Adaptability judgment system of ultrasonic transducer based on transient monitoring

Abstract

The invention provides an ultrasonic transducer suitability judgment system based on transient monitoring, and belongs to the technical field of ultrasonic waves. The system comprises a data acquisition module used for acquiring key transient parameters and steady-state parameters of operation of the ultrasonic transducer; the dynamic threshold value calculation module is used for calculating an initial current threshold value based on the collected key transient parameters and steady state parameters in combination with preset weight coefficients corresponding to the parameters and a preset basic current threshold value; and the intelligent judgment and feedback module is used for comparing the collected working current with an initial current threshold value and determining the ultrasonic adaptation condition. According to the method, the energy consumption state of the transducer is directly reflected in real time through transient parameter judgment and the collected working current value, sudden load change can immediately cause current fluctuation, temperature accumulation does not need to be waited, millisecond-level response can be achieved, load abnormity can be early warned in advance, and equipment damage or machining failure caused by lagging judgment is avoided.

Description

Technical Field

[0001] The present invention relates to the field of ultrasonic technology, and in particular to an ultrasonic transducer adaptability judgment system based on transient monitoring. Background Art

[0002] As a core component in ultrasonic machining and testing, the performance of ultrasonic transducers is closely related to the load state (such as whether they are carrying a workpiece and the load level) and the load medium (such as metal, plastic, ceramic, etc.). In practical applications, the transducer must dynamically adjust operating parameters (such as current, frequency, and amplitude) based on the current operating conditions to ensure stable operation and avoid equipment damage or processing failures caused by adaptation anomalies. Therefore, adaptability judgment (i.e., determining whether the transducer's current operating state matches the load requirements) is a key technical link in ultrasonic transducer control systems.

[0003] Currently, existing methods for determining ultrasonic transducer compatibility primarily rely on temperature monitoring: using temperature sensors to monitor temperature changes in the transducer body or key components (such as piezoelectric ceramics) to indirectly infer their load status. This method is based on the theory that when the transducer is overloaded (e.g., with a high-impedance medium or a heavy workpiece) or underloaded (e.g., unloaded or lightly loaded), energy loss or increased mechanical vibration can lead to abnormal temperature increases or fluctuations. However, existing temperature monitoring methods for determining compatibility suffer from a lag. Temperature changes provide "feedback" rather than a "real-time representation" of transducer energy loss. It takes a certain amount of time (typically seconds or even longer) for the transducer to experience a significant temperature increase after a sudden load change (e.g., contact with a high-impedance ceramic). During this time, overload may accelerate the aging of internal components (such as the piezoelectric ceramics), or underload may lead to underutilization of vibration energy, resulting in hidden damage to the equipment or reduced processing efficiency. Summary of the Invention

[0004] The present invention provides an ultrasonic transducer adaptability judgment system based on transient monitoring, which is used to solve the problem of hysteresis in the prior art method of judging adaptability through temperature detection.

[0005] In order to achieve the above-mentioned purpose, an embodiment of the present invention provides, on the one hand, an adaptability judgment system for an ultrasonic transducer based on transient monitoring, wherein the adaptability judgment system includes: a data acquisition module for collecting key transient parameters and steady-state parameters of the ultrasonic transducer operation; a dynamic threshold calculation module for calculating the initial current threshold based on the collected key transient parameters and steady-state parameters, combined with the preset weight coefficients corresponding to each parameter and the preset basic current threshold; an intelligent judgment and feedback module for comparing the collected working current with the initial current threshold to determine the ultrasonic adaptation situation.

[0006] Optionally, the data acquisition module includes: a current sensor installed in the power supply circuit between the ultrasonic transducer and the power supply, for collecting the working current of the ultrasonic transducer; a lock-in amplifier connected to the vibration output surface of the ultrasonic transducer, for extracting the working frequency of the ultrasonic transducer; The laser displacement sensor is installed vertically at the vibration end of the ultrasonic transducer to measure the vibration amplitude of the ultrasonic transducer; the load status sensor is installed at the contact surface between the ultrasonic transducer and the load to detect the load contact pressure; the acoustic impedance detection module is installed near the contact interface between the ultrasonic transducer and the workpiece to identify the load medium.

[0007] Optionally, the adaptability judgment system further includes a data preprocessing module for adaptively filtering the current signal to remove high-frequency noise; and discretizing the load state and the load medium; The frequency and amplitude signals are smoothed using a sliding window to eliminate high-frequency noise. Based on the current signal sampling time, the frequency, amplitude, load status, and load medium data are interpolated and aligned, and standardized data packets are output.

[0008] Optionally, the calculating of the initial current threshold based on the collected key transient parameters and steady-state parameters, in combination with preset weight coefficients corresponding to the parameters and a preset basic current threshold, includes: assigning weights to the four parameters of frequency, amplitude, load state, and load medium according to the load state and the load medium; The current threshold offset caused by the frequency is determined based on the difference between the current frequency and the preset standard frequency and the corresponding weight; the current threshold offset caused by the amplitude is determined based on the difference between the current amplitude and the preset standard amplitude and the corresponding weight; the current threshold increase caused by the load state is determined based on the load state and the corresponding weight; the current threshold increase caused by the load medium is determined based on the load medium and the corresponding weight; each offset and increase is superimposed on the preset basic current threshold to obtain the initial current threshold.

