A method for judging ultrasonic welding head abnormalities based on beat frequency

Abstract

A method for judging ultrasonic welding head anomalies based on beat frequency is disclosed, belonging to the technical field of ultrasonic welding head anomaly identification and judgment methods. It addresses the shortcomings of existing ultrasonic welding head status monitoring methods, such as difficulty in accurately locating faults, the need for additional sensors, or the inability to achieve real-time control. The method includes: acquiring sound signals during ultrasonic welding; preprocessing the acquired sound signals; performing time-domain analysis and frequency-domain transformation on the preprocessed sound signals; identifying low-frequency howling signals generated by the beat frequency effect in the spectral characteristics; calculating the distance between the anomaly location and the vibrating node using acoustic propagation principles, combined with the physical mechanism of the beat frequency effect and the propagation speed of ultrasound in the welding system material; and performing pattern matching using an anomaly type database to output detailed diagnostic results including fault location and anomaly type. This improves the scientific accuracy and precision of fault location, reduces system complexity, and enhances diagnostic accuracy.

Description

Technical Field

[0001] This invention relates to the technical field of ultrasonic welding head abnormality identification and judgment methods. Background Technology

[0002] Ultrasonic welding is a process technology that uses high-frequency mechanical vibration to generate localized heat at the material contact surface to achieve material bonding. It is widely used in the joining of materials such as plastics and metals. In an ultrasonic welding system, the ultrasonic welding head is a key component, and its performance directly affects the weld quality.

[0003] Currently, the monitoring of ultrasonic welding head status mainly relies on the following methods: first, indirectly judging the welding status by measuring electrical parameters (such as current, power, etc.); second, monitoring physical parameters such as temperature and pressure; and third, judging the welding quality by post-weld product inspection. These methods have obvious drawbacks: electrical parameter monitoring is difficult to accurately locate faults; physical parameter monitoring requires additional sensors, increasing system complexity; and post-weld product inspection cannot achieve real-time control.

[0004] During ultrasonic welding, the vibration system (including the amplitude transformer, welding head, etc.) generates sound when it is working. When the system malfunctions, it will produce special acoustic signals outside the normal operating frequency. When the vibration system has multiple cross-sections or abnormal reflection points, the resonance will produce a beat frequency effect, resulting in a low-frequency beat frequency howling phenomenon (hereinafter referred to as low-frequency howling or howling).

[0005] The beat frequency effect is an important phenomenon in wave mechanics. It occurs when two harmonic vibrations with similar frequencies are superimposed. Suppose a particle simultaneously participates in two harmonic vibrations with different frequencies; its vibration expression is as follows:

[0006] x1 = A1cos(ω1t + φ 01 (1) x2 = A2cos(ω2t + φ) 02 (2)

[0007] According to the superposition principle, the displacement of the resultant motion is:

[0008] x = x1 + x2 = A1cos(ω1t + φ 01 ) + A2cos(ω2t + φ 02 (3)

[0009] For ease of calculation, let A1 = A2 = A, φ 01 = φ 02 = φ0, then the above expression can be transformed into:

[0010] x = 2A cos((ω2 - ω1) / 2 × t) cos((ω2 + ω1) / 2 × t + φ0) (4)

[0011] In the usual practical situation, the two frequencies are relatively close, and |ω2 - ω1| ≪ ω1. The first factor in equation (4) changes slowly with time, and the second factor is a simple harmonic function with an angular frequency close to ω (i.e. close to ω1 and ω2). Therefore, the synthesized vibration can be approximated as a harmonic vibration with an angular frequency of (ω1 + ω2) / 2 ≈ ω1 ≈ ω2 and an amplitude of |2Acos((ω2 - ω1)t / 2)|.

[0012] When two harmonic vibrations with relatively high frequencies and small differences are combined, the resulting amplitude exhibits a slow, periodic change in strength, sometimes strong and sometimes weak. This phenomenon is called a beat. The frequency at which the beat occurs is called the beat frequency, which is equal to the difference between the two vibration frequencies, i.e.:

[0013] fbeat = |f2 - f1| = |ω2 - ω1| / 2π (5)

[0014] In an ultrasonic vibration system, when the system is intact and without abnormalities, it primarily generates vibrations at a single frequency. However, when an abnormality occurs at a certain location in the system (such as a crack, looseness, or wear), a new acoustic reflection interface forms at that location, producing reflected vibrations with a slightly different frequency from the dominant vibration. When these two vibrations with similar frequencies are superimposed, according to the beat frequency effect principle, a low-frequency modulated signal with a frequency of |f2 - f1| is generated, which acoustically manifests as a whistling phenomenon.

