Ultrasonic cleaning temperature monitoring and power adaptive adjustment control method and system

Abstract

The invention relates to the technical field of adjustment control, in particular to an ultrasonic cleaning temperature monitoring and power self-adaptive adjustment control method and system, and the method comprises the following steps: carrying out the discrete Fourier transform of a sound signal time sequence according to a sound signal collected by a high-bandwidth hydrophone and a liquid temperature sensed by a thermocouple, and obtaining a time sequence of the sound signal; and establishing a multi-dimensional cavitation state feature vector. According to the method, discrete Fourier transform is carried out on a sound signal time sequence, fundamental frequency, harmonic wave and broadband noise features and liquid temperature parameters are combined to construct cavitation state feature vectors, so that multi-parameter linkage is realized on an information acquisition dimension, a probability model is further constructed through amplitude distribution, and the probability model is used for realizing multi-parameter linkage. Information entropy and temperature coupling are introduced to form a chaos degree index, depiction of a complex cleaning state is achieved, and higher state sensitivity and fine control capacity are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of regulation and control, in particular to an ultrasonic cleaning temperature monitoring and power adaptive regulation and control method and system. BACKGROUND

[0002] The technical field of regulation and control mainly involves real-time monitoring, calculation analysis and dynamic feedback on the operating state to achieve stable output.

[0003] The prior art lacks deep perception of multivariate coupling characteristics, resulting in response lag and adjustment error when dealing with dynamic process changes. At the same time, since traditional regulation and control is mostly based on linear difference correction, it is difficult to effectively capture nonlinear behaviors in the cleaning process, such as critical mutation of cavitation state or abnormal response of wideband noise disturbance, which may easily lead to over-regulation or under-regulation in complex scenarios, and thus increase energy consumption or cause incomplete cleaning. Therefore, improvements are needed. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art and to provide an ultrasonic cleaning temperature monitoring and power adaptive regulation and control method and system.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical solution, an ultrasonic cleaning temperature monitoring and power adaptive regulation and control method, comprising the following steps: Discrete Fourier transform is performed on the time series of the acoustic signal according to the acoustic signal collected by the high-bandwidth hydrophone and the liquid temperature sensed by the thermocouple, and a multi-dimensional cavitation state feature vector is established; According to the multi-dimensional cavitation state feature vector, the subharmonic intensity, harmonic energy ratio and wideband noise level are taken as coordinates to determine the position of the current working point in the cavitation state space, and the amplitude values of the time series of the acoustic signal are discretized and binned, the signal point frequency in each bin is counted, and the amplitude probability distribution is constructed. The amplitude probability distribution is substituted into the Shannon entropy formula to calculate the information entropy, and the information entropy value is modified in combination with the liquid temperature to obtain a temperature-acoustic coupling chaos degree index; According to the temperature-acoustic coupling chaos degree index, a target chaos degree interval is set, the current temperature-acoustic coupling chaos degree index is compared with the upper and lower limits of the target chaos degree interval to obtain a chaos degree deviation value, and a power regulation deviation amount is established according to the chaos degree deviation value; According to the power regulation deviation amount, the power regulation deviation amount and the output power set value of the current ultrasonic generator are calculated to obtain an updated power set value, the updated power set value is converted into the duty cycle or voltage amplitude of the pulse width modulation signal driving the ultrasonic transducer to obtain a closed-loop power control signal.

[0006] Preferably, the step of obtaining the multi-dimensional cavitation state feature vector is: According to the time series of acoustic signals obtained by the high-bandwidth hydrophone and the liquid temperature values recorded by the thermocouple, the time series of acoustic signals is decomposed in the frequency domain and the energy is accumulated in each frequency band to obtain the fundamental frequency energy value, the harmonic energy value, the sub-harmonic intensity value and the broadband noise base value, and a frequency spectrum feature set is generated; According to the frequency spectrum feature set, the ratio of each harmonic energy value to the fundamental frequency energy value is calculated one by one while keeping the time index consistent, and the liquid temperature value is combined with the sub-harmonic intensity value and the broadband noise base value synchronously to generate a temperature-acoustic joint feature group; According to the temperature-acoustic joint feature group, the ratio sequence, the sub-harmonic intensity value, the broadband noise base value and the liquid temperature value are spliced in a fixed arrangement order and labeled with dimension labels to form a multi-dimensional cavitation state feature vector.

[0007] Preferably, the step of obtaining the amplitude probability distribution is: According to the multi-dimensional cavitation state feature vector, the amplitude values of the time series of acoustic signals are discretized and binned in equal-width intervals, and the frequency of the occurrence of the amplitude values is counted in each bin. Each bin frequency is stored corresponding to the bin number to generate an amplitude bin count. According to the amplitude bin count, each bin count is divided by the sum of all bin counts while keeping the bin order consistent to obtain the bin probability value, and the amplitude probability distribution is formed by combining.

[0008] Preferably, the step of obtaining the temperature-acoustic coupling chaos degree index is: According to the amplitude probability distribution and the liquid temperature value, the temperature-acoustic coupling chaos degree index is calculated.

[0009] Preferably, the step of obtaining the chaos degree deviation value is: According to the temperature-acoustic coupling chaos degree index and the upper and lower limits of the target chaos degree interval, the interval center point is calculated and the difference between the temperature-acoustic coupling chaos degree index and the interval center point is obtained. The difference result is taken as the chaos degree deviation value.

[0010] Preferably, the step of obtaining the power adjustment deviation amount is: According to the chaos degree deviation value, the current time chaos degree deviation value is extracted and compared with the previous time and the previous time chaos degree deviation value one by one, and the target chaos degree interval width and the sampling time interval are combined to obtain a chaos degree deviation sequence. According to the chaos degree deviation sequence and the liquid temperature value, the power adjustment deviation amount is calculated.

[0011] Preferably, the step of obtaining the updated power setting value is: According to the power adjustment deviation and the output power set value, the updated power set value is generated by aligning and checking units, adding and recording the results by algebraic signs, and replacing the boundary value if it exceeds the allowed range of the output power set value.

[0012] Preferably, the obtaining step of the closed-loop power control signal is: According to the updated power set value, the pulse width modulation mode or the voltage control mode is selected by mapping the ultrasonic transducer rated power range to the control amount range, and the duty cycle or the voltage amplitude of the pulse width modulation signal is calculated and quantized by the counter to obtain the duty cycle or the voltage amplitude of the pulse width modulation signal. According to the duty cycle or the voltage amplitude of the pulse width modulation signal, the closed-loop power control signal is generated by frame packaging and timing alignment, writing into the ultrasonic transducer drive register, combining the carrier frequency, the driving polarity, the dead time and the enable state.