[0009] Optionally, the adaptability judgment system further includes an inner loop response control module, and the inner loop response control module is used to fine-tune the initial current threshold through a sliding window and a PID controller.

[0010] Optionally, the initial current threshold is fine-tuned through a sliding window and a PID controller, including: taking a sliding window of length Δt with the current time t as the center, and calculating the current mean in the window; calculating the difference between the current measured current and the current mean in the window; calculating the proportional adjustment amount, integral adjustment amount and differential adjustment amount of the PID controller based on the calculated difference; and adjusting the initial current threshold through the PID controller based on the proportional adjustment amount, the integral adjustment amount and the differential adjustment amount to obtain a current adjustment threshold.

[0011] Optionally, the fine-tuning of the initial current threshold by using a sliding window and a PID controller includes: calculating the short-term fluctuation rate of the frequency and amplitude in real time; and adjusting the proportional gain coefficient of the PID controller according to the fluctuation rate.

[0012] Optionally, the adaptability judgment system further includes an outer loop stability optimization control module for optimizing the weight coefficients of various parameters by a gradient descent method.

[0013] Optionally, the intelligent judgment and feedback module is further configured to output an adaptation judgment result, and feed the adaptation judgment result back to the outer-loop stability optimization control module for continuous optimization. Optionally, the optimization of the weight coefficients of each parameter by the gradient descent method includes: calculating a comprehensive loss function based on historical misjudgment records and current threshold change records; calculating the partial derivative of the loss function with respect to the weight coefficient corresponding to each key transient parameter and steady-state parameter; synthesizing the partial derivative array of each parameter into a gradient vector; and adjusting the weight coefficient of each parameter according to the gradient direction.

[0014] The present invention provides an ultrasonic transducer adaptability judgment system based on transient monitoring. The system directly judges through transient parameters, and the collected working current value reflects the energy consumption status of the transducer in real time. A sudden change in load will immediately cause current fluctuations without waiting for temperature accumulation. It can achieve millisecond-level response, provide early warning of load anomalies, and avoid equipment damage or processing failure due to delayed judgment.

[0015] The present invention solves the core defects of existing temperature detection methods such as response lag, insufficient accuracy, weak anti-interference, poor dynamic adaptability and lack of long-term robustness through direct current judgment (real-time load characterization) and multi-parameter fusion (comprehensive coverage of operating conditions), combined with dynamic threshold calculation, real-time fine-tuning of the inner loop and continuous optimization of the outer loop. It significantly improves the working reliability and efficiency of ultrasonic transducers under different load states, medium characteristics and operating conditions, and has significant practical value and market competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS In order to more clearly illustrate the technical solutions of the present invention or the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings: Figure 1 Schematic diagram of the structure of the ultrasonic transducer adaptability judgment system based on transient monitoring provided by the present invention; Figure 2 Schematic diagram of the workflow of the dynamic threshold calculation module and the inner loop response control module provided in an embodiment of the present invention; Figure 3 3 is a schematic diagram of the workflow of the intelligent judgment and feedback module provided in an embodiment of the present invention.

[0016] Description of Reference Numerals 10. Data acquisition module; 20. Dynamic threshold calculation module; 30. Intelligent judgment and feedback module. DETAILED DESCRIPTION

[0017] The following describes the specific implementation of the embodiment of the present invention in detail with reference to the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the embodiment of the present invention and is not used to limit the embodiment of the present invention.

[0018] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations. In the embodiments of this application, certain software, components, models, and other existing solutions in the industry may be mentioned. These should be considered as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use such solutions.

[0019] As mentioned above, existing methods of determining compatibility through temperature detection suffer from lag. Temperature changes are "result feedback" of transducer energy loss, not "real-time characterization." It takes a certain amount of time (typically seconds or even longer) for a transducer to experience a significant temperature increase from a sudden load change (such as contact with a high-impedance ceramic). During this time, the transducer may have experienced accelerated aging of internal components (such as piezoelectric ceramics) due to overload, or underutilization of vibration energy due to underutilization, resulting in hidden damage to the equipment or reduced processing efficiency.

[0020] Therefore, to address this problem, the present invention provides an ultrasonic transducer adaptability judgment system based on transient monitoring, which directly judges through transient parameters, and the collected working current value reflects the energy consumption status of the transducer in real time - sudden changes in load will immediately cause current fluctuations, without waiting for temperature accumulation, and can achieve millisecond-level response, early warning of load abnormalities, and avoid equipment damage or processing failure due to delayed judgment.

[0021] Figure 1 This is a schematic diagram of the structure of the ultrasonic transducer adaptability judgment system based on transient monitoring provided by an embodiment of the present invention. Please refer to Figure 1 The adaptability judgment system may include: a data acquisition module 10 , a dynamic threshold calculation module 20 and an intelligent judgment and feedback module 30 .