[0015] Currently, there is no existing technology that utilizes the physical principle of beat frequency effect to diagnose and accurately locate faults by analyzing the correspondence between howling frequency and fault location. Summary of the Invention

[0016] In summary, the purpose of this invention is to address the shortcomings of existing ultrasonic welding head status monitoring methods, such as difficulty in accurately locating faults through electrical parameter monitoring, the need for additional sensors and increased system complexity for physical parameter monitoring, and the inability to achieve real-time control during post-product testing. The invention proposes a method for determining ultrasonic welding head anomalies based on beat frequency.

[0017] To address the shortcomings of the technology proposed in this invention, the following technical solution is adopted:

[0018] A method for determining abnormalities in an ultrasonic welding head based on beat frequency, characterized in that the method includes the following steps:

[0019] S1. Use a high-frequency microphone to collect sound signals during the ultrasonic welding process and capture the normal operating frequency of the system and its possible abnormal frequency components.

[0020] S2. Preprocess the acquired sound signals, including noise reduction, filtering and signal enhancement, remove environmental interference and extract effective acoustic feature signals;

[0021] S3. Perform time-domain analysis and frequency-domain transformation on the preprocessed sound signal, and use the fast Fourier transform algorithm to obtain the complete spectral characteristics of the acoustic feature signal;

[0022] S4. Identify the low-frequency beat frequency howling signal generated by the beat frequency effect in the spectral characteristics. Based on the beat frequency principle fbeat = |f2 - f1|, extract its frequency characteristics and analyze its persistence and stability.

[0023] S5. Based on the extracted low-frequency beat frequency, combined with the physical mechanism of the beat frequency effect and the propagation speed of ultrasonic waves in the welding system material, the distance between the abnormal location and the vibration node is calculated using the acoustic propagation principle.

[0024] S6. Based on the calculation results, locate the fault location, perform pattern matching in conjunction with the anomaly type database, and output detailed diagnostic results including the fault location and anomaly type.

[0025] The technical features that further define the present invention include:

[0026] In step S2, the preprocessing includes using a bandpass filter to remove environmental noise, extracting ultrasonic signals in the 20kHz-100kHz frequency band and beat frequency signals in the 100Hz-10kHz frequency band, and processing them independently.

[0027] In step S3, the frequency domain transformation uses the Fast Fourier Transform algorithm, and the analysis window length for low-frequency monitoring is 500 milliseconds (overlapping can be set); for the estimation of the main frequency and amplitude / frequency characteristics of the ultrasonic carrier band, a shorter analysis window is used for calculation.

[0028] In step S4, the method for identifying the low-frequency howling signal generated by the beat frequency effect is as follows: search for the energy peak in the 100Hz-10kHz frequency band; set the detection threshold to 5 times the average background noise of the frequency band; require the howling signal to remain continuous in the time domain and have a duration of not less than 0.5 seconds; and verify the correlation between the howling signal and the main operating frequency through correlation analysis.

[0029] In step S5, the fault location method based on the beat frequency effect physical mechanism is as follows: when the beat frequency howling frequency fbeat is detected, the abnormal reflection frequency f2 is deduced from fbeat = |f2 - f1|; the formula for calculating the distance between the abnormal location and the vibration node is: D = v / (2 × Δf), where v is the propagation speed of sound in the material and Δf is the frequency difference. It should be noted that this relationship is an engineering approximation under the premise of a specific simplified structural model and experimental calibration. Often, ultrasonic transducers, amplitude transformers, and molds are made of different materials and have different sound velocities, but the most easily damaged are the welding head and its connecting surface. This invention is mainly used to detect faults in the welding head to determine whether a low-frequency howling fault has occurred.

[0030] It also includes establishing an anomaly type database based on beat frequency features. The database contains: beat frequency feature patterns corresponding to different anomaly types; frequency range, duration, and energy distribution feature parameters of various anomalies; and the correspondence between anomaly severity and beat frequency intensity; and continuously optimizes feature recognition accuracy through machine learning algorithms.

[0031] It also includes real-time monitoring and early warning functions; establishes a time-series analysis model of beat frequency signals to track the development trend of anomalies; sets multi-level early warning thresholds to judge the severity of anomalies based on beat frequency intensity and duration; automatically records the time, location, and type of anomalies when an anomaly is detected; and provides anomaly development trend prediction to achieve preventive maintenance.

