Vacuum consumable electrode furnace molten drop testing method for high-temperature alloy smelting
By constructing voltage drift index and droplet confidence, the error problem of droplet number detection during vacuum arc furnace smelting was solved, and accurate evaluation of the smelting effect of high-temperature alloys was achieved.
Patent Information
- Application Number
- CN202512032922.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-30
- Publication Date
- 2026-02-24
AI Technical Summary
In the existing technology for vacuum arc furnace smelting, the voltage baseline drift causes a large error in the detection of the number of molten droplets, which affects the evaluation of the smelting effect of high-temperature alloys.
By acquiring voltage data and feed rate in real time, a voltage drift index, a droplet probability index for significant valleys, and a droplet confidence level are constructed. Using techniques such as sequence decomposition algorithm and sliding t-test algorithm, the number of droplets is accurately identified.
It improves the accuracy of droplet count testing, enhances the accuracy of alloy smelting stability assessment, and reduces false positives and false negatives.
Smart Images
Figure CN121557723A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of droplet testing technology, specifically to a method for testing droplets in a vacuum arc furnace used in high-temperature alloy smelting. Background Technology
[0002] Vacuum arc remelting furnaces (ARFs) are the main equipment for smelting titanium alloys and alloy steels. During alloy smelting in an ARF, the metal at the electrode tip melts and falls into the molten pool as droplets to solidify into ingots. The formation, transition, and solidification of these droplets directly affect the purity of the final ingot. Droplet parameters are closely related to electrode wear and molten pool stability; insufficient droplet quantity can lead to frequent short circuits and affect arc stability. Therefore, developing a droplet testing method for ARFs used in alloy smelting is of great significance for optimizing alloy smelting performance.
[0003] Existing technologies typically measure the number of molten droplets by comparing the voltage drop signal caused by the droplets' descent with a fixed threshold. However, during alloy smelting, as the consumable electrode is continuously consumed, the distance (arc length) between the consumable electrode and the molten pool surface changes. Even with continuous adjustments from the control system, it is difficult to completely avoid arc length fluctuations, which in turn causes drift in the arc voltage baseline. Furthermore, since the direction of voltage signal drift is uncertain, the number of molten droplets detected by the fixed threshold detection method used in existing technologies may also have significant errors, thus affecting the evaluation of the high-temperature alloy smelting effect. Summary of the Invention
[0004] To address the aforementioned technical problems, this application provides a method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting, thereby resolving the existing issues.
[0005] The present application provides a method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting, which employs the following technical solution: One embodiment of this application provides a method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting, the method comprising the following steps: Real-time acquisition of voltage data and feed rate of the vacuum arc furnace during the smelting process; The entire acquisition time is divided into multiple cycles. Based on the trend of voltage data changes in each cycle and the dispersion of feed rate in each cycle, the voltage drift index of each cycle is obtained. When the voltage drift index of each cycle is within the preset range, the voltage data of each cycle is compared with the preset voltage threshold to obtain the number of molten droplets in each cycle. When the voltage drift index of each cycle exceeds the preset range, the significant valleys of each cycle are obtained based on the difference between all the minimum points in the voltage data of each cycle and the previous data. The droplet probabilities of each significant valley are obtained based on the degree of abrupt change and the number of minimum points in the voltage data within the neighborhood window of each significant valley. The similarity factor between each significant valley and any other significant valley is obtained based on the difference between the droplet probabilities of each significant valley and any other significant valley, and the similarity between the neighborhood window data of each significant valley and any other significant valley. The droplet confidence of each significant valley in each cycle is obtained based on the mean of the similarity factors between each significant valley and all other significant valleys in each cycle, and then the number of droplets in each cycle is obtained.
[0006] Preferably, the method for obtaining the voltage drift index of each cycle is as follows: using a sequence decomposition algorithm to obtain the trend sequence corresponding to the voltage data of each cycle, and calculating the absolute value of the slope of the fitted line corresponding to the trend sequence; calculating the variance of the feed rate of each cycle; the voltage drift index of each cycle is positively correlated with the absolute value of the slope and the variance.
[0007] Preferably, the specific process of comparing the voltage data corresponding to each cycle with a preset voltage threshold to obtain the number of molten droplets in each cycle is as follows: if the voltage data corresponding to any minimum point in the voltage data of each cycle is less than the preset voltage threshold, then the minimum point is taken as the molten droplet point of each cycle; the total number of molten droplet points in each cycle is recorded as the number of molten droplets in each cycle.