[0022] The data acquisition module 10 is used to collect key transient parameters and steady-state parameters of the ultrasonic transducer operation; the dynamic threshold calculation module 20 is used to calculate the initial current threshold based on the collected key transient parameters and steady-state parameters, combined with the preset weight coefficients corresponding to each parameter and the preset basic current threshold; the intelligent judgment and feedback module 30 is used to compare the collected working current with the initial current threshold to determine the ultrasonic adaptation status.

[0023] The data acquisition module collects key transient and steady-state parameters of the ultrasonic transducer in real time (transient parameters such as working current, vibration frequency, and amplitude; steady-state parameters such as load pressure and load medium characteristics), providing the system with original working condition data; the dynamic threshold calculation module generates the initial current threshold based on the collected parameters, combined with the preset weight coefficient (such as the differential impact of load status and medium type on the threshold) and the basic current threshold, through multi-parameter weighted calculation, and dynamically adapts to load mutations and medium switching scenarios; the intelligent judgment and feedback module judges whether the current working condition of the ultrasonic transducer is adapted (such as "normal adaptation" or "abnormal adaptation") by comparing the working current with the dynamically generated initial current threshold in real time.

[0024] Preferably, the data acquisition module includes: a current sensor, installed in the power supply circuit between the ultrasonic transducer and the power supply, for collecting the working current of the ultrasonic transducer; a phase-locked amplifier, connected to the vibration output surface of the ultrasonic transducer, for extracting the working frequency of the ultrasonic transducer; a laser displacement sensor, installed vertically at the vibration end of the ultrasonic transducer, for measuring the vibration amplitude of the ultrasonic transducer; a load status sensor, installed at the contact surface between the ultrasonic transducer and the load, for detecting the load contact pressure; and an acoustic impedance detection module, installed near the interface between the ultrasonic transducer and the workpiece, for identifying the load medium.

[0025] The data acquisition module uses a coordinated deployment of multiple sensors to acquire real-time data on the ultrasonic transducer's current, frequency, amplitude, load pressure, and dielectric acoustic impedance, providing global data support for dynamic threshold calibration. The functions and installation locations of each component are as follows: The current sensor is connected in series with the ultrasonic transducer's power supply circuit, providing high-precision transient operating current measurements (±0.5% accuracy), reflecting load energy consumption. The lock-in amplifier is connected to the vibrating end face of the ultrasonic transducer, extracting the fundamental frequency signal (±0.1Hz accuracy) and demodulating the vibration frequency. The laser displacement sensor is mounted perpendicular to the vibrating end face, providing non-contact measurement of the end face's transient amplitude (±1μm accuracy), minimizing contact interference. The load status sensor is located at the interface between the ultrasonic transducer and the workpiece, determining the load status through pressure detection (threshold: no load <0.5N, full load >5N). The acoustic impedance detection module is located close to the workpiece interface, calculating the acoustic impedance (Z=ρv) from reflected sound waves, and identifying the material type (e.g., steel, aluminum, or plastic).

[0026] Preferably, the adaptability judgment system also includes a data preprocessing module, which is used to adaptively filter the current signal to remove high-frequency noise; discretize the load status and load medium; perform sliding window smoothing on the frequency and amplitude signals to eliminate high-frequency noise; interpolate and align the frequency, amplitude, load status and load medium data based on the current signal sampling time, and output a standardized data packet.

[0027] The data preprocessing module performs multi-level cleaning and alignment on the raw sensor data to ensure that the data input to the dynamic threshold calculation module has a high signal-to-noise ratio, a unified time base and a standardized format. The specific process includes: 1) Adaptive filtering: Dynamic low-pass filtering is applied to the current signal (the cutoff frequency is adaptively adjusted according to the grid fluctuations and mechanical vibration characteristics) to filter out high-frequency noise (such as power supply harmonic interference); 2) Discrete encoding: Continuous state parameters (such as the pressure value under load and the acoustic impedance of the load medium) are mapped into discrete categories (for example, pressure is divided into three levels: no load / light load / heavy load, and the medium matching is indexed as steel / aluminum / plastic); 3) Sliding window smoothing: Sliding average filtering (window length 10ms) is applied to the frequency and amplitude signals to suppress high-frequency jitter (such as amplitude glitches caused by mechanical resonance of the transducer); 4) Time alignment interpolation: Based on the current signal sampling time (100kHz), asynchronous sampling data such as frequency, amplitude, and load status (such as 50kHz for a phase-locked amplifier and 200kHz for a pressure sensor) are linearly interpolated to unify the timestamp; 5) Standardized packaging: The processed data is packaged according to a unified protocol (such as timestamp + current value + encoded load status + interpolated frequency / amplitude) and output to the downstream dynamic threshold calculation module.