[0013] The application also provides an adjustment control system, comprising: The signal processing module is used for performing discrete Fourier transform on the time sequence of the acoustic signal according to the acoustic signal collected by the high-bandwidth hydrophone and the liquid temperature sensed by the thermocouple, and establishing a multi-dimensional cavitation state feature vector. The feature analysis module is used for determining the position of the current working point in the cavitation state space by taking the subharmonic intensity, the harmonic energy ratio and the broadband noise level as coordinates according to the multi-dimensional cavitation state feature vector, discretizing and binning the amplitude value of the time sequence of the acoustic signal, counting the frequency of the signal points in each bin, constructing an amplitude probability distribution, substituting the amplitude probability distribution into a Shannon entropy formula to calculate an information entropy, and correcting the information entropy value in combination with the liquid temperature to obtain a temperature-acoustic coupling chaos degree index. The chaos degree determination module is used for setting a target chaos degree interval according to the temperature-acoustic coupling chaos degree index, comparing the current temperature-acoustic coupling chaos degree index with the upper and lower limits of the target chaos degree interval to obtain a chaos degree deviation, and establishing a power adjustment deviation according to the chaos degree deviation. The power control module is used for calculating the power adjustment deviation and the output power set value of the current ultrasonic generator according to the power adjustment deviation, obtaining an updated power set value, converting the updated power set value into the duty cycle or the voltage amplitude of the pulse width modulation signal for driving the ultrasonic transducer, and obtaining a closed-loop power control signal.

[0014] Compared with the prior art, the application has the advantages and positive effects that: This invention utilizes Discrete Fourier Transform (DFT) on the time series of acoustic signals to jointly construct a cavitation state feature vector by combining fundamental frequency, harmonics, and broadband noise characteristics with liquid temperature parameters. This achieves multi-parameter linkage in information acquisition. Furthermore, a probabilistic model is constructed through amplitude distribution, and information entropy is introduced to couple with temperature to form a chaos index, enabling the characterization of complex cleaning states. This results in stronger state sensitivity and finer control capabilities. In terms of adjustment strategy, the deviation is constructed by the difference between the chaos index and the target interval. Based on the deviation, the dynamic change trends of proportional, integral, and differential dimensions are combined to form a predictive and corrective adjustment output. In the output stage, the continuously changing adjustment value is transformed into a discrete and controllable signal carrier, achieving real-time response to the driving signal. This improves the state identification accuracy, response control flexibility, and power adjustment stability during ultrasonic cleaning, enabling the cleaning system to dynamically respond to complex liquid environment changes and maintain stable cavitation efficiency output. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the steps of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0017] Please see Figure 1 This invention provides a technical solution: an ultrasonic cleaning temperature monitoring and power adaptive adjustment control method, comprising the following steps: Based on the acoustic signal collected by the high-bandwidth hydrophone and the liquid temperature sensed by the thermocouple, a discrete Fourier transform is performed on the time series of the acoustic signal to establish a multidimensional cavitation state feature vector. Based on the multidimensional cavitation state feature vector, the subharmonic intensity, harmonic energy ratio, and broadband noise level are used as coordinates to determine the position of the current operating point in the cavitation state space. At the same time, the amplitude value of the acoustic signal time series is discretized into bins, the frequency of the signal points in each bin is counted, and an amplitude probability distribution is constructed. The amplitude probability distribution is substituted into the Shannon entropy formula to calculate the information entropy, and the information entropy value is corrected by combining the liquid temperature to obtain the temperature-acoustic coupling chaos index. Based on the temperature-acoustic coupling chaos index, a target chaos range is set. The current temperature-acoustic coupling chaos index is compared with the upper and lower limits of the target chaos range to obtain the chaos deviation value. Based on the chaos deviation value, the power adjustment deviation is established. Based on the power adjustment deviation, the power adjustment deviation is calculated with the current output power setting value of the ultrasonic generator to obtain the updated power setting value. The updated power setting value is then converted into the duty cycle or voltage amplitude of the pulse width modulation signal driving the ultrasonic transducer to obtain the closed-loop power control signal.

[0018] The steps for obtaining the multidimensional cavitation state feature vector are as follows: Based on the acoustic signal time series obtained by the high-bandwidth hydrophone and the liquid temperature value recorded by the thermocouple, the acoustic signal time series is decomposed in the frequency domain and the energy is accumulated band by band to obtain the fundamental frequency energy value, the energy values ​​of each harmonic, the intensity values ​​of each harmonic and the broadband noise floor value, and generate a set of spectral features. Based on the spectral feature set, the ratio of each harmonic energy value to the fundamental frequency energy value is calculated one by one while keeping the time index consistent. The liquid temperature value, the harmonic intensity value, and the broadband noise floor value are then synchronously merged to generate a temperature acoustic joint feature set. Based on the temperature-acoustic joint feature set, the feature vector of multidimensional cavitation state is formed by splicing the ratio sequence, subharmonic intensity values, broadband noise floor values ​​and liquid temperature values ​​in a fixed order and labeling the dimensions.

[0019] Specifically, based on the acoustic signal time series acquired by a high-bandwidth hydrophone at a sampling rate of no less than 2MHz and the liquid temperature values ​​synchronously recorded by thermocouples, a fixed-length segment of the acoustic signal time series, for example, 4096 sampling points, is first processed using a Hanning window function. Then, a Fast Fourier Transform is performed on the windowed signal sequence to transform it from the time domain to the frequency domain, obtaining spectral data containing the amplitude and phase of each frequency component. Next, based on the driving frequency of the ultrasonic generator, for example 40kHz, the range of each characteristic frequency band is defined. The calculation range of the fundamental frequency energy is set to fluctuate by 2.5% above and below the driving frequency, i.e., the interval from 39kHz to 41kHz. The calculation range of each harmonic energy is set to the corresponding interval of an integer multiple of the fundamental frequency; for example, the second harmonic is 78kHz to 82kHz, the third harmonic is 117kHz to 123kHz, and so on. The calculation range of the subharmonic intensity is set to the corresponding interval of half the fundamental frequency. For the calculation of the broadband noise floor, specifically from 19kHz to 21kHz, the previously defined fundamental frequency, harmonics, and subharmonic frequency bands are first identified and excluded. Then, multiple reference frequency bands without significant energy peaks are selected from the remaining spectrum, such as 25kHz to 35kHz, 50kHz to 70kHz, and 95kHz to 115kHz. The average of the squared amplitudes of all frequency points within these reference frequency bands is calculated and used as the broadband noise floor value. After defining each frequency band, the amplitudes of the Fourier transform results of all frequency points within each preset frequency band are squared and these squared values ​​are accumulated to calculate the fundamental frequency energy value, harmonic energy value, and subharmonic intensity value. Finally, the calculated fundamental frequency energy value, the sequence of energy values ​​of all harmonics, the subharmonic intensity value, and the broadband noise floor value, along with their corresponding timestamp information, are combined to generate a spectral feature set.