[0008] Preferably, the method for obtaining the significant valley points of each cycle is as follows: calculate the absolute difference between each minimum point in the voltage data of each cycle and the previous data, and record the minimum point corresponding to the absolute difference that is greater than or equal to a preset segmentation threshold as the significant valley point of each cycle.
[0009] Preferably, the method for obtaining the suspected droplet index of each significant valley point is as follows: obtaining the mutation value of each significant valley point based on the degree of mutation of the voltage data in the neighborhood window of each significant valley point; obtaining the total number of minimum points in the neighborhood window of each significant valley point; and recording the ratio of the mutation value to the total number as the suspected droplet index of each significant valley point.
[0010] Preferably, the process of obtaining the mutation value of each significant valley point is as follows: all data within the neighborhood window of each significant valley point are used as input to the sliding t-test algorithm, the t-values of the data within the neighborhood window of each significant valley point are output, and the absolute value of the t-value is recorded as the mutation value of each significant valley point.
[0011] Preferably, the method for obtaining the similarity factor between each significant valley point and any other significant valley point is as follows: calculating the absolute difference between the suspected droplet index of each significant valley point and any other significant valley point; obtaining the first similarity between each significant valley point and any other significant valley point based on the similarity between the neighborhood window data of each significant valley point and any other significant valley point; and recording the ratio of the first similarity to the absolute difference as the similarity factor between each significant valley point and any other significant valley point.
[0012] Preferably, the first similarity between each significant valley point and any other significant valley point refers to the absolute value of the correlation coefficient between the voltage sequences of each significant valley point and the neighboring voltage sequences of any other significant valley point; wherein, the neighboring voltage sequence of each significant valley point refers to the sequence composed of voltage data within the neighboring window of each significant valley point arranged in temporal order.
[0013] Preferably, the melt droplet confidence level of each significant valley point in each period refers to the mean of the similarity factors between each significant valley point and all other significant valley points in each period.
[0014] Preferably, the specific process of obtaining the number of molten droplets in each period is as follows: the significant valley point in each period where the confidence level of the molten droplets is greater than or equal to the preset molten droplet threshold is recorded as the molten droplet point of each period, and the total number of all molten droplet points in each period is recorded as the number of molten droplets in each period.
[0015] This application has at least the following beneficial effects: This application addresses the problem that existing technologies do not adequately consider voltage baseline drift, leading to significant errors in droplet count detection using fixed threshold methods. By analyzing the trend characteristics of voltage data and the feed rate characteristics of the consumable electrode, a voltage drift index is constructed to assess the likelihood of voltage signal baseline drift. Furthermore, by analyzing the abrupt changes and regularity in the voltage data, a droplet probability index is constructed to preliminarily assess whether extreme points are caused by actual droplet events. Finally, by analyzing the similarity between each significant valley point and other significant valley points, a droplet confidence score is constructed, enabling accurate identification of droplet points caused by actual droplet events. This avoids false positives or false negatives caused by fixed threshold failure due to baseline drift, improving the accuracy of droplet count testing and consequently enhancing the accuracy of subsequent alloy smelting stability assessments. Attached Figure Description
[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 A flowchart of the steps for a vacuum arc furnace droplet testing method for high-temperature alloy smelting provided in this application; Figure 2 A flowchart illustrating the process of obtaining the confidence level of the melt droplets at each significant valley point in each period provided in this application. Detailed Implementation
[0018] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a vacuum arc furnace droplet testing method for high-temperature alloy smelting proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.
[0020] The following description, in conjunction with the accompanying drawings, details a specific scheme for a vacuum arc furnace droplet testing method for high-temperature alloy smelting provided in this application.
[0021] One embodiment of this application provides a method for testing molten droplets in a vacuum arc remelting furnace for high-temperature alloy smelting. Specifically, the method is described below. Please refer to [link to relevant documentation]. Figure 1 The method includes the following steps: Step 1: Real-time acquisition of voltage data and feed rate of the vacuum arc furnace during the smelting process.
[0022] In this embodiment, the output voltage reference value of the vacuum arc furnace power supply for alloy smelting is set to 24V, and the output fundamental frequency is set to 300Hz. During the alloy smelting process, the arc furnace voltage signal output by the power supply is acquired and transformed by a 10:1 ratio to convert the voltage to the range of 0-2.4V. Then, the transformed voltage signal is sampled by a DSP processor at a frequency of 10kHz. Next, the sampled voltage data is subjected to multi-order notch filtering to filter out harmonic effects. The notch orders are third (900Hz), fifth (1500Hz), ninth (2700Hz), eleventh (3300Hz), fifteenth (4500Hz), and seventeenth (5100Hz). In addition, the real-time feed speed of the arc furnace electrode is acquired through the arc furnace control system at a data acquisition frequency of 20Hz.