[0028] For example, take the ultrasonic welding of aluminum parts as an example: the original current signal contains high-frequency noise (100kHz sampling), and after adaptive filtering, the 20kHz fundamental frequency component is retained; the loaded pressure sensor outputs a fluctuating signal (0.3-5.2N), which is discretized and encoded as "heavy load" (L=2); the vibration frequency (20.5kHz) and amplitude (12μm) data are smoothed by a sliding window, and the frequency is stabilized to 20.4±0.1kHz, and the amplitude fluctuation is reduced by 60%; because the current signal has the highest sampling rate, the frequency and pressure data are aligned to the 100kHz time grid through interpolation to eliminate timing misalignment.

[0029] The present invention combines adaptive filtering with sliding window smoothing to suppress high-frequency noise while retaining key operating characteristics (such as amplitude attenuation trends); unifies the time base and discrete coding to eliminate timing confusion caused by asynchronous sampling of multiple sensors; standardizes data packets to reduce the compatibility processing burden of downstream modules and improve real-time performance; and avoids misjudgments caused by sensor noise or sampling rate differences (such as "false alarms" caused by instantaneous jitter of the pressure signal) through interpolation alignment and noise filtering.

[0030] Please refer to Figure 2 Preferably, based on the collected key transient parameters and steady-state parameters, combined with the preset weight coefficients corresponding to each parameter and the preset basic current threshold, the calculation of the initial current threshold may include S301-S306: S301, assigning weights to the four parameters of frequency, amplitude, load state and load medium according to the load state and load medium; S302, determining the current threshold offset caused by frequency according to the difference between the current frequency and the preset standard frequency and the corresponding weight; S303, determining the current threshold offset caused by amplitude according to the difference between the current amplitude and the preset standard amplitude and the corresponding weight; S304, determining the current threshold increase caused by the load state according to the load state and the corresponding weight; S305, determining the current threshold increase caused by the load medium according to the load medium and the corresponding weight; S306 superimposing each offset and increase on the preset basic current threshold to obtain the initial current threshold.

[0031] S301 dynamically adjusts the weight coefficients of the four major parameters of frequency (f), amplitude (A), load state (L), and load medium (M) according to the load state (L) and load medium (M). , , , ), reflecting the core influence of various parameters on the current threshold under different working conditions. The specific rules are: 1) Load state weight : The weight is 0 when there is no load (L=0) (the load has no direct effect on the threshold); the weight increases to 0.3 when there is a light load (L=1) (some energy loss needs to be compensated); the weight increases to 0.6 when there is a heavy load (L=2) (the load significantly increases the energy demand). 2) Load medium weight km: Dynamically assigned according to the acoustic impedance of the material (such as steel =0.4, aluminum =0.3, ceramic =0.5), high impedance media (such as ceramics) require a higher threshold to compensate for energy loss due to their high energy reflectivity. 3) Frequency Weight With amplitude weight : The default weights are 0.8 and 0.5 respectively (frequency has a greater impact on the threshold), but if the outer loop optimization module detects long-term changes in working conditions (such as frequent amplitude attenuation in aluminum welding), the weights will be dynamically adjusted (for example, increasing to 0.6 to enhance amplitude compensation).

[0032] For example, when welding aluminum parts (L=2, M=aluminum), =0.6, =0.3, if the current frequency deviates from the reference value (20kHz→21kHz), the frequency weight = 0.8 dominates the threshold correction, while the amplitude weight =0.5 suppresses minor fluctuations.

[0033] The present invention uses working condition-driven weight distribution to ensure that the initial current threshold is anchored to historical experience (such as the baseline threshold for unloaded steel parts) and can also adapt to load and medium characteristics, avoiding misjudgments caused by fixed weights (such as insufficient threshold in high-impedance media).

[0034] S302 calculates the difference between the current operating frequency and the preset standard frequency ( ), combined with the frequency weight coefficient , quantify the direct impact of frequency fluctuation on current threshold. The specific logic is: 1) Frequency difference calculation: real-time monitoring of the operating frequency of the ultrasonic transducer , with the preset reference frequency (such as 20kHz) and calculate the difference ; 2) Linear offset mapping: multiply the frequency difference by the weight (The default value is 0.8), and the threshold offset caused by the frequency is obtained For example, if the current =21kHz( =+1kHz), then =0.8A, indicating that the threshold needs to be increased to compensate for high-frequency energy loss.

[0035] For example, in ultrasonic cleaning of plastic parts, if the transducer frequency drifts from 20kHz to 19.5kHz (Δf = −0.5kHz) due to load changes, then ΔIf = 0.8 × (−0.5) = −0.4A. The threshold automatically adjusts downward to match the low-frequency energy demand, avoiding false positives. This invention uses a linear binding between frequency and threshold to quickly respond to fluctuations in energy transfer efficiency caused by load changes, ensuring that the threshold always matches the actual operating conditions.

[0036] S303 is based on the difference between the current vibration amplitude and the preset standard amplitude ( and amplitude weight coefficient , quantify the direct impact of amplitude fluctuation on current threshold, and realize dynamic compensation for energy transfer efficiency changes caused by amplitude attenuation or overshoot. The specific process is: 1) Amplitude difference calculation: take the standard working condition amplitude as the (such as 10μm) as a benchmark to calculate the real-time amplitude Deviation For example, if the current amplitude =8μm, then = −2μm. Multiply the amplitude deviation by the weighting coefficient (As preset = 0.5), and the current threshold offset caused by the amplitude is obtained .For example, =0.5×(−2)=−1A, indicating that the threshold needs to be lowered to match the energy demand at low amplitudes.

