Asynchronous motor time domain waveform transient interval identification method based on peak detection

Abstract

The invention provides an asynchronous motor time domain waveform transient interval identification method based on peak detection, and the method comprises the steps: carrying out the fast Fourier transform of the original time domain waveform of a three-phase current, and obtaining the frequency and amplitude spectrum of a signal; determining a dominant frequency and obtaining the point number of each period; identifying all peak values in the time domain waveform, and setting a distance threshold value and a height threshold value; calculating the average amplitude of all the peak values through the amplitude; judging whether the mean value of the three cycles corresponding to the current peak value position is in an expected cycle point interval or not, judging whether the absolute value of the average amplitude is smaller than a set threshold value or not, and determining a start-up ending position; and the current data of each phase are integrated, the maximum current value in the starting process is found out, the total duration from the starting point to the starting end point is calculated, and the starting performance of the motor is comprehensively evaluated. According to the method, the transient interval in the motor starting process can be accurately identified, and more accurate data support is provided for motor performance evaluation and maintenance strategies.

Description

Technical Field

[0001] The present invention relates to the technical field of motor control, and particularly relates to a method for identifying the transient interval of the time-domain waveform of an asynchronous motor based on peak detection. Background Art

[0002] The existing methods for identifying the transient interval of the time-domain waveform mainly include the following different techniques and strategies:

[0003] 1. Fixed threshold method: A fixed threshold is set to distinguish the transient interval and the steady-state interval, and there are the following disadvantages:

[0004] (1) It is not applicable to different types of motors or different load conditions because the threshold needs to be adjusted according to specific situations.

[0005] (2) It cannot adapt to the non-linear changes during the motor startup process, especially in variable-frequency motors.

[0006] 2. Fourier transform-based method: The Fourier transform is used to analyze the spectral characteristics of the signal to identify the transient interval, and there are the following disadvantages:

[0007] (1) The Fourier transform assumes that the signal is periodic, which is not applicable in the case of non-periodic or transient signals.

[0008] (2) For variable-frequency motors, since the frequency is changing, this method cannot accurately capture the details of the frequency change.

[0009] 3. Wavelet transform-based method: The wavelet transform is used to analyze the time-frequency characteristics of the signal, and there are the following disadvantages:

[0010] (1) Although the wavelet transform can provide good time-frequency resolution, it is affected by noise in some cases.

[0011] (2) It is relatively complex to select appropriate wavelet basis functions and scale parameters.

[0012] 4. Machine learning-based method: Machine learning models (such as support vector machines, neural networks, etc.) are used to train classifiers to identify the transient interval, and there are the following disadvantages:

[0013] (1) A large amount of labeled training data is required to build an accurate model.

[0014] (2) When there is not enough data set to cover various situations, the model generalization ability is limited.

[0015] 5. Signal processing technology-based method: It includes techniques such as filter design, moving average, autocorrelation, etc., and there are the following disadvantages:

[0016] (1) These techniques require manual adjustment of parameters to adapt to different motor types and operating conditions.

[0017] (2) In a noisy environment, simple filtering techniques may not be able to effectively remove interference.

[0018] In summary, although existing time-domain waveform transient interval identification methods can identify the transient intervals during the motor startup process to a certain extent, they generally have the following problems:

[0019] Lack of generality: Most methods need to be adjusted for specific motor types and operating conditions.

[0020] Poor robustness: In the face of noise, non-linear changes or complex waveforms, existing methods may not be robust enough.

[0021] Limited accuracy: Especially in terms of non-linear changes and frequency changes during the motor startup process, existing methods may not be able to accurately capture key information.

[0022] Lack of flexibility: The methods may be too dependent on specific parameter settings or model selections and are difficult to adapt to a wide range of motor application scenarios.

[0023] Based on the above situation, it can be seen that there is currently a lack of effective methods for accurately determining the transient intervals of motor time-domain waveforms, which means that existing technologies cannot fully meet the needs of precise analysis. In addition, for variable-frequency motors, since their operating frequencies are not fixed but dynamically adjusted according to actual needs, this leads to frequency changes during the motor startup process. Traditional analysis methods often have difficulty accurately capturing these changes, especially during the transition stage from motor startup to stable operation, and cannot effectively identify the transient intervals during the motor startup process, thus limiting the comprehensive evaluation of motor startup performance and electrical health. Summary of the Invention

[0024] In order to solve the problem that there is a lack of effective methods for accurately determining the transient intervals of motor time-domain waveforms in the existing technology and cannot fully meet the needs of precise analysis, the purpose of the present invention is to provide a method for identifying the transient intervals of asynchronous motor time-domain waveforms based on peak detection. This method uses local maximum detection, distance threshold, height threshold and significance test to accurately find the peaks of the current time-domain waveform, and combines period and amplitude analysis to accurately identify the transient intervals during the motor startup process, thereby providing more accurate data support for motor performance evaluation and maintenance strategies.