[0020] Based on the spectral feature set generated in the previous step, which contains the fundamental frequency energy value, harmonic energy values, harmonic intensity values, and broadband noise floor value corresponding to each timestamp, a ratio calculation is first performed. Specifically, for each time point in the spectral feature set, the energy values ​​of the second harmonic, third harmonic, and so on up to the preset highest harmonic, such as the fifth harmonic, are extracted. Then, these harmonic energy values ​​are divided one by one by the fundamental frequency energy value at the same time point, thus obtaining a set of dimensionless harmonic energy ratios. For example, if the fundamental frequency energy at time t is 1.2, the second harmonic energy is 0.3, and the third harmonic energy is 0.15, then the corresponding second harmonic energy ratio is 0.25, and the third harmonic energy ratio is 0.125. During the calculation process, each ratio is strictly kept consistent with its original value. Based on the time index, the next step is to synchronize and merge the data. From the liquid temperature values ​​recorded by the thermocouples, the temperature reading closest to each timestamp in the spectral feature set is found. A time alignment tolerance is set, for example, no more than 5 milliseconds. The successfully matched liquid temperature values ​​are extracted. Then, this liquid temperature value synchronized with the acoustic feature time is merged with the subharmonic intensity value and broadband noise floor value extracted from the spectral feature set. Finally, the harmonic energy ratio sequence calculated above is also added to this merged data structure. All feature data are aligned based on the timestamp, forming a structured dataset containing timestamps, harmonic energy ratio sequences, subharmonic intensity values, broadband noise floor values, and liquid temperature values, generating a temperature-acoustic joint feature group.

[0021] Based on the temperature-acoustic joint feature set generated in the previous process, each record contains a series of acoustic and temperature features associated with a specific timestamp. To construct a standardized input that is easy for subsequent model processing, these features need to be standardized in their assembly and annotation. First, a fixed feature arrangement order is determined, which is set as harmonic energy ratio sequence, subharmonic intensity values, broadband noise floor values, and liquid temperature values. For example, if the analysis considers harmonics from the second to the fifth order, then the harmonic energy ratio sequence contains four values. The concatenated vector will have seven dimensions. Next, for each timestamp record in the temperature-acoustic joint feature group, the corresponding feature values ​​are extracted sequentially according to this fixed arrangement, and they are concatenated into a row vector. For example, at time t, if the energy ratios of the second, third, fourth, and fifth harmonics are 0.25, 0.125, 0.08, and 0.05 respectively, the harmonic intensity is 0.02, the broadband noise floor is 0.001, and the liquid temperature is 55℃, then the concatenated vector will be [0.25, ...]. [0.125, 0.08, 0.05, 0.02, 0.001, 55] Subsequently, each dimension of this vector is explicitly labeled. The first dimension is labeled as "second harmonic energy ratio", the second dimension as "third harmonic energy ratio", and so on. The sixth dimension is labeled as "broadband noise floor", and the seventh dimension as "liquid temperature". This labeling ensures the clarity and consistency of the physical meaning of each element in the vector. All time points are recorded in the same way, and finally a matrix composed of multiple time-sequentially arranged feature vectors is formed. Each row vector in this matrix is ​​a snapshot representing the cavitation state at a specific moment, forming a multidimensional cavitation state feature vector.

[0022] The steps to obtain the amplitude probability distribution are as follows: Based on the multidimensional cavitation state feature vector, the amplitude value of the acoustic signal time series is discretized into bins according to equal width intervals, and the frequency of amplitude values ​​is counted for each bin. The frequency of each bin is stored in correspondence with the bin number to generate an amplitude bin count. Based on the amplitude bin count, divide each bin count by the sum of all bin counts while maintaining the bin count order to obtain the probability value of each bin, and combine them to form the amplitude probability distribution.

[0023] Specifically, based on the acoustic signal time series contained in the multidimensional cavitation state feature vector, the amplitude range for discretization and binning is first determined. This involves iterating through all acoustic signal time series amplitude values ​​within the current time window, identifying the maximum and minimum amplitude values, and then determining the number of bins. The number of bins needs to balance the smoothness of the probability distribution with the preservation of detail. An empirical method is based on the total number of data points. For example, for a signal segment containing 8192 data points, the number of bins can be set to 32. This value is chosen to avoid feature blurring due to too few bins and to prevent statistical sparsity problems caused by too many bins. Next, based on the maximum and minimum amplitude values ​​and the set number of bins (32), the width of each bin is calculated. This is done by dividing the difference between the maximum and minimum amplitude values ​​by the number of bins to obtain the width of an equal-width interval. For example, if the minimum amplitude is -1.2V and the maximum amplitude is 1.2V, then the width of each bin is (1.2 - (-1.2)). / 32 = 0.075V. Based on this width value, the boundary of each sub-box is defined starting from the minimum amplitude value. The range of the first sub-box is [-1.2V, -1.125V), the second sub-box is [-1.125V, -1.05V), and so on, until the entire amplitude range is covered. Then, a counting array of length 32 is initialized, with all its elements being zero. Then, each amplitude value in the time series of the sound signal is traversed, the sub-box interval to which it belongs is determined, and the counter of the corresponding sub-box is incremented by one. After traversing all amplitude values, each element in the counting array represents the frequency of the signal point in the corresponding amplitude interval. Finally, this array containing 32 count values ​​is associated with their respective sub-box numbers and stored to generate the amplitude sub-box count.