[0037] The present invention converts amplitude deviation into a quantitative basis for threshold correction, ensuring that the threshold fits the actual energy output capacity in real time and suppressing misjudgments caused by amplitude fluctuations (such as mechanical wear or sudden load changes) (such as the false alarm of "current abnormality" when the amplitude is insufficient).

[0038] S304 determines the load status (L) by grade and the corresponding weight coefficient , quantify the rigidity of the load level on the current threshold, and ensure that the threshold is dynamically increased under high load conditions to compensate for energy loss. The specific logic is: 1) Load state classification: According to the pressure sensor signal, the load state is divided into three levels: No load (L=0): contact force F<0.5N, weight =0 (no additional energy compensation required); light load (L=1): 0.5N≤F≤5N, weight =0.3 (need to slightly increase the threshold to cover contact friction loss); heavy load (L=2): F>5N, weight =0.6 (the load significantly increases the energy demand, and the threshold needs to be greatly increased). 2) Threshold increment calculation: The load state weight Directly mapped to current threshold increment For example, when overloading (L=2) =0.6×10A=6A, indicating that the threshold value needs to be increased from the reference value =10A is increased to 16A to match the high load energy demand.

[0039] For example, when ultrasonically welding aluminum parts, if the pressure sensor detects a contact force F = 8N (heavy load state L = 2), the threshold increment =0.6×10A=6A, initial threshold On the basis of frequency and amplitude correction, 6A is further increased to avoid misjudgment of insufficient current due to excessive load.

[0040] The present invention ensures that the threshold value has the ability to respond quickly when the physical load changes by binding the load state classification with the rigidity weight, eliminating the risk of threshold misadjustment caused by contact force fluctuations (such as incomplete fit or overpressure of the workpiece).

[0041] S305 uses the acoustic impedance characteristics of the load medium and the corresponding weight coefficient , quantify the rigid requirements of material types on current thresholds, and ensure that the threshold is dynamically increased in high acoustic impedance media (such as ceramics) to compensate for energy reflection loss. The specific logic is: 1) Medium classification and weight binding: Based on the medium type identified by the acoustic impedance detection module (such as steel Z=45×10 6 kg / (m 2 s), aluminum Z = 17 × 10 6 、Ceramic Z=35×10 6 ), mapping preset weight km (steel =0.4, aluminum =0.3, ceramic =0.5), the weight value is positively correlated with the acoustic impedance, reflecting the characteristic that high impedance materials require higher energy input. 2) Threshold increment calculation: The weight coefficient km is directly superimposed on the current threshold increment For example, ceramic dielectrics ( = 0.5), the threshold increment =0.5×10A=5A, indicating that the reference value =10A is increased to 15A to overcome high impedance energy reflection.

[0042] For example, when ultrasonically welding ceramic parts, the acoustic impedance detection module confirms that the load is ceramic (Z=35×10 6 ),trigger =0.5, threshold increment =5A. After corrections to the frequency, amplitude, and load status, the final threshold adapts to the high energy requirements of the ceramic, avoiding false alarms of insufficient current due to high dielectric reflectivity.

[0043] The present invention accurately compensates for differences in material properties through the strong correlation between the acoustic impedance of the medium and the weight, ensures automatic adaptation of the threshold when processing heterogeneous workpieces, and improves the system's cross-scenario generalization capabilities.

[0044] S306 integrates the frequency, amplitude, load status and the offset and increment of the load medium and adds them to the preset basic current threshold , generate the initial current threshold adapted to the current working conditions The specific logic is: 1) Incremental superposition formula: = + + + + .in, (frequency offset), (Amplitude offset), (load increment), (Medium increment) is calculated by S302-S305 respectively.

[0045] For example, when ultrasonic welding ceramic parts, if the basic threshold =10A, frequency offset =+1.2A (frequency increased to 21kHz), amplitude attenuation =−0.8A (amplitude reduced to 8μm), load increment =6A (heavy duty), medium increment =5A (ceramic high impedance), then the initial threshold =10+1.2−0.8+6+5=21.4A.

[0046] The present invention uses multi-source incremental superposition to uniformly map dispersed parameter fluctuations (such as frequency drift, amplitude attenuation, load change, and medium difference) into comprehensive corrections to the current threshold, ensuring that the initial threshold covers both historical experience ( ), and dynamically adapt to the current complex working conditions, providing accurate initial values ​​for subsequent nonlinear corrections.

[0047] Please refer to Figure 2 Preferably, the adaptability judgment system further includes an inner loop response control module, which is used to fine-tune the initial current threshold through a sliding window and a PID controller.