[0025] The present invention achieves the above purpose through the following technical solutions:

[0026] A method for identifying the transient intervals of asynchronous motor time-domain waveforms based on peak detection, comprising the following steps:

[0027] Obtain the three-phase current of the asynchronous motor;

[0028] Perform fast Fourier transform on the original time-domain waveforms of the three-phase current respectively to obtain the frequency and amplitude spectrum of the signal; determine the main frequency and obtain the number of points in each period;

[0029] Use the local maximum detection algorithm to identify all the peaks in the time-domain waveform, and set the distance threshold and height threshold;

[0030] By calculating the position difference between adjacent peaks, obtain the number of periods between two adjacent peaks; find the amplitude of each peak, and calculate the average amplitude of all peaks through the amplitude;

[0031] Start traversing from the position of the first peak in the time-domain waveform, and judge whether the average value of the number of periods at the current peak position and its two adjacent periods before and after is within the expected range of the number of points in a period, and judge whether the absolute value of the average amplitude is less than the set threshold. If these two conditions are met, it is considered that the startup is completed, and the current peak position is used as the position where the startup ends;

[0032] Set the current threshold, and in the three-phase current data, find the first point that exceeds this current threshold as the starting moment of startup, ensuring that the analysis range covers the complete startup process from the starting moment to the startup end position;

[0033] Integrate the three-phase current data, find the maximum current value during the startup process, and calculate the total duration from the startup start point to the startup end point to comprehensively evaluate the motor startup performance.

[0034] According to a transient interval identification method for the time-domain waveform of an asynchronous motor based on peak detection provided by the present invention, when determining the main frequency, select the frequency point corresponding to the maximum amplitude between 10 - 100 Hz as the main frequency FL, and divide the sampling frequency fs by the main frequency to obtain the number of points Ncycle in each period, expressed as the following formula:

[0035]

[0036] Among them, fs represents the sampling frequency, and FL represents the main frequency calculated by FFT.

[0037] According to a transient interval identification method for the time-domain waveform of an asynchronous motor based on peak detection provided by the present invention, the local maximum detection algorithm includes:

[0038] Set an empty array or list to store the detected local maxima;

[0039] Start traversing from the first point of the one-dimensional signal, and traverse one by one to the last point;

[0040] For each point, determine whether the point is larger than the points on its left and right sides;

[0041] If the current point is larger than the points on its left and right sides, then consider the point as a local maximum, and add the local maximum to the previously initialized array. Among them, the array storing the local maxima will contain all detected potential peak points.

[0042] According to a transient interval recognition method for the time-domain waveform of an induction motor based on peak detection provided by the present invention, the setting of the distance threshold and the height threshold includes:

[0043] Set the minimum horizontal distance between adjacent peaks to ensure that there are at least distance sampling points between all adjacent peaks. If the distance between two peaks is less than distance, then remove the lower one. Assume that d represents the distance between adjacent peaks, then d≥distance;

[0044] Set the peak height condition height to determine whether the height of the current peak exceeds the peak height condition.

[0045] According to a transient interval recognition method for the time-domain waveform of an induction motor based on peak detection provided by the present invention, set the threshold parameter to specify the minimum threshold on both sides of the peak. This threshold refers to the minimum value when the signal value drops from the peak to a certain point, and its calculation process is as follows:

[0046] First, for each peak, calculate the values at which the signal values adjacent to its left and right sides drop to the lowest point. For each peak, compare whether the thresholds on both sides are within the range of the threshold parameter;

[0047] Retain the peaks that meet the conditions.

[0048] According to a transient interval recognition method for the time-domain waveform of an induction motor based on peak detection provided by the present invention, calculate the significance of the value, that is, the height of the peak relative to the baseline around it, and its calculation process is as follows:

[0049] First, determine the baseline points. For each potential peak point, search to the left and right to find the first local minimum point that no longer rises above the height of the peak; these points are the left baseline point and the right baseline point. If such a point cannot be found on one side, then the baseline point can be considered as the boundary point of the signal;

[0050] For each peak, calculate the difference between its height and the height of the left baseline point to obtain the left significance; then calculate the difference between its height and the height of the right baseline point to obtain the right significance.