[0023] To eliminate the influence of different dimensions among the data, all data are normalized. The normalization method is not limited to Z-score, maximum value normalization, or maximum and minimum value normalization.
[0024] Step 2: Divide the entire acquisition time into multiple cycles. Based on the trend of voltage data changes in each cycle and the dispersion of feed rate in each cycle, obtain the voltage drift index for each cycle.
[0025] During the melting process of consumable electrodes, the power supply output voltage is a set value, and the control system adjusts the position of the consumable electrode in real time through the feed electrode rod to maintain a stable arc length, thereby ensuring stable arc heat. Therefore, if the arc length is stable and no baseline drift occurs in the voltage signal, although the falling of the molten droplet will cause a voltage drop, the overall voltage signal will remain stable. However, if baseline drift occurs, the actual voltage signal will show a slow upward or downward trend. Therefore, the initial analysis of the overall trend change of the voltage signal can be used to determine whether baseline drift has occurred.
[0026] The entire data acquisition time is divided into cycles of s seconds, where s is 5 in this embodiment. The following analysis takes the i-th cycle as an example.
[0027] Voltage data from each period are arranged in chronological order to obtain voltage sequences for each period. A sequence decomposition algorithm is used to decompose the voltage sequence in the i-th period to obtain a trend sequence. This trend sequence is then fitted with a straight line, and the absolute value of the slope of the fitted line is calculated. The obtained absolute slope value reflects the overall trend of voltage data in the i-th period; a larger value indicates a greater degree of trend change and a higher likelihood of drift. Sequence decomposition algorithms are not limited to STL sequence decomposition or SEATS sequence decomposition.
[0028] Furthermore, abnormal changes in the feed rate of the consumable electrode may also cause arc length fluctuations, which could in turn cause the arc voltage baseline to drift. Therefore, further evaluation can be conducted by considering the stability of the consumable electrode feed rate.
[0029] The dispersion of the feed rate in the i-th cycle is calculated. The resulting dispersion reflects whether the feed rate of the consumable electrode will fluctuate significantly. The larger the value, the more unstable the feed rate in that cycle, resulting in a decrease in the ability to maintain arc length stability and a greater risk of baseline drift. The calculation of dispersion is not limited to variance, standard deviation, and coefficient of variation; in this embodiment, variance is used for calculation.
[0030] In a preferred embodiment, a voltage drift index is obtained for each cycle based on the trend of voltage data changes and the dispersion of feed rate within each cycle. This index characterizes the probability of voltage data drift in each cycle. The method for obtaining the voltage drift index for each cycle is as follows: the absolute value of the slope of the fitted straight line of voltage data for each cycle is calculated; the variance of feed rate for each cycle is calculated; the voltage drift index for each cycle is positively correlated with both the absolute value of the slope and the variance. This positive correlation means that the dependent variable increases (decreases) as the independent variable increases (decreases).
[0031] Preferably, in this embodiment, the voltage drift index of the i-th cycle is denoted as... Its specific expression is: In the formula, Let be the voltage drift index for the i-th cycle. Let be the absolute value of the slope of the fitted straight line of the voltage sequence in the i-th period; Let be the variance of the feed rate in the i-th cycle.
[0032] The voltage drift index reflects the probability that baseline drift occurred in the voltage data during the i-th period. The larger the value, the more obvious the overall trend of the voltage data in the i-th acquisition period has drift characteristics, and the worse the feed stability of the consumable electrode is, thus increasing the probability of arc voltage baseline drift.
[0033] Step 3: When the voltage drift index of each cycle is within the preset range, the voltage data of each cycle is compared with the preset voltage threshold to obtain the number of molten droplets in each cycle.
[0034] Furthermore, if the voltage data does not exhibit baseline drift and the overall trend remains stable, the number of molten droplets can be identified using existing threshold comparison methods. However, if the voltage data shows baseline drift and its overall trend exhibits an upward or downward change, using a fixed threshold comparison method to test the number of molten droplets will result in significant errors.
[0035] The voltage drift index of Z (300 in this embodiment) historical cycles is obtained from the historical database of the vacuum arc furnace. Then, the voltage drift index of the i-th cycle and the Z historical cycles are used as... The input to the criterion outlier detection algorithm is obtained. The range, the result The range is denoted as the preset range.