[0048] Please refer to Figure 2 , preferably, the initial current threshold is fine-tuned by a sliding window and a PID controller, including S401-S404: S401, taking the current time t as the center, taking the length The sliding window of the current is used to calculate the current mean value in the window; S402, the difference between the current measured current and the current mean value in the window is calculated; S403, the proportional adjustment amount, the integral adjustment amount and the differential adjustment amount of the PID controller are calculated based on the calculated difference; S404, the initial current threshold is adjusted by the PID controller based on the proportional adjustment amount, the integral adjustment amount and the differential adjustment amount to obtain the current adjustment threshold.

[0049] S401 uses a dynamic sliding window to count the real-time current signal, taking the current time t as the center and intercepting the length of (e.g. 10ms) window, calculate the current mean within the window , eliminate high-frequency noise (such as power supply harmonics or transient jitter caused by mechanical vibration). For example, if the current sampling values ​​in the window are 12.1A, 12.3A, and 11.9A, then =12.1A. This average value is used as the input reference of the PID controller for subsequent fine-tuning of the initial current threshold , ensuring that the system responds quickly to real current fluctuations (such as load mutations) while avoiding threshold misadjustment due to noise interference. The time step size is one slide) which balances the response speed and noise suppression capability, ensuring real-time performance and stability.

[0050] S402 compares the current measured current with the sliding window average to quantify the instantaneous deviation between the actual current and the historical statistical benchmark, providing a dynamic correction basis for the PID controller. The specific process is as follows: 1) Data acquisition: Real-time acquisition of the instantaneous current value output by the current sensor (e.g. 12.5A); 2) Average call: read the sliding window current average calculated in step S401 (such as 12.1A); 3) Deviation calculation: calculate the difference between the two ,For example: =12.5A−12.1A=+0.4A. This positive value indicates that the current current is higher than the historical average, which may be caused by a sudden load change or medium reflection fluctuation.

[0051] The present invention uses real-time deviation detection to quickly identify current anomalies (such as sudden current increases or decreases), provide accurate error signals for the PID controller, ensure that the threshold dynamically adapts to the actual working conditions (such as compensating for current fluctuations caused by sudden load changes), and avoid misjudgment due to delayed response.

[0052] S403 uses proportional, integral, and differential channels to calculate the current deviation in real time and convert it into an accurate threshold adjustment to achieve dynamic error compensation. The specific logic is as follows: 1) Proportional adjustment: Based on the current deviation , directly multiply by the proportional coefficient like =0.8), quickly respond to transient errors. For example, When the proportional term contributes 0.8×0.4=+0.32A. 2) Integral adjustment amount: accumulated historical deviation (For example, the cumulative deviation in the past 1 second is +0.2A), multiplied by the integral coefficient (like =0.1), eliminating steady-state error. The integral term contributes 0.1×0.2=+0.02A. 3) Differential adjustment: Based on the rate of change of the deviation (For example, the current deviation growth rate is +0.1A / ms), multiplied by the differential coefficient (like = 0.05), predicting future trends and suppressing overshoot. The differential term contributes 0.05 × 0.1 = +0.005A.

[0053] The present invention uses the proportional term to quickly respond to deviations, the integral term to eliminate long-term accumulated errors, and the differential term to suppress fluctuation trends. The three work together to ensure that the threshold dynamically adapts to current mutations (such as load mutations or medium reflection fluctuations) while avoiding system oscillations caused by overshoot.

[0054] S404 uses the PID controller to perform comprehensive calculations on the proportional, integral, and differential adjustment amounts, and adds the correction amount to the initial current threshold. , generating the final dynamically adjusted current threshold The specific process is as follows: 1) Adjustment synthesis: The proportional term calculated in S403 , integral item , differential terms Add them together to get the total PID adjustment: For example, if (proportional term), (Integral Item), (differential term), then the total adjustment 2) Dynamic update of threshold: Apply the total adjustment amount to the initial threshold , generating the inner loop control output: For example, if the initial threshold , then the adjusted threshold .

[0055] For example, when ultrasonically welding aluminum parts, if the initial threshold , PID total adjustment , then the adjusted threshold At this time, if the actual current , then the error e(t)=21.8−21.745=+0.055A, triggering the next round of PID fine-tuning.

[0056] The present invention quickly responds to current deviations through PID closed-loop control, ensuring that the threshold adapts to the actual working conditions (such as sudden load changes or medium reflection fluctuations) in real time. At the same time, through dynamic parameter optimization and anti-overmodulation protection, it balances response speed and stability to avoid misjudgment of "adaptation anomalies".

[0057] Preferably, fine-tuning the initial current threshold value through the sliding window and the PID controller includes: calculating the short-term fluctuation rate of the frequency and amplitude in real time; and adjusting the proportional gain coefficient of the PID controller according to the fluctuation rate.