[0051] A transient interval recognition method for the time-domain waveform of an asynchronous motor based on peak detection provided by the present invention, calculating the average amplitude of all peaks, which is expressed by the following formula:

[0052]

[0053] amp (amplitude) = Pvalue (peak value) - Mvalue (average current)

[0054]

[0055] Where, Ncycle represents the number of points per cycle, Mvalue represents the average current, amp represents the amplitude, Pvalue represents the calculated current peak sequence, Pi represents the i-th current value between two adjacent peaks, and Mamp represents the average amplitude of the current values within two adjacent cycles.

[0056] A transient interval recognition method for the time-domain waveform of an asynchronous motor based on peak detection provided by the present invention, determining whether the average value of the current peak position cycle number and the two cycle numbers before and after it is within the interval [Ncycle - 1, Ncycle + 1], which is expressed by the following formula:

[0057]

[0058] Where, MCnumber represents the average number of points of three adjacent consecutive cycles, Current Cycle Number represents the number of points of the current cycle, Previous Cycle Number represents the number of points of the previous cycle, and Next Cycle Number represents the number of points of the next cycle.

[0059] A transient interval recognition method for the time-domain waveform of an asynchronous motor based on peak detection provided by the present invention, selecting three adjacent consecutive cycle numbers to calculate the average value, and determining whether the average value falls within an allowable error range ([Ncycle - 1, Ncycle + 1]) of the motor stable state cycle. If it falls within this interval, it indicates that the motor has approached or reached the stable state;

[0060] Where, Ncycle represents the number of cycle points in the stable state of the motor.

[0061] A transient interval recognition method for the time-domain waveform of an asynchronous motor based on peak detection provided by the present invention, when determining whether the absolute value of the average amplitude is less than the set threshold, if the absolute value of the average amplitude is less than the set threshold, the motor startup process has been completed and it enters the stable operation state, where the set threshold is greater than 0.

[0062] It can be seen that, compared with the prior art, the method provided by the present invention can accurately identify the transient interval during the motor startup process, improve the accuracy of motor performance evaluation and maintenance strategies, and has the following beneficial effects:

[0063] 1. High efficiency:

[0064] Fast Fourier Transform (FFT): Through the fast Fourier transform, the present invention can accurately determine the main frequency during the motor startup process within a short time, and thus calculate the number of points in each period. This efficient spectrum analysis method enables the entire process to be completed quickly, improving the overall efficiency.

[0065] Peak detection algorithm: The present invention adopts a local maximum detection algorithm. By setting reasonable distance thresholds and height thresholds, it can quickly and accurately detect the peaks in the time-domain waveform. This efficient peak detection mechanism reduces unnecessary calculations and accelerates the entire analysis process.

[0066] 2. High robustness:

[0067] Adapt to variable-frequency motors: The present invention determines the main frequency through FFT and can adapt to the frequency changes during the startup process of variable-frequency motors, ensuring the accuracy of the analysis.

[0068] Flexible peak detection: The distance thresholds and height thresholds set by the peak detection algorithm of the present invention can ensure accurate peak detection even under large frequency changes, excluding the influence of noise and enhancing the robustness of the algorithm.

[0069] Determination of the startup end point: The present invention determines the end point of startup through the mean number of cycles and the absolute value of the average amplitude. This method can well adapt to the influence brought by frequency changes during the motor startup process, ensuring the accurate judgment of the startup end point.

[0070] 3. Easy operation:

[0071] Intuitive parameter setting: The setting of the distance threshold and height threshold is intuitive and simple, easy to understand and adjust, enabling non-professional technical personnel to operate easily.

[0072] High degree of automation: The present invention realizes automation in the whole process from raw signal acquisition to startup process feature extraction, reducing the operation difficulty and the need for human intervention.

[0073] Clear determination of startup and end points: The present invention automatically identifies the startup moment of the machine through the current threshold, and determines the end point of startup through the number of cycles and amplitude, making the whole process easy to understand and execute.

[0074] 4. High generalization

[0075] Wide applicability: The present invention is applicable not only to fixed-frequency motors but also particularly to variable-frequency motors, and can cope with different frequency change situations.

[0076] Multi-scenario verification: The present invention is verified by testing at different main frequencies, including 50Hz / 36Hz / 30Hz / 20Hz, etc., which proves the effectiveness of the algorithm under various motor types and operating conditions.

[0077] Enhanced robustness: The present invention has been tested in a variety of application scenarios, ensuring the robustness of the algorithm in complex motor starting scenarios and enhancing its applicability and generalization ability in different environments.