[0024] Based on the amplitude bin count generated in the previous step, which is an array containing the number of signal points in each bin, the sum of all bin counts is calculated first. This involves accumulating all element values ​​in the amplitude bin count array to obtain a total number of signal points. This total should correspond to the length of the acoustic signal time series used for analysis. For example, if the analyzed signal segment contains 8192 sampling points, the theoretical sum of all bin counts should also be 8192. Next, to obtain the probability value for each bin, the amplitude bin counts need to be normalized. Specifically, this involves iterating through each count value in the amplitude bin count array and dividing that count value by the previously calculated sum of all bin counts. For example, if the count value for the first bin is 150 and the sum of all bin counts is 8192, then the probability value for the first bin is approximately 150 / 8192. 0.0183. Perform the same division operation on the count values ​​of all sub-boxes to obtain a set of probability values ​​corresponding to each sub-box. Throughout the calculation process, the original order of the sub-boxes must be maintained, that is, the first probability value corresponds to the first sub-box, the second probability value corresponds to the second sub-box, and so on, to ensure that the correspondence between the probability distribution and the amplitude interval is not disordered. Finally, combine all the calculated sub-box probability values ​​into an array or list according to their original order. This array is the required amplitude probability distribution. The sum of this distribution should be 1, representing the distribution of the acoustic signal amplitude in different intervals. The array is then combined to form the amplitude probability distribution.

[0025] The steps for obtaining the temperature-acoustic coupling chaos index are as follows: Based on the amplitude probability distribution and the liquid temperature value, the temperature-acoustic coupling chaos index is calculated using the following formula: ; in, The temperature-acoustic coupling chaos index. Let B be the probability of the j-th amplitude bin, where j is the amplitude bin index and B is the number of amplitude bins. This is the real-time liquid temperature value. For the optimal cavitation temperature, The width coefficient is affected by temperature. This is the temperature penalty weighting factor.

[0026] Specifically, the above formula constructs a composite index that can comprehensively evaluate the cavitation state and temperature environment suitability during ultrasonic cleaning by deviating between the Shannon entropy of the coupled acoustic signal and the liquid temperature. The first part of the formula... , which is Shannon entropy in information theory, is used to quantify the randomness or uncertainty of the probability distribution of acoustic signal amplitude. A more uniform amplitude distribution, i.e., a more complex sound field state, corresponds to a higher entropy value, which is usually associated with more active cavitation bubble collapse activity. The second part of the formula... This is a temperature penalty term that introduces the crucial influence of temperature on cavitation efficiency. Centered on the optimal cavitation temperature, a secondary penalty is applied when the actual temperature deviates from the optimal temperature, reducing the calculated exponent value and thus... It can reflect the true cleaning efficiency more accurately than a single indicator.

[0027] The probability of the j-th amplitude bin represents the frequency at which the amplitude value of the acoustic signal time series falls within the j-th preset interval. This parameter is directly obtained from the amplitude probability distribution calculated in the previous step, which is an array containing B probability values. This refers to the j-th element in the array. For example, after completing amplitude bin counting and normalization, a probability array of length 32 is obtained, where the 10th element has a value of 0.085. Therefore, in the calculation... The value is 0.085.

[0028] B represents the number of amplitude bins. This is an integer parameter set during data preprocessing, determining the granularity of discretizing the amplitude values ​​of continuous acoustic signals. The value of B needs to balance computational accuracy and statistical stability. A value that is too small will lead to information loss and an inability to accurately characterize the amplitude distribution. A value that is too large may result in too few sample points in some bins, leading to inaccurate probability estimation and statistical noise. The specific value is usually determined based on the analysis of historical data and experimental verification. For example, for a signal sampling segment with a length of 8192, the following can be calculated: Considering computational efficiency and alignment with powers of 2, B was ultimately set to 16 or 32. In this example, the value of B was set to 32.

[0029] The real-time liquid temperature is measured in degrees Celsius (°C) by thermocouple sensors deployed within the cleaning tank. The thermocouples convert the sensed temperature into an electrical signal, which, after signal conditioning circuitry and an analog-to-digital converter (ADC), is read digitally by the control system. The data acquisition frequency needs to match the analysis frequency of the acoustic signal; for example, the temperature reading is updated once per second. To reduce the impact of measurement noise, the average of multiple continuously acquired temperature readings can be taken as the current value. For example, if five temperature readings were collected consecutively within the last 100 milliseconds, namely 65.1℃, 65.2℃, 65.0℃, 65.2℃, and 65.1℃, then the average value is taken. ℃.

[0030] The optimal cavitation temperature represents the liquid temperature at which the ultrasonic cavitation effect is strongest under a specific combination of cleaning fluid and object being cleaned. It needs to be obtained through experimental calibration. The calibration process typically involves preparing multiple sets of standard contaminant samples (e.g., aluminum foil or ceramic discs coated with standard stains). Under a fixed ultrasonic power, these samples are cleaned in cleaning fluids at different liquid temperatures (e.g., from 30°C to 80°C, with a test point every 5°C). The time required for each sample to reach the predetermined cleanliness level or the mass of contaminants removed within a fixed time is recorded. The cleaning efficiency (e.g., mass removal rate) is plotted against temperature; the temperature corresponding to the peak point of the curve is the optimal cavitation temperature. For example, experiments have shown that the aluminum foil perforation speed is fastest at 60℃, thus determining... ℃.

[0031] The temperature-dependent width factor describes the cleaning efficiency as it deviates from the optimal cavitation temperature. The degree of descent is also determined using the methods described above. The experimental data was used to obtain the cleaning efficiency-temperature curve, and the point where the efficiency dropped to approximately 60.7% of its peak efficiency was found. Calculate the temperature points corresponding to (times) and (times) these temperature points. The average of the differences can be used to obtain... The estimated value, a larger one A lower temperature rating indicates a higher tolerance for temperature changes in the cleaning process, while a lower temperature rating indicates greater sensitivity to temperature changes. For example, if efficiency is highest at 60°C, and drops to about 60% of its peak efficiency at 55°C and 65°C, then... It can be set to 5℃.

[0032] This is the temperature penalty weighting factor, a dimensionless adjustment coefficient used to balance the relative importance of the acoustic entropy term and the temperature penalty term in the formula. Its setting depends on the objective of the control strategy; if the system prioritizes maintaining the temperature at the optimal point, a larger value should be set. The value is such that even a slight deviation in temperature can lead to... A sharp drop in the index, conversely, if the chaotic state of the sound field is considered more important, then... The value can be set smaller, and the specific method for determining its value is as follows: First, analyze the typical range of values ​​for the acoustic entropy term. For a bin with B=32, the maximum value of the entropy is... It typically fluctuates between 2.5 and 4.5, setting a target penalty intensity, for example, when the actual temperature deviates from... one At that distance, I hope The index decreases by 0.5, at which point the temperature penalty term becomes... Therefore, it can be set .