[0058] The present invention dynamically adjusts the proportional gain coefficient of the PID controller by real-time monitoring of the short-term fluctuation rate of frequency and amplitude. , achieving agility and robustness of threshold fine-tuning. The specific process is: 1) Volatility calculation: Frequency volatility: frequency signal within a sliding window (such as 10ms) Standard deviation A measure of short-term frequency jitter (e.g. ); Amplitude volatility: Calculate the amplitude signal in the same way ) (like ). 2) Dynamic optimization of PID parameters: High fluctuation scenario: If or , indicating that the system is disturbed by high-frequency noise or mechanical vibration, and the (For example, increase from 0.8 to 1.2), accelerate the threshold response to suppress overshoot; low fluctuation scenario: if and , indicating that the working condition is stable and the (For example, reduce it from 0.8 to 0.5) to avoid misadjustment due to oversensitivity.

[0059] For example, when ultrasonic cleaning plastic parts, if the frequency fluctuation rate (high noise), the system automatically Increasing from 0.8 to 1.2 enables PID to compensate for frequency drift faster and avoids misjudgment of current anomalies caused by threshold hysteresis.

[0060] The present invention dynamically adjusts the It not only improves the system's adaptability to dynamic working conditions (such as sudden load changes and medium reflection fluctuations), but also suppresses erroneous operations caused by noise interference, and achieves balanced control of "fast response steady state, anti-noise and non-oscillation".

[0061] Preferably, the adaptability judgment system further includes an outer loop stability optimization control module for optimizing the weight coefficients of various parameters by a gradient descent method.

[0062] Preferably, the optimization of the weight coefficients of each parameter by the gradient descent method includes S501-S504: S501, calculating the comprehensive loss function based on the historical misjudgment records and the current threshold change records; S502, calculating the partial derivative of the loss function with respect to the weight coefficient corresponding to each key transient parameter and steady-state parameter; S503, synthesizing the partial derivative array of each parameter into a gradient vector; S504, adjusting the weight coefficient of each parameter according to the gradient direction.

[0063] S501 constructs a loss function by quantifying the number of historical misjudgments and the threshold fluctuation amplitude to drive the outer loop optimization module to adjust the weight coefficient ( , , , ) and baseline thresholds The specific formula is: . Misjudgment penalty ( ): Count the number of false positives or omissions of "adaptation anomaly" in history (such as misjudging normal working conditions as abnormalities) and assign high weights (such as α = 1), directly reflects the impact of misjudgment on system reliability; the threshold fluctuation penalty term ( ): Calculate the sum of the absolute values ​​of the threshold changes at adjacent moments and assign the second highest weight β (such as β = 0.5) to suppress energy waste and false alarms caused by frequent threshold fluctuations.

[0064] S502 provides parameter adjustment direction for the outer loop optimization module by quantifying the sensitivity of the loss function to each weight coefficient. The specific steps are: loss function explicitly associated with weights: the comprehensive loss function Loss is not only related to the number of misjudgments and threshold fluctuations, but also depends on the weight coefficients , , , (Because weights directly affect threshold calculations, which in turn affect misjudgments and fluctuations.) For example, weight deviations may cause the threshold to deviate from actual needs, increase the number of misjudgments, or exacerbate threshold fluctuations. Partial derivative calculation: Calculate the partial derivative for each weight coefficient. ( ), reflects the change in the loss function when the weight changes by 1 unit. For example: , indicating that increasing the weight of ceramic dielectric This will exacerbate misjudgment (such as overestimating the threshold required for ceramic parts) and needs to be reduced. ;like , indicating that the overload weight is increased Can reduce misjudgment (such as compensating for high load energy demand), need to increase .

[0065] S503 synthesizes the gradient vector by combining the partial derivative array of the loss function with respect to each parameter, clarifies the mathematical direction and magnitude of the parameter adjustment, and provides a basis for parameter update for the outer loop optimization module. The specific process is: the four key parameters ( , , , ) corresponding to the partial derivative Arrange them in order to form a gradient vector: .

[0066] S504 adjusts the weight coefficients of key parameters along the negative gradient direction to minimize the comprehensive loss function and achieve adaptive optimization of parameters. The specific logic is: 1) Gradient direction analysis: Gradient vector Each element in ( ) directly indicates the parameter adjustment direction. For example: , indicating that increasing the weight of ceramic dielectric Can reduce losses; if , then the frequency weight needs to be reduced To suppress misjudgment. 2) Parameter update formula: along the negative gradient direction according to the learning rate Adjust weights: For example, if the current and , learning rate , then after updating .

[0067] The gradient descent method of the present invention uses a mathematically driven parameter optimization mechanism to enable the system to have rapid response capability, noise robustness and long-term adaptability under complex dynamic conditions.

[0068] Please refer to Figure 3 Preferably, the intelligent judgment and feedback module is also used to output the adaptation judgment result, and feed the adaptation judgment result back to the outer loop stability optimization control module for continuous optimization. The present invention generates an adaptation judgment result (such as "normal adaptation" or "abnormal adaptation") by comparing the current threshold adjusted by the inner loop with the actual current value in real time, and feeds the result and relevant working condition data (load status, load medium, frequency, amplitude, etc.) to the outer loop optimization module to drive it to continuously optimize the weight coefficient and the reference threshold. For example, if the aluminum welding adaptation is judged to be normal for 5 consecutive times, the outer loop module will reduce the load medium weight. The learning rate is set to stabilize the current parameter combination; if a misjudgment occurs (such as the actual current does not exceed the threshold but is judged to be abnormal), the outer loop is triggered to recalibrate the frequency weight Or load state weight , using gradient descent to correct historical deviations and improve the system's long-term robustness. A closed-loop feedback mechanism ensures the algorithm dynamically adapts to long-term operating conditions such as changes in material properties (such as differences in ceramic batches) and equipment aging, achieving intelligent adaptation through "self-learning and self-optimization."