[0078] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. Description of the drawings

[0079] Figure 1 It is a flowchart of an embodiment of a method for identifying the transient interval of the time-domain waveform of an asynchronous motor based on peak detection according to the present invention.

[0080] Figure 2 It is a schematic flowchart of peak detection in an embodiment of a method for identifying the transient interval of the time-domain waveform of an asynchronous motor based on peak detection according to the present invention.

[0081] Figure 3 It is a schematic flowchart of the starting current characteristic analysis in an embodiment of a method for identifying the transient interval of the time-domain waveform of an asynchronous motor based on peak detection according to the present invention.

[0082] Figure 4 It is a partially enlarged schematic diagram of the time-domain waveform of a fixed-frequency motor and its peak detection result in an embodiment of a method for identifying the transient interval of the time-domain waveform of an asynchronous motor based on peak detection according to the present invention.

[0083] Figure 5 It is a schematic diagram of the time-domain waveform of a variable-frequency motor and its peak detection result in an embodiment of a method for identifying the transient interval of the time-domain waveform of an asynchronous motor based on peak detection according to the present invention.

[0084] Figure 6 It is a partially enlarged schematic diagram of the time-domain waveform of a variable-frequency motor and its peak detection result in an embodiment of a method for identifying the transient interval of the time-domain waveform of an asynchronous motor based on peak detection according to the present invention. Specific embodiments

[0085] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0086] Reference to "embodiment" herein means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0087] This embodiment discloses a method for identifying the transient interval of the time-domain waveform of an asynchronous motor based on peak detection. By using local maximum detection, distance threshold, height threshold, and significance test to accurately find the peaks of the current time-domain waveform, and combining period and amplitude analysis, it can accurately identify the transient interval during the motor startup process, thereby providing more accurate data support for motor performance evaluation and maintenance strategies.

[0088] During the motor startup process, the significant increase in the startup current reflects the sharp increase in power demand when the motor switches from the stationary state to the rotating state. This current peak not only reflects the transient characteristics of the motor during startup but also is directly related to the startup efficiency and electrical health of the motor. As the motor approaches the synchronous speed, the startup current gradually returns to normal, indicating that the motor enters the stable operation state. The smoothness of this transition stage and the final stable current level are key indicators for evaluating the startup performance and electrical health of the motor.

[0089] During this process of current distortion, it is particularly important to accurately identify the transient interval. The transient interval refers to the transition stage of the motor from startup to the stable operation state. The current change in this stage usually exhibits non-linearity and uncertainty. The transient interval during the motor startup process contains rich information, such as current peaks and current change rates. These information are crucial for diagnosing abnormal conditions during the motor startup process. And accurately identifying the transient interval helps to achieve the following goals: improving startup efficiency, ensuring electrical safety, and evaluating the health status of the motor.

[0090] See Figures 1 to 6 , this embodiment provides a method for identifying the transient interval of the time-domain waveform of an asynchronous motor based on peak detection. The method includes the following steps:

[0091] Step S1, obtain the three-phase current of the asynchronous motor;

[0092] Step S2: Perform fast Fourier transform on the original time-domain waveforms of the three-phase currents respectively to obtain the frequency and amplitude spectra of the signals; determine the main frequency and the number of points in each period.

[0093] Step S3: Use the local maximum detection algorithm to identify all the peaks in the time-domain waveform, and set a distance threshold (distance) and a height threshold (height) to ensure the minimum horizontal distance between any two adjacent peaks and exclude the smaller peaks that may be caused by noise.

[0094] Step S4: Calculate the number of periods between adjacent peaks by calculating the position difference between adjacent peaks; find the amplitude of each peak and calculate the average amplitude of all peaks through the amplitudes.

[0095] Step S5: Start traversing from the position of the first peak in the time-domain waveform, and judge whether the average value of the number of periods at the current peak position and the two periods before and after it is within the expected range of the number of points in a period, and judge whether the absolute value of the average amplitude is less than the set threshold. If these two conditions are met, it is considered that the startup is completed, and the current peak position is used as the position where the startup ends.

[0096] Step S6: Set a current threshold (20% of the rated current) to automatically identify the startup moment of the motor and ensure that the analysis range covers the entire startup process. Among them, in the three-phase current data, find the first point that exceeds this current threshold as the startup moment of the motor, and ensure that the analysis range covers the entire startup process from the startup moment to the position where the startup ends.

[0097] Step S7: Synthesize the three-phase current data to find the maximum current value during the startup process and calculate the total duration from the startup start point to the startup end point to comprehensively evaluate the motor startup performance.