[0033] Calculations based on parameters: The currently acquired parameter values ​​are set as follows: The number of amplitude sub-boxes is B = 32.

[0034] Amplitude probability distribution (To simplify calculations, some probability values ​​are shown here, and for example, the entropy value has already been calculated): The calculated acoustic entropy term .

[0035] Real-time liquid temperature values ℃.

[0036] Optimal cavitation temperature ℃.

[0037] Temperature affects width coefficient ℃.

[0038] Temperature penalty weighting factor .

[0039] Substitute the above parameter values ​​into the formula: ; ; ; ; ; ; The results indicate that, at the current moment, the temperature-acoustic coupling chaos index, which comprehensively considers both acoustic field chaos and temperature suitability, is 3.626. This value is a quantitative and comprehensive indicator for evaluating the cleaning state, and its relative magnitude and trend serve as the basis for subsequent control decisions. For example, if the preset target chaos range is [3.8, 4.2], then the currently calculated value of 3.626 is lower than the lower limit of the target range, indicating that the current cleaning state has not reached its optimal state. It is necessary to adjust the ultrasonic power to increase the index value and bring it into the target range.

[0040] The steps to obtain the chaos deviation value are as follows: Based on the temperature-acoustic coupling chaos index and the upper and lower limits of the target chaos interval, the center point of the interval is calculated and the difference between the temperature-acoustic coupling chaos index and the center point of the interval is obtained. The difference result is used as the chaos deviation value.

[0041] Specifically, based on the temperature-acoustic coupling chaos index calculated in the previous step and the pre-set upper and lower limits of the target chaos range, the source of the target chaos range needs to be clarified first. This range is determined through a series of calibration experiments. Specifically, under standard cleaning load and cleaning fluid conditions, the ultrasonic generator is gradually adjusted from its lowest power to its highest power in 5W increments. At each power set point, after the system stabilizes, the temperature-acoustic coupling chaos index and the corresponding cleaning efficiency are continuously recorded for 5 minutes. The cleaning efficiency is quantified by measuring the removal rate of standard contaminants. All data points were plotted as a scatter plot of cleaning efficiency versus temperature-acoustic coupling chaos index, and curve fitting was performed to find the temperature-acoustic coupling chaos index range corresponding to a cleaning efficiency exceeding 90%. The lower and upper bounds of this range were set as the upper and lower limits of the target chaos interval. For example, experiments determined that the cleaning effect was optimal when the index was between 3.8 and 4.2, so the lower limit of the target chaos interval was 3.8 and the upper limit was 4.2. Next, the center point of this interval was calculated by adding the upper and lower limits and then dividing by 2, i.e., (3.8...). + 4.2) / 2 = 4.0, obtaining the interval center point as 4.0. Then, the temperature-acoustic coupling chaos index calculated in real time, such as 3.626 obtained in the previous step, is subtracted from the interval center point to calculate the difference between the current index and the center point, i.e., 3.626 - 4.0 = -0.374. The sign of this difference result indicates the direction of deviation of the current state from the ideal center. A negative value indicates that it is lower than the ideal center, and a positive value indicates that it is higher than the ideal center. The magnitude of its absolute value quantifies the degree of deviation. Finally, the calculated difference result -0.374 is taken as the chaos deviation value.

[0042] The steps for obtaining the power regulation deviation are as follows: Based on the chaos degree deviation value, extract the chaos degree deviation value at the current time and compare it with the chaos degree deviation values ​​at the previous time and the time before that. At the same time, combine the target chaos degree interval width and the sampling time interval to obtain the chaos degree deviation sequence. The power regulation deviation is calculated based on the chaos degree deviation sequence and the liquid temperature value. The calculation formula is as follows: ; in, Let $\frac{ ... The chaos deviation value at the k-th sampling time. The chaos deviation value at the (k-1)th sampling time. The chaos deviation value at the (k-2)th sampling time. The width of the target chaos interval. Let be the liquid temperature value at the k-th sampling time. The critical protection temperature, It is the temperature decay factor. Based on the gain constant, The sampling time interval, The integral time constant is... is the differential time constant.

[0043] Specifically, based on the currently calculated chaos deviation value, to provide the subsequent controller with dynamic trend information, a chaos deviation sequence containing historical information needs to be constructed. This process is implemented using a time-series data queue of fixed length 3. Such a queue is maintained in memory to store the chaos deviation values ​​at the three most recent sampling times. Whenever a new chaos deviation value is calculated at the current sampling time, it is pushed to the head of the queue. Simultaneously, the oldest value in the queue, the deviation value from the time two times ago, is removed. In this way, the queue always maintains the chaos deviation values ​​at the current time, the previous time, and the time two times ago. For example, if the chaos deviation value is calculated to be -0.374 at time k, the queue is updated, with -0.374 at the head, followed by the value at time k-1 (e.g., -0.350) and the value at time k-2 (e.g., -0.320). Next, these three values ​​in the queue are compared one by one to analyze the trend of the deviation value. The current value is compared with the value at the previous time; -0.374 < -0.350 indicates that the deviation is increasing in the negative direction, meaning that the system state is further deviating from the target center point. Comparing the values ​​of the previous two time points, -0.350 < -0.320, indicating that this trend of increasing deviation has continued for some time. This comparison reveals the first and second rates of change of the error, which are the basis for subsequent differential control term calculations. At the same time, the target chaos degree interval width, such as the previously calculated 0.4, and the sampling time interval of the control system, such as setting it to 0.5 seconds, are packaged together with this time series containing three deviation values ​​to form a complete data structure, resulting in the chaos degree deviation sequence.