[0032] The distance calculation results are corrected for material properties using the following formula: D = v / (2 × Δf) × K, where v is the speed of sound propagation in the material, Δf is the beat frequency, and K is a correction coefficient obtained through experimental calibration based on a prototype or standard sample. Its value is related to the material, structure, and boundary constraints, and it is used for approximate positioning reference in engineering.

[0033] The beneficial effects of this invention are as follows:

[0034] 1. Based on the physical principle of beat frequency effect, this invention establishes a solid theoretical foundation and clarifies the quantitative relationship between howling frequency and fault location through mathematical derivation, thereby improving the scientificity and accuracy of fault location.

[0035] 2. The method of the present invention utilizes acoustic signal analysis technology to achieve non-contact monitoring of the ultrasonic welding head status, avoiding the problem of needing to install additional sensors on the equipment in traditional methods and reducing system complexity;

[0036] 3. By deeply analyzing the physical mechanism of the beat frequency effect, this invention can distinguish different types of anomalies (such as structural anomalies, material anomalies, etc.), realize intelligent identification of fault types, and improve the accuracy of diagnosis;

[0037] 4. The method of the present invention can monitor the ultrasonic welding process in real time. Based on the temporal characteristics of the beat frequency signal, it can predict the abnormal development trend, realize preventive maintenance, and prevent production losses caused by sudden failures.

[0038] 5. The system has low implementation cost, requiring only the addition of a high-frequency microphone and signal processing unit, without changing the structure of the existing welding equipment, and has good engineering applicability;

[0039] 6. The theoretical model and analysis method established in this paper can be extended to the fault diagnosis of other types of ultrasonic equipment, and has broad application prospects and promotion value. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the system structure of the present invention.

[0041] Figure 2 This is a flowchart of the method of the present invention.

[0042] Figure 3 This is a schematic diagram of the beat frequency effect mechanism.

[0043] Figure 4 This is a schematic diagram of the spectrum under normal and abnormal conditions. Detailed Implementation

[0044] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0045] This invention discloses a method for judging ultrasonic welding head abnormalities based on beat frequency, which is based on the physical principle of the beat frequency effect and used for real-time monitoring and diagnosis of abnormal states of welding heads in ultrasonic welding equipment. By deeply understanding the physical mechanism of the beat frequency effect and analyzing the acoustic signals generated during ultrasonic welding, especially the low-frequency howling generated by the beat frequency effect, a quantitative relationship between the howling frequency and the fault location is established, enabling real-time detection and precise positioning of abnormal conditions in ultrasonic welding heads. (Refer to...) Figure 2 As shown in the diagram, S1 acquires acoustic signals; S2 performs signal preprocessing; S3 performs spectrum / time-frequency analysis; S4 detects beat frequencies; S5 locates faults, using D = v / (2 × Δf) × K for approximate engineering location; and S6 outputs the anomaly result. The specific method of this invention includes the following steps:

[0046] S1. Use a high-frequency microphone to collect sound signals during the ultrasonic welding process, and capture the normal operating frequency of the system and its possible abnormal frequency components; preferably, the sampling frequency of the high-frequency microphone is not less than 100kHz, so as to ensure that the fundamental wave and its harmonics of the ultrasonic working frequency can be collected, while capturing the low-frequency components generated by the beat frequency effect.

[0047] S2. Preprocess the acquired sound signal, including noise reduction, filtering and signal enhancement, remove environmental interference and extract effective acoustic feature signals; preferably, the preprocessing includes using a bandpass filter to remove environmental noise, extracting ultrasonic signals in the 20kHz-100kHz frequency band and beat frequency signals in the 100Hz-10kHz frequency band, and processing them independently to adapt to common working frequency bands such as 20-40kHz.

[0048] S3. Perform time-domain analysis and frequency-domain transformation on the preprocessed audio signal, and obtain the complete spectral characteristics of the signal using Fast Fourier Transform (FFT). Preferably, the frequency-domain transformation uses the Fast Fourier Transform (FFT) algorithm, with an analysis window length of 500 milliseconds for low-frequency monitoring, which can be overlapped. For the estimation of the dominant frequency and amplitude / frequency characteristics of the ultrasonic carrier band, a shorter analysis window is used for calculation. The preferred window function is a Hamming window, and the sampling sample should be large enough, for example, a 500ms sliding window performing data analysis once every 500 milliseconds to improve the accuracy and continuity of time-frequency analysis and to monitor the low-frequency jitter of the welding head.