[0036] If the voltage drift index of the i-th period is within a preset range, it indicates that the voltage data collected in the i-th period has not experienced drift. In this case, the number of molten droplets can be detected using an existing preset voltage threshold comparison method. Specifically, an extreme value detection algorithm is used to obtain all minimum values of the voltage sequence in the i-th period. If the voltage data corresponding to any minimum point is less than the preset voltage threshold, then that minimum point is taken as the molten droplet point in the i-th period. The total number of molten droplets in the i-th period is counted, and the total number of molten droplets in the i-th period is recorded as the number of molten droplets in the i-th period. In this embodiment, the preset voltage threshold is 1.2V. The extreme value detection algorithm is a known technique, and its specific process will not be described in detail.
[0037] Step 4: When the voltage drift index of each cycle exceeds the preset range, obtain the significant valleys of each cycle based on the difference between all minimum points in the voltage data within each cycle and the previous data; obtain the droplet probabilities of each significant valley based on the degree of abrupt change and the number of minimum points in the voltage data within the neighborhood window of each significant valley; obtain the similarity factor between each significant valley and any other significant valley based on the difference between the droplet probabilities of each significant valley and any other significant valley, and the similarity between the neighborhood window data of each significant valley and any other significant valley; obtain the droplet confidence of each significant valley in each cycle based on the mean of the similarity factors between each significant valley and all other significant valleys in each cycle, and then obtain the number of droplets in each cycle.
[0038] If the voltage drift index in the i-th cycle is not within the preset range, it indicates that the voltage data in the i-th cycle has drifted. In this case, under high-frequency sampling conditions, if baseline correction is directly applied to the original voltage signal, the steep falling edge and short duration of the droplet pulsation will cause pulse shape distortion after baseline correction, which will affect the subsequent test results of the droplet count by setting a threshold. Therefore, to improve the accuracy of droplet count testing, for voltage data with baseline drift, the voltage drop characteristics caused by droplet falling can be combined to more accurately obtain the droplet count.
[0039] Specifically, while the collected voltage data may experience momentary drops due to factors such as feed jitter of the consumable electrode, boiling of the molten pool, and transient power supply disturbances, the voltage drops caused by molten droplet discharge exhibit periodicity and stability compared to random drops caused by these random interference factors. Therefore, analyzing the local characteristics of the voltage drop can help determine whether it is a genuine molten droplet event.
[0040] Let's take the i-th cycle as an example for analysis.
[0041] The extreme value detection algorithm is used to obtain all the minimum values of the voltage sequence within the i-th period. Considering that there may be many minimum values in the voltage data with a high sampling rate, it is necessary to filter the minimum values to improve the detection accuracy and computational efficiency.
[0042] The absolute difference between each minimum point and its preceding data is obtained, and all absolute differences are used as inputs to the Otsu thresholding method to obtain a segmentation threshold, which is denoted as the preset segmentation threshold. The minimum points corresponding to the absolute differences greater than or equal to the preset segmentation threshold are denoted as the significant valleys of the i-th period. The Otsu thresholding method is a well-known technique, and the specific process will not be described in detail.
[0043] Since molten droplets are usually accompanied by an arc short circuit, the resulting voltage drop is relatively steep, and the steepness is usually greater than that caused by most interference factors.
[0044] Let's take the u-th significant valley point in the i-th cycle as an example for analysis. A window of size 1×N (N is 13 in this embodiment) is constructed centered on the u-th significant valley point, denoted as the neighborhood window of the u-th significant valley point. All data within this neighborhood window are used as input to the sliding t-test algorithm. The output is the t-value of the data within the neighborhood window of the u-th significant valley point, and the absolute value of the t-value is recorded as the mutation value of the u-th significant valley point. The obtained mutation value reflects the degree of mutation in the voltage data within the neighborhood window of the u-th significant valley point; the larger the value, the more significant the mutation. The sliding t-test algorithm is a well-known technique, and its specific process will not be elaborated further.
[0045] It should be noted that when the center of the window is near the ends of the sequence, causing the window to exceed the sequence range, the portion of the window that exceeds the sequence range will be filled with the mean of the data in the window that does not exceed the sequence range.
[0046] Furthermore, the dripping process of molten droplets is a complete physical process, and the voltage drop is not an isolated, steep drop, but rather a continuous trend with certain variations. Before the short circuit occurs, the voltage will decrease due to the stretching of the molten droplet, and will increase rapidly after the arc reignites. Therefore, the voltage drop caused by the molten droplet has a certain smoothness, while the voltage drop waveform caused by other interferences is irregular.