[0044] The power regulation deviation calculation formula constructs an incremental PID controller with a temperature safety protection mechanism to calculate the adjustment of ultrasonic power at each sampling moment. The core of the formula is a digital PID control algorithm, in which... This represents the proportional term (response to the rate of change of error). This represents the integral term (the response to accumulated error), while These are the differential terms in the second-order difference form (predicting the trend of error change). The combination of these three terms enables rapid, stable, and error-free adjustment of the deviation. The two adjustment factors introduced are, firstly, the normalized gain. , base gain Use the width of the target chaos interval Normalization is performed so that the controller's sensitivity can adapt to the width of the target range; the second is a temperature protection factor. Utilizing the properties of the Sigmoid function, when the liquid temperature... Far below the critical protection temperature When the temperature is close to or exceeds the critical value, the factor approaches 1 and the controller works normally. When the temperature approaches or exceeds the critical value, the factor quickly approaches 0, thus smoothly attenuating or even cutting off the power regulation output to avoid damage to the equipment due to overheating.

[0045] , , These represent the chaos degree deviation values ​​at sampling times k, k-1, and k-2, respectively. These values ​​are dimensionless and are directly provided by the chaos degree deviation sequence constructed in the previous step. They are the core inputs of the controller, reflecting the current state, past states, and trends of the system. A new chaos degree deviation sequence is calculated for each control cycle. And update the stored historical values, for example, at the current time k, by calculation. At the same time, the deviation value of the previous moment is retrieved from the historical records. Deviation from the previous time point This data shows that the chaos index is consistently below the target center, and the degree of deviation is increasing.

[0046] The target chaos interval width is a dimensionless parameter representing the allowable fluctuation range of the chaos index corresponding to the optimal cleaning effect. It is obtained by subtracting the lower limit from the upper limit of the target chaos interval. Its value is set based on previous experimental calibration, such as by plotting the relationship curve between cleaning efficiency and chaos index to determine the boundary of the efficient cleaning interval, which is a relatively wide value. This means the system has a high tolerance for state fluctuations, allowing for relatively mild control; conversely, a lower tolerance requires more precise control. In this example, the target interval is [3.8, 4.2], therefore... .

[0047] The value of the liquid temperature at the k-th sampling time is expressed in degrees Celsius (°C). This data is collected in real time by thermocouple sensors deployed in the cleaning tank and transmitted to the controller through the data acquisition system. It is the direct input to the temperature protection mechanism and reflects the real-time thermal state of the cleaning process. In this example, the temperature value obtained in the previous steps is used. ℃.

[0048] The critical protection temperature, expressed in degrees Celsius (°C), is a preset, fixed safety threshold. It is set based on the physicochemical properties of the cleaning solution (such as boiling point, flash point, and decomposition temperature) and the heat resistance of the workpiece being cleaned. This ensures that the temperature of the cleaning tank will not exceed a level that would pose a danger to the equipment or workpiece under any circumstances. For example, for water-based cleaning agents, considering evaporation and safety margins, this value is typically set between 85°C and 90°C. In this example, it is set... ℃.

[0049] This is the temperature decay factor, expressed in units of 1 / ℃. This parameter is used to adjust the severity of the temperature protection mechanism's response, i.e., to control the steepness of the Sigmoid function curve. A larger value... A certain value will cause the power regulation to decay more rapidly as the temperature approaches the critical point, resulting in stronger protection but potentially being overly conservative; conversely, a lower value will result in a smoother response. The value is determined by working backwards from the decay rate set at a specific temperature point. For example, the controller gain might be set to decay to 50% when the actual temperature reaches the critical temperature, and to decay to approximately 5% when the temperature exceeds the critical point by 3°C. Calculations are then performed based on this requirement: when... The attenuation factor is Solving for To facilitate rounding, set / ℃.

[0050] The base gain constant, measured in watts (W), is the overall gain of the PID controller. It determines the overall response strength of the controller to deviations. Its value needs to be tuned through system debugging. Common methods include the Ziegler-Nichols method, which aims to achieve the fastest possible response speed and the smallest overshoot while ensuring system stability. For example, in practical systems, this can be achieved by gradually increasing the base gain constant from small to large. Observe the step response of the system until the system reaches a critical oscillation, record the gain and oscillation period at this point, and then calculate the appropriate value using the Ziegler-Nichols method. The value, in this example, was determined through debugging. W.

[0051] The sampling time interval, measured in seconds (s), is determined by the system's hardware performance and software processing speed. It represents the cycle of one complete calculation and adjustment by the controller and is a fixed system parameter. In this example, it is set as follows: s.

[0052] This is the integration time constant, measured in seconds (s), which affects the weight of the integration term and is mainly used to eliminate the steady-state error of the system. The smaller the value, the stronger the integral action and the faster the elimination of steady-state error. However, too small a value may lead to integral saturation and system oscillation. Its tuning is usually related to... and This can be done in a coordinated manner, and can be set to a certain proportion of the critical oscillation period, such as 0.5 times. In this example, this can be set through debugging. s.

[0053] The differential time constant, measured in seconds (s), affects the weight of the differential term. It is used to predict the trend of error changes and provides damping to suppress overshoot and oscillations. The larger the value, the stronger the differential action and the better the system stability, but it also becomes more sensitive to noise. In this example, the settings are adjusted accordingly. s.

[0054] Calculations based on parameters: The currently acquired parameter values ​​are set as follows: , , .

[0055] W, .

[0056] ℃, ℃, / ℃.

[0057] s, s, s.

[0058] Calculate the PID control part: Proportional contribution: ; Integral contribution: ; Differential contribution: ; Sum of PID terms: ; Calculate the temperature protection factor: ; Calculate the total power regulation deviation: ; ; W; The results show that at the k-th sampling time, the controller calculates a power adjustment deviation of +14.3125 watts. This positive value indicates that the controller recommends increasing the current output power of the ultrasonic generator by 14.3125 W. This decision is based on a comprehensive analysis of the chaos deviation: although the deviation itself (integral term) and its rate of change (proportional term) both point to the need to increase power to correct the negative deviation, the acceleration of its change (differential term) is very small, indicating that the system tends to stabilize. The final adjustment is a precise value that integrates the current error, cumulative error, and future trend. Since the current temperature is far below the critical value, the temperature protection mechanism has not been activated, and the adjustment has not been attenuated. This will serve as the direct basis for updating the power setting value in the next step.

[0059] The steps to obtain the updated power setting value are as follows: Based on the power adjustment deviation and the output power setting value, the sampling time is aligned and the units are checked to be consistent. The algebraic signs are used to perform addition and the accumulation result is recorded. If it exceeds the allowable range of the output power setting value, the boundary value is used to replace it and an updated power setting value is generated.