[0049] S4. Identify the low-frequency howling signal generated by the beat frequency effect in the frequency spectrum. Based on the beat frequency principle fbeat = |f2 - f1|, extract its frequency characteristics and analyze its persistence and stability. Preferably, the method for identifying the low-frequency howling signal generated by the beat frequency effect is as follows:

[0050] • Search for energy peaks in the 100Hz-10kHz frequency band;

[0051] • Set the detection threshold to 5 times the average background noise level of this frequency band;

[0052] • The howling signal must remain continuous in the time domain for at least 0.5 seconds.

[0053] Correlation analysis was used to verify the relationship between the howling signal and the main operating frequency, confirming that the howling signal originated from the beat frequency effect.

[0054] S5. Based on the extracted howling frequency, combined with the physical mechanism of the beat frequency effect and the propagation speed of ultrasound in the welding system material, the distance between the abnormal location and the vibrating node is calculated using the acoustic propagation principle; preferably, the fault location method based on the physical mechanism of the beat frequency effect is as follows:

[0055] • When the beat frequency fbeat is detected, the abnormal reflection frequency f2 is deduced from fbeat = |f2 - f1|.

[0056] • The formula for calculating the distance between the abnormal location and the vibration node is: D = v / (2 × Δf), where v is the propagation speed of the sound wave in the material and Δf is the frequency difference;

[0057] The propagation characteristics of sound waves in different materials are used to correct the distance calculation results using the following formula: D = v / (2 × Δf) × K, where K is the correction coefficient for material and geometry.

[0058] Reference Figure 3 As shown, the incident wave and the reflected wave superimpose to generate an envelope. The beat frequency fbeat = |f2 - f1| has an engineering approximation relationship with the distance D from the vibration node A to the fault location B: D = v / (2 × Δf) × K.

[0059] It should be noted that this relationship is an engineering approximation under the premise of a simplified model of a specific structure and experimental calibration. Often, ultrasonic transducers, amplitude transformers, and molds are made of different materials and have different sound velocities. However, the welding head and its connecting surface are the most prone to failure. This invention is mainly used to detect welding head failures, mainly to determine whether a low-frequency howling fault has occurred, and to locate suspicious sections, rather than directly outputting the absolute position.

[0060] S6. Based on the calculation results, locate the fault location, perform pattern matching in conjunction with the anomaly type database, and output detailed diagnostic results including the fault location and anomaly type.

[0061] In its specific implementation, this invention also includes establishing an anomaly type database based on beat frequency characteristics, the database comprising:

[0062] Beat frequency characteristic patterns corresponding to different anomaly types (cracks, loosening, jitter caused by insufficient tracking bandwidth, etc.);

[0063] Characteristic parameters of various anomalies, such as frequency range, duration, and energy distribution;

[0064] The correlation between the severity of the anomaly and the beat frequency intensity;

[0065] The accuracy of feature recognition is continuously optimized through machine learning algorithms.

[0066] To achieve real-time monitoring and early warning functions, the method of the present invention further includes:

[0067] Establish a time-series analysis model for beat frequency signals to track anomaly development trends;

[0068] Set multi-level early warning thresholds and determine the severity of the anomaly based on the beat frequency intensity and duration;

[0069] When an anomaly is detected, it automatically records detailed information such as the time, location, and type of the anomaly.

[0070] Provides predictions of abnormal development trends, enabling preventative maintenance.

[0071] Reference Figure 4As shown, (a) is normal; (b) is abnormal. The abnormal state is characterized by beat frequency peaks in the low-frequency range (example 1.2kHz), with the high-frequency dominant frequency located in the 40kHz region.

[0072] In the specific implementation process, in order to acquire the sound signal during the ultrasonic welding process in step S1, a high-frequency microphone can be installed around the ultrasonic welding equipment at a distance of 25cm from the welding head. The microphone is selected as a high-frequency microphone with a frequency response range of 20Hz-100kHz, preferably a piezoelectric microphone. For example, multiple piezoelectric microphones of different frequency bands can be used to form a microphone array, and the array is calibrated to achieve a flatter response curve. The sampling frequency is set to 192kHz, the quantization accuracy is 24bit, which satisfies the Nyquist sampling theorem, and can be implemented using an inexpensive audio ADC. Figure 1 In this process, the ultrasonic welding equipment is connected to a transducer, providing a drive signal to the transducer. The transducer is connected to a welding head via an amplitude transformer, and the welding head performs ultrasonic welding on the workpiece. A high-frequency microphone collects the sound signal during the ultrasonic welding process and transmits it to the signal acquisition module 1. Then, the signal processing module 2 performs preprocessing on the collected sound signal in step S2. After the sound signal preprocessing, the frequency spectrum / time-frequency analysis module 3 executes steps S3 and S4 to perform time-domain analysis and frequency-domain conversion on the preprocessed sound signal and identify the low-frequency howling signal generated by the beat frequency effect in the frequency spectrum. Then, the fault location module 4 executes step 5 to calculate the distance between the abnormal location and the vibration node using the acoustic propagation principle. Finally, the abnormal diagnosis result output module 5 executes step S6 to output a detailed diagnostic result containing the fault location and abnormality type.