[0047] This function retrieves the total number of local minima within the neighborhood window of the u-th significant valley. The total number of local minima reflects the irregularity of the voltage waveform within the neighborhood window of the u-th significant valley. A minimum value of 1 indicates that there is only one local minima, i.e., the u-th significant valley; a larger value indicates a greater degree of irregularity in the voltage waveform.
[0048] As a preferred implementation, the droplet simulacrum of each significant valley point is obtained based on the degree of abrupt change in voltage data and the number of minimum points within the neighborhood window of each significant valley point, which is used to characterize the probability that each significant valley point is the droplet falling time.
[0049] In this embodiment, the suspected droplet index of the u-th significant valley point is denoted as Its specific expression is: In the formula, The suspected melting droplet index for the u-th significant valley point; Let be the mutation value of the u-th significant valley point; is the total number of local minima in the neighborhood window of the u-th significant valley point.
[0050] The droplet probabilities index reflects the likelihood that the u-th significant valley is caused by a droplet event. The larger the value, the more the voltage data in the neighborhood window of the u-th significant valley matches the voltage drop characteristics caused by the droplet event, and the more likely the u-th significant valley is an extreme point caused by the droplet event.
[0051] Furthermore, since the local changes in the voltage signal caused by the dripping of molten droplets are periodic, this feature can be used to further extract the true minimum point caused by the dripping of molten droplets.
[0052] Following the calculation method for the suspected droplet index of the u-th significant valley point, the suspected droplet indexes of all other significant valley points are calculated. The absolute difference between the suspected droplet index of the u-th valley point and any other significant valley point is then recorded as the first difference between the u-th valley point and any other significant valley point. This first difference reflects the degree of difference between the local voltage data characteristics of the u-th valley point and any other significant valley point.
[0053] The voltage data within the neighborhood windows of each significant valley point are arranged in temporal order to form a sequence, denoted as the neighborhood voltage sequence of each significant valley point. The absolute value of the correlation coefficient between the u-th significant valley point and the neighborhood voltage sequences of any other significant valley point is calculated and denoted as the first similarity between the u-th significant valley point and any other significant valley point. The first similarity reflects whether the overall trend of the voltage data within the neighborhood windows of the u-th significant valley point is consistent with that of any other significant valley point. The algorithm for calculating the correlation coefficient is not limited to Pearson correlation coefficient, Spearman correlation coefficient, or cosine similarity; this embodiment uses the Pearson correlation coefficient.
[0054] The ratio of the first similarity to the first difference between the u-th significant valley and any other significant valley is denoted as the similarity factor between the u-th significant valley and any other significant valley. The similarity factor reflects the similarity in morphological consistency and characteristic stability of the local voltage data corresponding to the u-th significant valley and any other significant valley. A larger value indicates a greater likelihood that the u-th significant valley and any other significant valley are caused by the same factor causing the voltage drop.
[0055] It should be noted that when calculating the similarity factor, if the denominator is 0, the denominator should be set to 0.05 to ensure the normal operation of the index.
[0056] Similarly, the similarity factor between the u-th significant valley point and all other significant valley points in the i-th period is calculated and averaged, denoted as the droplet confidence score of the u-th significant valley point in the i-th period. The droplet confidence score reflects the consistency between the neighborhood voltage data of the u-th significant valley point and all other significant valley points in the i-th period. A larger value indicates greater similarity between the u-th significant valley point and other significant valley points in local voltage data, and the more likely the u-th significant valley point is to be a valley value caused by a genuine droplet event. The process for obtaining the droplet confidence score of each significant valley point in each period is as follows: Figure 2 As shown.
[0057] Similarly, the melt drop confidence of all significant valley points in the i-th period is obtained and used as the input of Otsu thresholding method. The output segmentation threshold is recorded as the preset melt drop threshold. Significant valley points with melt drop confidence greater than or equal to the preset melt drop threshold are recorded as melt drop points in the i-th period. Then, the total number of melt drop points in the i-th period is recorded as the number of melt droplets in the i-th period.
[0058] Similarly, obtain the number of molten droplets for all cycles.
[0059] Measuring the number of molten droplets during alloy smelting using the above method can improve the accuracy of the measurement, making subsequent assessments of alloy smelting stability more accurate.
[0060] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.
[0061] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
[0062] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them; modifications to the technical solutions described in the foregoing embodiments, or equivalent substitutions of some of the technical features, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.