[0098] In the above step S2, select the frequency point corresponding to the maximum amplitude within the frequency range of 10 - 100 Hz as the main frequency FL. This is because during the motor startup process, the fundamental frequency of the motor is usually within this frequency range; divide the sampling frequency fs by the main frequency to obtain the number of points Ncycle in each period, which is expressed by the following formula:

[0099]

[0100] where fs represents the sampling frequency and FL represents the main frequency calculated through FFT.

[0101] The above steps can flexibly determine the number of points in each period according to the actual operating frequency of the motor and are applicable to the time-domain waveform analysis during the startup process of different types of variable-frequency motors.

[0102] In the above step S3, the local maximum detection algorithm includes:

[0103] Set an empty array or list to store the detected local maxima; traverse from the first point of the one-dimensional signal, one by one to the last point; for each point, determine whether the point is larger than the points on its left and right sides; if the current point is larger than the points on its left and right sides, then consider this point as a local maximum and add the local maximum to the previously initialized array. Among them, the array storing the local maxima will contain all the detected potential peak points, and finally an array storing a series of local maxima is obtained. These local maxima can initially represent all potential peak points.

[0104] Specifically, setting the distance threshold and height threshold includes:

[0105] To prevent detecting too many peaks in the fast-fluctuating part of the signal, especially when the signal has a large amount of noise, it is necessary to set the minimum horizontal distance (distance) between adjacent peaks to ensure that all adjacent peaks are at least separated by distance sampling points. If the distance between two peaks is less than distance, then remove the lower one. Assuming d represents the distance between adjacent peaks, then d ≥ distance.

[0106] In practical applications, the value of distance should be slightly less than the number of points per cycle (Ncycle). According to practical experience, it is more reasonable for the value of distance to be in the range of [Ncycle - 20, Ncycle - 10]. This can ensure that at most one peak can be detected within each cycle, and at the same time avoid additional peaks introduced by signal noise.

[0107] To exclude those smaller peaks that may be caused by noise, it is also necessary to set the peak height condition height and determine whether the height of the current peak exceeds the peak height condition.

[0108] In this embodiment, height can be a number or a tuple:

[0109] If height is a number, then the height h of all peaks must satisfy h > height.

[0110] If height is a tuple, then the first element is the minimum peak height (heightmin), and the second element is the maximum peak height (heightmax), that is, the height h of all peaks must satisfy heightmin < h < heightmax.

[0111] By setting the height condition, it can be ensured that only when the height of the peaks exceeds this threshold will they be recorded. This helps to exclude those smaller peaks that may be caused by noise. In practical applications, it is recommended to set height to the average value of the absolute values of the one-dimensional signal. This can be achieved by calculating the mean value of the absolute values of the signal. If a tuple needs to be set as the peak height condition, a certain multiple of the mean value of the absolute values of the signal can be considered as the minimum peak height (1.2μabs) and the maximum peak height (1.5μabs):

[0112] The average value μabs of the absolute values of the one-dimensional signal can be calculated by the following formula:

[0113] Assume that the one-dimensional signal is x = [x1, x2,..., xn], where n is the length of the signal.

[0114]

[0115] Here, ∣xi∣ represents the absolute value of the i-th element in the signal.

[0116] Apply threshold: Set the threshold parameter to specify the minimum threshold on both sides of the peak. This threshold refers to the minimum value when the signal value drops from the peak to a certain point, and its calculation process is as follows:

[0117] First, for each peak, calculate the values when the adjacent signal values on its left and right drop to the lowest point. For each peak, compare whether the thresholds on both sides are within the range of the threshold parameter; retain the peaks that meet the conditions, that is, retain those peaks whose thresholds on both sides are within the specified range. The purpose of doing this is to exclude those peaks whose vertical distance from the adjacent samples is less than the threshold.

[0118] Specifically, threshold can be a number or a tuple:

[0119] If threshold is a number, the threshold t of all peaks must satisfy t > threshold;

[0120] If threshold is a tuple, the first element is the minimum threshold (thresholdmin), and the second element is the maximum threshold (thresholdmax), that is, the threshold t of all peaks must satisfy thresholdmin < t < thresholdmax.

[0121] By setting the threshold condition, it can be ensured that only when the vertical distance between the peak and its neighbors exceeds this threshold will they be recorded. This helps to exclude those peaks whose vertical distance from the adjacent samples is less than the threshold.

[0122] In practical applications, a reasonable threshold value can be set based on the statistical characteristics of the signal. For example, 1.5 times the standard deviation of the absolute value of the signal can be used as the threshold. It can be expressed by the following mathematical calculation formula:

[0123] Assume that the one-dimensional signal is x = [x1, x2,..., xn], where n is the length of the signal.