[0073] In the specific implementation process, in order to achieve the preprocessing of the collected sound signal in step S2, the following is specifically adopted:

[0074] Use a fourth-order Butterworth high-pass filter (e.g., cutoff frequency 120Hz) to remove power frequency interference and interference transmitted to the transducer by the rectified power frequency multiplier.

[0075] Use an 8th-order elliptic low-pass filter (e.g., cutoff frequency 100kHz, select a matching bandwidth based on the device frequency) to remove high-frequency noise;

[0076] Adaptive filtering algorithms are used to further reduce the impact of environmental noise;

[0077] Automatic gain control ensures the consistency of signal amplitude.

[0078] In step S3, the following methods are used when performing time-domain analysis and frequency-domain transformation on the preprocessed audio signal:

[0079] Continuous sampling for 500ms, using the Hamming window function;

[0080] Calculate the power spectral density (PSD) and short-time Fourier transform (STFT);

[0081] Using wavelet transform for time-frequency analysis improves the detection accuracy of low-frequency signals.

[0082] In step S4, low-frequency howling signals are detected based on beat frequency theory: For example... Figure 3 As shown, when an abnormality occurs in an ultrasonic vibration system at a certain location, the following physical process will occur:

[0083] The incident wave propagates in the system at a frequency f1;

[0084] Reflection occurs at abnormal locations, forming a reflected wave with frequency f2;

[0085] The superposition of two waves produces a beat frequency effect, with the beat frequency being fbeat = |f2 - f1|.

[0086] Acoustically, it manifests as periodic amplitude modulation, producing an audible whistling sound.

[0087] Implementation of low-frequency howling signal detection algorithm:

[0088] Peak search was performed in the 100Hz-10kHz frequency band to detect signals with energy exceeding 5 times the background noise.

[0089] Correlation analysis was used to verify the correlation between the detected low-frequency signal and the main operating frequency;

[0090] Transient interference is eliminated by judging the continuity in the time domain (requiring the signal duration to be >0.5 seconds);

[0091] Frequency tracking algorithms are used to monitor the stability of the howling frequency.

[0092] In step S5, fault location based on the physical model: after a stable beat frequency howling signal is detected, precise fault location is performed.

[0093] The offset Δf of the abnormal reflection frequency is calculated by inferring the beat frequency f.

[0094] Considering the propagation characteristics of sound waves in the vibrating system material, for aluminum alloy material, the sound wave propagation speed is approximately 6300 m / s;

[0095] The corrected distance calculation formula is: D = v / (2 × Δf) × K, where K is the correction factor for material and geometry;

[0096] By combining the system's geometry, the calculated distance is mapped to a specific physical location.

[0097] In step S6, the intelligent diagnosis and early warning system:

[0098] Establish a feature database of typical anomaly patterns;

[0099] Use Support Vector Machine (SVM) for anomaly type classification;

[0100] A three-level early warning system is implemented: green (normal), yellow (minor abnormality), and red (serious abnormality).

[0101] Preferably, the diagnostic system provides a signal output for the generator to record, but does not cause the generator to stop in case of an anomaly.

[0102] This invention fully utilizes the physical principle of beat frequency effect and establishes a complete technical system from theory to practice. It can not only accurately detect abnormalities in ultrasonic welding heads, but also achieve intelligent identification of abnormality types and prediction of development trends, providing strong support for the intelligent maintenance of ultrasonic welding equipment.

[0103] This invention is not limited to the above embodiments. Various modifications and variations can be made to this invention without departing from the basic idea of ​​this invention based on the physical principle of beat frequency effect. All such modifications and variations fall within the protection scope of this invention.