[0069] In short, the above description is only a preferred embodiment of the technical solution of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention.

Claims

1. A system for determining the suitability of an ultrasonic transducer based on transient monitoring, characterized in that: The adaptability judgment system includes: Data acquisition module, used to collect key transient parameters and steady-state parameters of ultrasonic transducer operation; A dynamic threshold calculation module is used to calculate the initial current threshold based on the collected key transient parameters and steady-state parameters, combined with the preset weight coefficients corresponding to each parameter and the preset basic current threshold; The intelligent judgment and feedback module is used to compare the collected working current with the initial current threshold to determine the ultrasonic adaptation situation.

2. The adaptability judgment system according to claim 1, characterized in that: The data acquisition module includes: A current sensor is installed in the power supply circuit between the ultrasonic transducer and the power supply, and is used to collect the working current of the ultrasonic transducer; a lock-in amplifier connected to the vibration output surface of the ultrasonic transducer and used to extract the operating frequency of the ultrasonic transducer; The laser displacement sensor is installed vertically on the vibration end of the ultrasonic transducer to measure the vibration amplitude of the ultrasonic transducer; The load status sensor is installed on the contact surface between the ultrasonic transducer and the load to detect the load contact pressure; The acoustic impedance detection module is installed near the interface between the ultrasonic transducer and the workpiece to identify the load medium.

3. The ultrasonic transducer adaptability judgment system based on transient monitoring according to claim 1, characterized in that: The adaptability judgment system also includes a data preprocessing module for Adaptively filter the current signal to remove high-frequency noise; Discrete encoding of load status and load medium; Perform sliding window smoothing on frequency and amplitude signals to eliminate high-frequency noise; Based on the current signal sampling time, the frequency, amplitude, load status and load medium data are interpolated and aligned, and a standardized data packet is output.

4. The adaptability judgment system according to claim 1, characterized in that: The calculation of the initial current threshold based on the collected key transient parameters and steady-state parameters, combined with the preset weight coefficients corresponding to the parameters and the preset basic current threshold, includes: According to the load state and load medium, weights are assigned to the four parameters of frequency, amplitude, load state and load medium; Determine the current threshold offset caused by the frequency based on the difference between the current frequency and the preset standard frequency and the corresponding weight; Determine the current threshold offset caused by the amplitude based on the difference between the current amplitude and the preset standard amplitude and the corresponding weight; Determine, based on the load state and the corresponding weight, an increase in the current threshold caused by the load state; Determine, based on the load medium and the corresponding weight, an increase in the current threshold caused by the load medium; Each offset and increase is added to the preset basic current threshold to obtain the initial current threshold.

5. The adaptability judgment system according to claim 1, characterized in that: The adaptability judgment system further includes an inner loop response control module, which is configured to fine-tune the initial current threshold through a sliding window and a PID controller.

6. The adaptability judgment system according to claim 5, characterized in that: The fine-tuning of the initial current threshold by using a sliding window and a PID controller includes: Taking the current time t as the center, take a sliding window of length Δt and calculate the current mean within the window; Calculating the difference between the current measured and the current average within the window; According to the calculated difference, the proportional adjustment amount, integral adjustment amount and differential adjustment amount of the PID controller are calculated; According to the proportional adjustment amount, the integral adjustment amount and the differential adjustment amount, the initial current threshold is adjusted by a PID controller to obtain a current adjustment threshold.

7. The adaptability judgment system according to claim 5, characterized in that: The fine-tuning of the initial current threshold by using the sliding window and the PID controller includes: Calculate short-term volatility of frequency and amplitude in real time; According to the fluctuation rate, adjust the proportional gain coefficient of the PID controller.

8. The adaptability judgment system according to claim 1, characterized in that: The adaptability judgment system also includes an outer loop stability optimization control module, which is used to optimize the weight coefficients of various parameters through a gradient descent method.

9. The adaptability judgment system according to claim 8, characterized in that: The intelligent judgment and feedback module is further configured to output an adaptation judgment result, and feed the adaptation judgment result back to the outer loop stability optimization control module for continuous optimization.

10. The ultrasonic transducer adaptability judgment system based on transient monitoring according to claim 8, characterized in that: The optimization of each parameter weight coefficient by the gradient descent method includes: Calculate the comprehensive loss function based on historical misjudgment records and current threshold change records; For each key transient parameter and steady-state parameter corresponding to the weight coefficient, calculate the partial derivative of the loss function; Synthesize the gradient vector from the partial derivative array of each parameter; Adjust the weight coefficient of each parameter according to the gradient direction.

Images