[0060] Specifically, based on the power adjustment deviation calculated in the previous step and the currently stored output power setting, data alignment and unit verification are first performed to confirm that the power adjustment deviation, for example +14.3125W, and the current output power setting, for example 250.0W, are both in watts (W) and correspond to the same control sampling time k. Then, these two values ​​are added using algebraic notation to calculate the initial updated power value, i.e., 250.0W + 14.3125W. =264.3125W, and this accumulated result is temporarily recorded. Next, a boundary check is performed on this accumulated result, comparing it with the preset allowable range of output power settings. This allowable range is jointly determined by the hardware specifications and safe operating procedures of the ultrasonic generator. For example, for a generator with a rated power of 500W, its safe and stable operating range is set to 50W to 500W. The lower limit of this range, 50W, is to avoid the generator from operating unstablely or failing to effectively excite the transducer at too low a power, while the upper limit, 500W, is the maximum power that the hardware can handle. The calculated accumulated result 264.3125W is compared with this range [50W, 500W]. Since 264... The 0.3125W value falls within this range, so no adjustment is needed. If, in a certain calculation cycle, the current power setting is 495.0W and the calculated power adjustment deviation is +10.5W, the accumulated result is 505.5W. This value exceeds the upper limit of 500W, so the accumulated result will be replaced with the boundary value of 500W. Conversely, if the current power setting is 55.0W and the calculated adjustment deviation is -8.2W, the accumulated result is 46.8W, which is lower than the lower limit of 50W. In this case, the accumulated result will be replaced with the boundary value of 50W. After completing the boundary check and possible replacement operations, the final value obtained is the power command finally adopted in this control cycle, and the updated power setting is generated.

[0061] The steps for obtaining the closed-loop power control signal are as follows: Based on the updated power setting, the rated power range of the ultrasonic transducer is mapped to the control range. The pulse width modulation mode or voltage control mode is selected, the duty cycle or voltage amplitude of the corresponding pulse width modulation signal is calculated and quantized by a counter to obtain the duty cycle or voltage amplitude of the pulse width modulation signal. Based on the duty cycle or voltage amplitude of the pulse width modulation signal, combined with the carrier frequency, drive polarity, dead time and enable state, frame encapsulation and timing alignment are performed, and the data is written into the ultrasonic transducer drive register to generate a closed-loop power control signal.

[0062] Specifically, based on the updated power setpoint generated in the previous process, it needs to be converted into a physical control signal that the underlying hardware can recognize. First, the specific implementation method of power control is determined; here, pulse width modulation (PWM) mode is selected. This mode controls the average power output to the transducer by adjusting the duty cycle of the drive signal. Next, a linear mapping from power to control quantity is performed. A correspondence needs to be established between the updated power setpoint (physical quantity) and the duty cycle of the pulse width modulation signal (control quantity). This mapping is based on the rated power range of the ultrasonic transducer and the counter range of the PWM timer in the drive circuit. For example, if the effective power range of the transducer is known to be 50W to 500W, and the control core, such as a microcontroller, has a 12-bit precision PWM timer with a corresponding counter quantization range of 0 to 4095, then a linear interpolation formula is used to calculate the PWM counter value corresponding to the updated power setpoint. The calculation formula is: Counter value = ((Current power value - Minimum power value) / (Maximum power value - Minimum power value)) × (Maximum count value - Minimum count value) + The minimum count value is calculated by substituting the updated power setting value of 264.3125W obtained in the previous step. The minimum power value is 50W, the maximum power value is 500W, the minimum count value is 0, and the maximum count value is 4095. The counter value is calculated as follows: Counter value = ((264.3125 - 50) / (500 - 50)) × 4095 = (214.3125 / 450) ×4095 ≈ 0.47625 × 4095 ≈ 1950.24375. Finally, the calculated floating-point result is quantized by the counter. Since the PWM timer register can only accept integers, 1950.24375 needs to be converted to an integer. Here, rounding is used to obtain 1951. This integer value is the value that will be written to the hardware timer compare register to obtain the duty cycle or voltage amplitude of the pulse width modulation signal.

[0063] Based on the counter value corresponding to the duty cycle of the pulse width modulation signal calculated and quantized in the previous step, such as 1951, a series of drive parameters are needed to configure the hardware drive circuit to generate the final electrical signal applied to the transducer. First, the carrier frequency of the PWM signal is set. This frequency determines the switching speed of the power switching device and is usually selected based on the design of the drive circuit and the characteristics of the switching device (e.g., MOSFET). A typical setting is 100kHz to ensure smooth power regulation while keeping switching losses within a reasonable range. Second, the drive polarity is set. Depending on the design of the drive circuit (e.g., full-bridge or half-bridge circuit), it is determined whether the PWM signal is active high or active low. Here, it is set to active high. Next, the dead time is configured. This is to prevent the switches in the upper and lower arms of the bridge circuit from conducting simultaneously and causing a short circuit. A small delay must be inserted between turning off one switch and turning on another. The setting of this time is based on the switching transistors. The turn-off and turn-on delay parameters were determined by consulting the device datasheet and allowing for a safety margin. The dead time was set to 500 nanoseconds. Then, the enable state of the driver was set, which determines whether the driver output is on or off. During normal operation, this state bit is set to enable. After setting all parameters, these parameter values ​​were frame-encapsulated. This involves organizing the counter value 1951 corresponding to the duty cycle of the pulse width modulation signal, the configuration word for the carrier frequency of 100kHz, the setting of the driver polarity, the configuration word for the dead time of 500 nanoseconds, and the enable state bit into one or more data words to be written, according to the address and bit field definition of the microcontroller driver register. Timing alignment was then performed to ensure that at a safe update point in the PWM timer cycle, such as when the counter returns to zero, these data words are written to the corresponding address of the ultrasonic transducer driver register through one or more bus write operations. After the hardware receives these new configurations, its internal logic will immediately run according to the new parameters to generate a closed-loop power control signal.