[0124] The standard deviation σabs of the absolute value of the signal can be calculated by the following formula:

[0125]

[0126] where, ∣xi∣ represents the absolute value of the i-th element in the signal. μabs represents the average value of the absolute values of the one-dimensional signal.

[0127] In this embodiment, the prominence of a peak refers to the height of the peak relative to the surrounding baseline. In signal processing, prominence is an important indicator to measure the prominence degree of a peak. The calculation of prominence involves finding the baselines on both the left and right sides of the peak, and then measuring the vertical distance from the top of the peak to these two baselines. The calculation process is as follows:

[0128] First, determine the baseline points. For each potential peak point, search to the left and right to find the first local minimum point that no longer rises above the height of the peak; these points are the left and right baseline points. If no such point can be found on one side, the baseline point can be considered as the boundary point of the signal; then calculate the prominence. Specifically, for each peak, first calculate the height of the peak point minus the height of the left baseline point to obtain the left prominence; then calculate the height of the peak point minus the height of the right baseline point to obtain the right prominence. Generally, the prominence of the peak takes the smaller value of the left and right prominences.

[0129] Specifically, in order to exclude those peaks whose prominences do not meet the requirements, it is necessary to set the prominence condition. Prominence can be a number or a tuple:

[0130] If prominence is a number, the prominence p of all peaks must satisfy p > prominencep;

[0131] If prominence is a tuple, the first element is the minimum prominence (prominencemin), and the second element is the maximum prominence (prominencemax), that is, the prominence p of all peaks must satisfy prominencemin < p < prominencemax.

[0132] The prominence Pi can be calculated by the following formula:

[0133] Pi = max(hi - bl, hi - br)

[0134] Where hi is the height of the i-th peak, and bl and br are the heights of the left baseline and right baseline respectively.

[0135] By setting the prominence condition, it can be ensured that only when the prominence of the peak exceeds this threshold will they be recorded, which helps to exclude those peaks with insufficient prominence.

[0136] In practical applications, a reasonable prominence value can be set based on the statistical characteristics of the signal. For example, 1.5 times the standard deviation of the absolute value of the signal can be used as the prominence threshold (the calculation method is the same as above).

[0137] To sum up, in actual calculations, it can be appropriately considered whether to apply the threshold condition and the execution steps of the prominence condition according to the complexity of the signal change. For most mutation signals, peak detection can be well completed through the distance threshold and height threshold. If the signal is relatively simple and has little noise, it may not be necessary to further apply the threshold condition and prominence condition.

[0138] After a series of the above conditional judgments, the index positions representing the peaks of the input one-dimensional signal are finally obtained. These conditions together ensure that the detected peaks are the truly meaningful mutation points in the signal, excluding the smaller or unimportant peaks caused by noise or other factors.

[0139] In the above step S4, starting from the first peak position of the time-domain waveform, if the mean value of the three cycle numbers (the previous cycle number, the current cycle number, and the next cycle number) corresponding to this peak position is within the interval [Ncycle - 1, Ncycle + 1], and the absolute value of the average amplitude is less than 1, it is considered that the startup ends, and the current peak position is used as the startup end position.

[0140] Specifically, calculate the average amplitude of all peaks, expressed by the following formula:

[0141]

[0142] amp (amplitude) = Pvalue (peak) - Mvalue (average current)

[0143]

[0144] Among them, Ncycle represents the number of points per cycle, Mvalue represents the average current, amp represents the amplitude, Pvalue represents the calculated current peak sequence, Pi represents the i-th current value between two adjacent peaks, and Mamp represents the average amplitude of the current values within two adjacent cycles.

[0145] Judge whether the average value of the current peak position cycle number and the two adjacent cycle numbers before and after it is within the interval [Ncycle - 1, Ncycle + 1], which is expressed by the following formula:

[0146]

[0147] Among them, MCnumber represents the average number of points in three adjacent consecutive cycles, Current Cycle Number represents the number of points in the current cycle, Previous Cycle Number represents the number of points in the previous cycle, and Next Cycle Number represents the number of points in the next cycle.

[0148] In this embodiment, the average number of cycles is within the interval [Ncycle - 1, Ncycle + 1], and the cycle stability includes: before the motor reaches the stable state, the cycle of the current waveform changes with the change of the motor speed. When the motor approaches the stable state, the cycles of the current waveforms tend to be consistent, that is, close to the cycle when the motor operates in the stable state. Ncycle represents the number of cycle points in the stable state of the motor. Judgment basis: Select the average value of three adjacent consecutive cycle numbers and judge whether this average value falls within an allowable error range ([Ncycle - 1, Ncycle + 1]) of the cycle in the stable state of the motor. If it falls within this interval, it means that the motor has approached or reached the stable state; among them, Ncycle represents the number of cycle points in the stable state of the motor.