Claims
1. A method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting, characterized in that, The method includes the following steps: Real-time acquisition of voltage data and feed rate of the vacuum arc furnace during the smelting process; The entire acquisition time is divided into multiple cycles. Based on the trend of voltage data changes in each cycle and the dispersion of feed rate in each cycle, the voltage drift index of each cycle is obtained. When the voltage drift index of each cycle is within the preset range, the voltage data of each cycle is compared with the preset voltage threshold to obtain the number of molten droplets in each cycle. When the voltage drift index of each cycle exceeds the preset range, the significant valleys of each cycle are obtained based on the difference between all the minimum points in the voltage data of each cycle and the previous data. The droplet probabilities of each significant valley are obtained based on the degree of abrupt change and the number of minimum points in the voltage data within the neighborhood window of each significant valley. The similarity factor between each significant valley and any other significant valley is obtained based on the difference between the droplet probabilities of each significant valley and any other significant valley, and the similarity between the neighborhood window data of each significant valley and any other significant valley. The droplet confidence of each significant valley in each cycle is obtained based on the mean of the similarity factors between each significant valley and all other significant valleys in each cycle, and then the number of droplets in each cycle is obtained.
2. The method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting as described in claim 1, characterized in that, The method for obtaining the voltage drift index of each cycle is as follows: the trend sequence corresponding to the voltage data of each cycle is obtained by using a sequence decomposition algorithm, and the absolute value of the slope of the fitted line corresponding to the trend sequence is calculated; the variance of the feed rate of each cycle is calculated; the voltage drift index of each cycle is positively correlated with the absolute value of the slope and the variance.
3. The method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting as described in claim 1, characterized in that, The specific process of comparing the voltage data corresponding to each cycle with a preset voltage threshold to obtain the number of molten droplets in each cycle is as follows: if the voltage data corresponding to any minimum point in the voltage data of each cycle is less than the preset voltage threshold, then the minimum point is taken as the molten droplet point of each cycle; the total number of molten droplet points in each cycle is recorded as the number of molten droplets in each cycle.
4. The method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting as described in claim 1, characterized in that, The method for obtaining the significant valley points of each cycle is as follows: calculate the absolute difference between each minimum point in the voltage data of each cycle and the previous data, and record the minimum point corresponding to the absolute difference that is greater than or equal to the preset segmentation threshold as the significant valley point of each cycle.
5. The method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting as described in claim 1, characterized in that, The method for obtaining the suspected droplet index of each significant valley point is as follows: obtain the mutation value of each significant valley point according to the degree of mutation of the voltage data in the neighborhood window of each significant valley point; obtain the total number of minimum points in the neighborhood window of each significant valley point; and record the ratio of the mutation value to the total number as the suspected droplet index of each significant valley point.
6. The method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting as described in claim 5, characterized in that, The process of obtaining the mutation value of each significant valley point is as follows: take all the data in the neighborhood window of each significant valley point as the input of the sliding t-test algorithm, output the t value of the data in the neighborhood window of each significant valley point, and record the absolute value of the t value as the mutation value of each significant valley point.
7. The method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting as described in claim 1, characterized in that, The method for obtaining the similarity factor between each significant valley point and any other significant valley point is as follows: calculate the absolute difference between the suspected droplet index of each significant valley point and any other significant valley point; obtain the first similarity between each significant valley point and any other significant valley point based on the similarity between the neighborhood window data of each significant valley point and any other significant valley point; and record the ratio of the first similarity to the absolute difference as the similarity factor between each significant valley point and any other significant valley point.
8. The method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting as described in claim 7, characterized in that, The first similarity between each significant valley point and any other significant valley point refers to the absolute value of the correlation coefficient between each significant valley point and the neighboring voltage sequence of any other significant valley point; wherein, the neighboring voltage sequence of each significant valley point refers to the sequence composed of voltage data within the neighborhood window of each significant valley point arranged in temporal order.
9. The method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting as described in claim 1, characterized in that, The melt drop confidence level of each significant valley point in each period refers to the mean of the similarity factor between each significant valley point and all other significant valley points in each period.
10. The method for testing molten droplets in a vacuum arc furnace for high-temperature alloy smelting as described in claim 1, characterized in that, The specific process for obtaining the number of molten droplets in each period is as follows: the significant valley point in each period where the confidence level of the molten droplets is greater than or equal to the preset molten droplet threshold is recorded as the molten droplet point of each period, and the total number of all molten droplet points in each period is recorded as the number of molten droplets in each period.