[0064] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for monitoring and adaptively adjusting the power of ultrasonic cleaning temperature, characterized in that, Includes the following steps: Based on the acoustic signal collected by the high-bandwidth hydrophone and the liquid temperature sensed by the thermocouple, a discrete Fourier transform is performed on the time series of the acoustic signal to establish a multidimensional cavitation state feature vector. Based on the multidimensional cavitation state feature vector, the subharmonic intensity, harmonic energy ratio, and broadband noise level are used as coordinates to determine the position of the current operating point in the cavitation state space. At the same time, the amplitude value of the acoustic signal time series is discretized into bins, the frequency of the signal points in each bin is counted, and an amplitude probability distribution is constructed. The amplitude probability distribution is substituted into the Shannon entropy formula to calculate the information entropy, and the information entropy value is corrected by combining the liquid temperature to obtain the temperature-acoustic coupling chaos index. Based on the temperature-acoustic coupling chaos index, a target chaos range is set. The current temperature-acoustic coupling chaos index is compared with the upper and lower limits of the target chaos range to obtain the chaos deviation value. Based on the chaos deviation value, a power adjustment deviation is established. Based on the power adjustment deviation, the power adjustment deviation is calculated with the current output power setting value of the ultrasonic generator to obtain an updated power setting value. The updated power setting value is then converted into the duty cycle or voltage amplitude of the pulse width modulation signal driving the ultrasonic transducer to obtain a closed-loop power control signal.

2. The ultrasonic cleaning temperature monitoring and power adaptive adjustment control method according to claim 1, characterized in that, The steps for obtaining the multidimensional cavitation state feature vector are as follows: Based on the acoustic signal time series obtained by the high-bandwidth hydrophone and the liquid temperature value recorded by the thermocouple, the acoustic signal time series is decomposed in the frequency domain and the energy is accumulated band by band to obtain the fundamental frequency energy value, the energy values ​​of each harmonic, the intensity values ​​of each harmonic and the broadband noise floor value, and generate a set of spectral features. Based on the spectral feature set, the ratio of each harmonic energy value to the fundamental frequency energy value is calculated one by one while keeping the time index consistent. The liquid temperature value, the harmonic intensity value, and the broadband noise floor value are synchronously merged to generate a temperature acoustic joint feature set. Based on the aforementioned temperature-acoustic joint feature set, the feature vectors are spliced ​​together in a fixed order of ratio sequence, subharmonic intensity value, broadband noise floor value and liquid temperature value, and labeled with dimension tags to form a multidimensional cavitation state feature vector.

3. The ultrasonic cleaning temperature monitoring and power adaptive adjustment control method according to claim 1, characterized in that, The steps for obtaining the amplitude probability distribution are as follows: Based on the multidimensional cavitation state feature vector, the amplitude value of the acoustic signal time series is discretized into bins according to equal width intervals, and the frequency of amplitude values ​​is counted for each bin. The frequency of each bin is stored in correspondence with the bin number to generate an amplitude bin count. Based on the amplitude bin count, each bin count is divided by the sum of all bin counts while maintaining the bin count order to obtain the probability value of each bin, which is then combined to form the amplitude probability distribution.

4. The ultrasonic cleaning temperature monitoring and power adaptive adjustment control method according to claim 1, characterized in that, The steps for obtaining the temperature-acoustic coupling chaos index are as follows: The temperature-acoustic coupling chaos index is calculated based on the amplitude probability distribution and the liquid temperature value.

5. The ultrasonic cleaning temperature monitoring and power adaptive adjustment control method according to claim 1, characterized in that, The steps for obtaining the chaos degree deviation value are as follows: Based on the temperature-acoustic coupling chaos index and the upper and lower limits of the target chaos interval, the center point of the interval is calculated and the difference between the temperature-acoustic coupling chaos index and the center point of the interval is obtained. The difference result is used as the chaos deviation value.

6. The ultrasonic cleaning temperature monitoring and power adaptive adjustment control method according to claim 1, characterized in that, The steps for obtaining the power adjustment deviation are as follows: Based on the chaos degree deviation value, extract the chaos degree deviation value at the current time and compare it with the chaos degree deviation values ​​at the previous time and the time before that. At the same time, combine the target chaos degree interval width and the sampling time interval to obtain the chaos degree deviation sequence. The power regulation deviation is calculated based on the chaos degree deviation sequence and the liquid temperature value.

7. The ultrasonic cleaning temperature monitoring and power adaptive adjustment control method according to claim 1, characterized in that, The steps for obtaining the updated power setting value are as follows: Based on the power adjustment deviation and the output power setting value, the units are aligned according to the sampling time and checked for consistency. The values ​​are then added using algebraic signs and the results are recorded. If the values ​​exceed the allowable range of the output power setting value, they are replaced with boundary values ​​to generate an updated power setting value.

8. The ultrasonic cleaning temperature monitoring and power adaptive adjustment control method according to claim 1, characterized in that, The steps for obtaining the closed-loop power control signal are as follows: Based on the updated power setting value, the rated power range of the ultrasonic transducer is mapped to the control range. The pulse width modulation mode or voltage control mode is selected, the duty cycle or voltage amplitude of the corresponding pulse width modulation signal is calculated and quantized by a counter to obtain the duty cycle or voltage amplitude of the pulse width modulation signal. Based on the duty cycle or voltage amplitude of the pulse width modulation signal, combined with the carrier frequency, drive polarity, dead time and enable state, frame encapsulation and timing alignment are performed, and the data is written into the ultrasonic transducer drive register to generate a closed-loop power control signal.

9. The adjustment and control system of the ultrasonic cleaning temperature monitoring and power adaptive adjustment control method according to any one of claims 1-8, characterized in that, include: The signal processing module is used to perform discrete Fourier transform on the time series of the acoustic signal based on the acoustic signal collected by the high-bandwidth hydrophone and the liquid temperature sensed by the thermocouple, and to establish a multidimensional cavitation state feature vector. The feature analysis module is used to determine the position of the current operating point in the cavitation state space based on the multidimensional cavitation state feature vector, using subharmonic intensity, harmonic energy ratio and broadband noise level as coordinates. At the same time, it discretizes the amplitude value of the acoustic signal time series into bins, counts the frequency of signal points in each bin, constructs an amplitude probability distribution, substitutes the amplitude probability distribution into the Shannon entropy formula to calculate the information entropy, and corrects the information entropy value in combination with liquid temperature to obtain the temperature-acoustic coupling chaos index. The chaos determination module is used to set a target chaos range based on the temperature-acoustic coupling chaos index, compare the current temperature-acoustic coupling chaos index with the upper and lower limits of the target chaos range to obtain the chaos deviation value, and establish the power adjustment deviation based on the chaos deviation value. The power control module is used to calculate the power adjustment deviation and the current output power setting value of the ultrasonic generator based on the power adjustment deviation, obtain the updated power setting value, and convert the updated power setting value into the duty cycle or voltage amplitude of the pulse width modulation signal driving the ultrasonic transducer to obtain a closed-loop power control signal.

Images