[0149] In this embodiment, at the initial stage of startup, due to large current fluctuations, the amplitude is also large. As the motor approaches the stable state, the current waveform tends to be stable and the amplitude also decreases accordingly. Judgment basis: When judging whether the absolute value of the average amplitude is less than the set threshold, if the absolute value of the average amplitude is less than the set threshold (for example, 1), it means that the current waveform tends to be stable and there are no significant fluctuations, then the motor startup process is completed and it enters the stable operation state. The reason why the threshold is set to 1 here is that theoretically, if the motor is in the stable operation stage, the average amplitude should be close to 0, but due to the possible presence of harmonics and noise in the signal (as well as the influence of the frequency converter), the average amplitude in actual measurement will not be exactly 0, so the set threshold here needs to be greater than 0.

[0150] Therefore, by combining these two conditions, it is possible to effectively determine when the motor transitions from the starting phase to the stable operating phase. When the period is stable and the amplitude decreases, it can be considered that the motor has completed the starting process. This is because after the end of the starting phase, both the electromagnetic field and mechanical motion of the motor reach a dynamic equilibrium, and the current waveform becomes more regular and stable. In practical applications, these thresholds (such as the allowable error range of the period and the amplitude threshold) need to be adjusted and optimized according to the specific motor characteristics and experimental data to obtain the best judgment result for the end point of starting the machine.

[0151] In summary, the method provided by the present invention can accurately identify the transient interval during the motor starting process, improving the accuracy of motor performance evaluation and maintenance strategies. Through fast Fourier transform, the present invention can accurately determine the main frequency during the motor starting process within a short time, thereby calculating the number of points in each period. This efficient spectrum analysis method enables the entire process to be completed quickly, improving the overall efficiency.

[0152] Furthermore, the present invention adopts a local maximum detection algorithm. By combining reasonable settings of distance thresholds and height thresholds, it can quickly and accurately detect the peaks in the time-domain waveform. This efficient peak detection mechanism reduces unnecessary calculations and accelerates the entire analysis process.

[0153] Furthermore, the present invention determines the main frequency through FFT and can adapt to the frequency changes during the starting process of variable-frequency motors, ensuring the accuracy of the analysis. The distance thresholds and height thresholds set by the peak detection algorithm of the present invention can ensure accurate peak detection even under large frequency changes, excluding the influence of noise and enhancing the robustness of the algorithm.

[0154] Furthermore, the present invention determines the end point of starting the machine through the average number of periods and the absolute value of the average amplitude. This method can well adapt to the influence brought by frequency changes during the motor starting process, ensuring accurate judgment of the end point of starting the machine. The settings of the distance threshold and height threshold are intuitive and simple, easy to understand and adjust, enabling non-professional technicians to operate easily.

[0155] Furthermore, the present invention realizes automation from the entire process of original signal acquisition to starting process feature extraction, reducing the operation difficulty and the need for human intervention.

[0156] Furthermore, the present invention automatically identifies the starting moment of starting the machine through the current threshold, and determines the end point of starting the machine through the number of periods and the amplitude, making the entire process easy to understand and execute.

[0157] Furthermore, the present invention is not only applicable to fixed-frequency motors, but also particularly applicable to variable-frequency motors and can handle different frequency change situations.

[0158] Furthermore, the present invention is verified through tests at different main frequencies, including 50Hz / 36Hz / 30Hz / 20Hz, etc., demonstrating the effectiveness of the algorithm under various motor types and operating conditions.

[0159] Furthermore, the present invention has been tested in a variety of application scenarios, ensuring the robustness of the algorithm in complex motor startup scenarios and enhancing its applicability and generalization ability in different environments.

[0160] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0161] The above embodiments are only the preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and substitutions made by those skilled in the art based on the present invention belong to the scope of protection required by the present invention.