Claims

1. A method for determining abnormalities in an ultrasonic welding head based on beat frequency, characterized in that... The method includes the following steps: S1. Use a high-frequency microphone to collect sound signals during the ultrasonic welding process and capture the normal operating frequency of the system and its possible abnormal frequency components. S2. Preprocess the acquired sound signals, including noise reduction, filtering and signal enhancement, remove environmental interference and extract effective acoustic feature signals; S3. Perform time-domain analysis and frequency-domain transformation on the preprocessed sound signal, and use the fast Fourier transform algorithm to obtain the complete spectral characteristics of the acoustic feature signal; S4. Identify the low-frequency beat frequency howling signal generated by the beat frequency effect in the spectral characteristics. Based on the beat frequency principle fbeat = |f2 -f1|, extract its frequency characteristics and analyze its persistence and stability. S5. Based on the extracted low-frequency beat frequency howling frequency, combined with the physical mechanism of the beat frequency effect and the propagation speed of ultrasonic waves in the welding system material, the distance between the abnormal location and the vibration node is calculated using the acoustic propagation principle. S6. Based on the calculation results, locate the fault location, perform pattern matching in conjunction with the anomaly type database, and output detailed diagnostic results including the fault location and anomaly type.

2. The method for determining ultrasonic welding head abnormalities based on beat frequency according to claim 1, characterized in that: In step S1, the sampling frequency of the high-frequency microphone signal is not less than 100kHz.

3. The method for judging ultrasonic welding head abnormalities based on beat frequency according to claim 1, characterized in that: In step S2, the preprocessing includes using a bandpass filter to remove environmental noise, extracting ultrasonic signals in the 20kHz-100kHz frequency band and beat frequency signals in the 100Hz-10kHz frequency band, and processing them independently.

4. The method for judging ultrasonic welding head abnormalities based on beat frequency according to claim 1, characterized in that: In step S3, the frequency domain transformation uses the Fast Fourier Transform algorithm, and the analysis window length for low-frequency monitoring is 500 milliseconds; for the estimation of the main frequency and amplitude / frequency characteristics of the ultrasonic carrier band, a shorter analysis window is used for calculation.

5. The method for determining ultrasonic welding head abnormalities based on beat frequency according to claim 1, characterized in that: In step S4, the method for identifying the low-frequency beat frequency howling signal generated by the beat frequency effect is as follows: search for the energy peak in the 100Hz-10kHz frequency band; set the detection threshold to 5 times the average background noise of the frequency band; require the low-frequency beat frequency howling signal to remain continuous in the time domain and have a duration of not less than 0.5 seconds; and verify the correlation between the low-frequency beat frequency howling signal and the main operating frequency through correlation analysis.

6. The method for judging ultrasonic welding head abnormalities based on beat frequency according to claim 1, characterized in that: In step S5, the fault location method based on the beat frequency effect physical mechanism is as follows: when a low-frequency beat frequency fbeat is detected, the abnormal reflection frequency f2 is inferred from fbeat = |f2 - f1|; the formula for calculating the distance between the abnormal location and the vibration node is: D = v / (2 × Δf), where v is the propagation speed of the sound wave in the material, Δf is the frequency difference, and Δf = fbeat = |f2 - f1|.

7. The method for determining ultrasonic welding head abnormalities based on beat frequency according to claim 1, characterized in that: It also includes establishing an anomaly type database based on beat frequency features. The database contains: beat frequency feature patterns corresponding to different anomaly types; frequency range, duration, and energy distribution feature parameters of various anomalies; and the correspondence between anomaly severity and beat frequency intensity; and continuously optimizes feature recognition accuracy through machine learning algorithms.

8. The method for determining ultrasonic welding head abnormalities based on beat frequency according to claim 1, characterized in that: It also includes real-time monitoring and early warning functions; establishes a time-series analysis model of beat frequency signals to track the development trend of anomalies; sets multi-level early warning thresholds to judge the severity of anomalies based on beat frequency intensity and duration; automatically records the time, location, and type of anomalies when an anomaly is detected; and provides anomaly development trend prediction to achieve preventive maintenance.

9. A method for determining ultrasonic welding head abnormalities based on beat frequency according to claim 6, characterized in that: Material corrections are applied to the distance calculation results using the distance calculation formula: D = v / (2 × Δf) × K, where v is the propagation speed of sound in the material, Δf is the beat frequency, and K is a correction coefficient obtained through experimental calibration based on a prototype or standard sample. Its value is related to the material, structure, and boundary constraints, and is used for engineering approximate positioning reference.

Images