to near zero. The backlash detection system recognizes the "small-amplitude micro-motion command" and the "low speed" condition, and automatically opens the backlash sensitive detection window.

[0145] 3.2 Phase Tracking and Abrupt Change Detection: During the window period, the dual-channel phase analysis module operates at high speed. Assume that at a certain commutation instant, the DPLL channel detects a phase difference jump of 0.63° within 1.2ms, with a change rate of 525° / s; simultaneously, the FFT verification channel shows a synchronous phase jump at the 843Hz frequency point, and the signal energy increases by 32%.

[0146] 3.3 Comprehensive Judgment and Calculation: The backlash determination module determines that all the above signals meet the composite conditions, confirming that a backlash event has occurred. Then, it uses the formula Δθ = k × Δφ = 0.038 × 0.63 ≈0.024° to calculate the mechanical backlash angle of the current joint.

[0147] 3.4 Data Recording and Health Assessment: The health management module records this event. Combining data from thousands of cycles over the past week, the module uses linear fitting to find that the backlash is increasing at an average rate of 0.001° per 100 hours of operation. The current health index H has decreased from an initial 100 to 88.

[0148] 4. Maintenance Early Warning: The host computer monitoring software receives real-time health data for each joint. When the health index H of the wrist joint remains below 85, the software automatically generates a predictive maintenance work order, prompting maintenance personnel: "The transmission backlash of robot #A-06's wrist joint is showing a significant increasing trend, and it is expected to reach the alarm threshold of 0.03° after 300 operating hours. It is recommended to check the harmonic reducer during the next planned shutdown." This transforms potential sudden shutdown failures into planned and manageable preventative maintenance activities.

[0149] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0150] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A mechanical gear drive backlash detection system, characterized by, The system comprises: an excitation module configured to inject an alternating excitation current of a preset frequency into the winding of the motor under test; a signal processing module configured to collect an excitation signal corresponding to the excitation current and an induced signal induced by the excitation current in the magnetic field of the motor; a phase analysis module configured to calculate the phase difference between the excitation signal and the induced signal in real time; a backlash determination module configured to determine that the mechanical transmission chain has a backlash when a mutation event of the phase difference is detected, and determine the amount of backlash based on the change in the phase difference.

2. The mechanical gear backed lash detection system of claim 1, wherein: The backlash determination module is configured to start detecting the mutation event of the phase difference when the absolute value of the actual speed of the motor is lower than a first preset threshold and the actual acceleration direction is reversed; and / or start detecting the mutation event of the phase difference when a reciprocating position micro-motion instruction with an amplitude less than a second preset threshold is received.

3. A mechanical geared back lash detection system according to claim 1 or 2, characterized in that: The phase analysis module comprises: a digital phase-locked loop unit configured to perform real-time phase tracking on the excitation signal and the induced signal, and output a first phase difference signal; and a fast Fourier transform analysis unit configured to perform frequency spectrum analysis on the induced signal, extract phase information at the preset frequency, and output a second phase difference signal.

4. The mechanical gear backlash detection system of claim 3, wherein: The backlash determination module is configured to determine that an effective backlash phase mutation event occurs when the following conditions are met simultaneously: (a) the rate of change of the first phase difference signal exceeds a third preset threshold; (b) the duration of the phase mutation state is less than a fourth preset threshold; (c) the change in signal energy at the preset frequency analyzed by the fast Fourier transform analysis unit exceeds a fifth preset threshold.

5. The mechanical gear backed lash detection system of claim 1, wherein, The system further comprises: a calibration module configured to control the excitation module to perform frequency sweep excitation within a preset frequency band during system initialization, and select the optimal preset frequency based on the signal response quality of each frequency point.

6. The mechanical gear backed lash detection system of claim 1, wherein: The backlash determination module converts the change in the phase difference into a mechanical backlash angle or a linear displacement amount according to a pre-stored phase difference-backlash amount conversion model.

7. The mechanical gear backed lash detection system of claim 1, wherein, The system further comprises: a health management module configured to record historical backlash amount data, perform trend analysis and fitting, and generate a device health index or a wear warning signal based on the change trend of the backlash amount.

8. The mechanical gear backed lash detection system of any of claims 1-7, wherein: The system is integrated into a control unit of a servo drive, a robot joint controller, or a numerical control system.

9. A mechanical gear drive backlash detection method, characterized by, The method comprises the following steps: injecting an alternating excitation current of a preset frequency into the winding of the motor under test; collecting an excitation signal corresponding to the excitation current and an induced signal induced by the excitation current in the magnetic field of the motor; calculating the phase difference between the excitation signal and the induced signal in real time; monitoring whether a mutation event of the phase difference occurs according to a preset feature; if so, determining that the mechanical transmission chain has a backlash, and determining the amount of backlash based on the change in the phase difference.

10. The mechanical gear backlash detection method of claim 9, wherein: In the monitoring step, high-sensitivity monitoring of the phase difference mutation event is performed only when it is detected that the absolute value of the actual rotation speed of the motor is lower than a first preset threshold and the actual acceleration direction is reversed, and / or when the system receives a reciprocating position fine motion instruction with an amplitude smaller than a second preset threshold.

Images