Claims

1. A method for identifying transient intervals of time-domain waveforms of asynchronous motors based on peak detection, characterized in that: The following steps are involved: Get the three-phase current of asynchronous motor; Perform fast Fourier transform on the original time domain waveforms of the three-phase currents to obtain the frequency and amplitude spectra of the signals; determine the main frequency and obtain the number of points in each cycle; Use the local maximum detection algorithm to identify all peaks in the time domain waveform and set the distance threshold and height threshold; By calculating the position difference between adjacent peaks, the number of cycles between two adjacent peaks is obtained; the amplitude of each peak is found, and the average amplitude of all peaks is calculated through the amplitude; Start traversing from the first peak position of the time domain waveform, determine whether the current peak position cycle number and the average of the two cycle numbers before and after it are within the expected cycle point range, and determine whether the absolute value of the average amplitude is less than the set threshold. If these two conditions are met, it is considered that the startup is completed, and the current peak position is used as the startup end position; Set the current threshold, find the first point that exceeds the current threshold in the three-phase current data, and use it as the start-up time to ensure that the analysis range covers the complete start-up process from the start-up time to the end of the start-up. By combining the current data of each phase, finding the maximum current value during the startup process and calculating the total time from the start point to the start end point, the motor starting performance is fully evaluated.

2. The method according to claim 1, characterized in that: When determining the main frequency, select the frequency point corresponding to the maximum amplitude between 10-100Hz as the main frequency FL, and divide the sampling frequency fs by the main frequency to obtain the number of points per cycle Ncycle, which is expressed as the following formula: Among them, fs represents the sampling frequency, and FL represents the main frequency calculated by FFT.

3. The method according to claim 1, characterized in that: The local maximum detection algorithm includes: Set an empty array or list to store the detected local maxima; Start traversing from the first point of the one-dimensional signal and traverse to the last point one by one; For each point, determine whether the point is larger than the points on its left and right sides; If the current point is larger than the points on its left and right, the point is considered to be a local maximum, and the local maximum is added to the previously initialized array, where the array storing the local maximum will contain all detected potential peak points.

4. The method according to claim 1, characterized in that: The setting of the distance threshold and the height threshold comprises: Set the minimum horizontal distance between adjacent peaks to ensure that all adjacent peaks are at least distanced apart. If the distance between two peaks is less than distance, remove the lower one. Assume that d represents the distance between adjacent peaks, then d≥distanced; Set the peak height condition height to determine whether the height of the current peak exceeds the peak height condition.

5. The method according to claim 4, characterized in that: Set the threshold parameter to specify the minimum threshold on both sides of the peak. This threshold refers to the minimum value when the signal value drops from the peak to a certain point. The calculation process is as follows: First, for each peak, calculate the value of the signal values ​​on its left and right sides that drop to the lowest point. For each peak, compare whether the thresholds on both sides are within the range of the threshold parameter. The peaks that meet the conditions are retained.

6. The method according to claim 5, characterized in that: The significance of the calculated value, that is, the height of the peak relative to the surrounding baseline, is calculated as follows: First, determine the baseline point. For each potential peak point, search left and right to find the first local minimum point, which no longer rises above the peak height. These points are the left baseline point and the right baseline point. If no such point is found on one side, the baseline point can be considered to be the boundary point of the signal. For each peak, the difference between its height and the height of the left baseline point is calculated to obtain the left significance; then the difference between its height and the height of the right baseline point is calculated to obtain the right significance.

7. The method according to any one of claims 1 to 6, characterized in that: The average amplitude of all peak values ​​is calculated as follows: amp(amplitude)=Pvalue(peak value)-Mvalue(current mean value) Wherein, Ncycle represents the number of points in each cycle, Mvalue represents the current mean, amp represents the amplitude, Pvalue represents the calculated current peak sequence, Pi represents the i-th current value between two adjacent peaks, and Mamp represents the average amplitude of the current value in two adjacent cycles.

8. The method according to any one of claims 1 to 6, characterized in that: Determine whether the current peak position cycle number and the average of the two cycle numbers before and after it are within the interval [Ncycle-1, Ncycle+1], which can be expressed as the following formula: Among them, MCnumber represents the average number of points in three consecutive cycles, Current Cycle Number represents the number of points in the current cycle, Previous Cycle Number represents the number of points in the previous cycle, and Next Cycle Number represents the number of points in the next cycle.

9. The method according to any one of claims 1 to 6, characterized in that: Select three adjacent continuous cycles to calculate the average value, and determine whether the average value falls within an allowable error range of the motor's stable state cycle ([Ncycle-1, Ncycle+1]). If it falls within this range, it means that the motor has approached or reached a stable state. Wherein, Ncycle represents the number of cycle points in the stable state of the motor.

10. The method according to claim 9, characterized in that: When judging whether the absolute value of the average amplitude is less than the set threshold, if the absolute value of the average amplitude is less than the set threshold, the motor startup process has been completed and enters a stable operation state, wherein the set threshold is greater than 0.

Images