A real-time temperature monitoring method in high-frequency heating anchor cable production process

Abstract

The present application relates to the technical field of data processing, and especially relates to a real-time temperature monitoring method in high-frequency heating anchor cable production process, which performs STL decomposition on the temperature data sequence of the anchor cable in the high-frequency heating stage to obtain residual data, uses the box plot method to obtain to-be-analyzed data in the temperature data sequence according to the residual data, obtains suspected abnormal data according to the persistence of the temperature monitoring data of each to-be-analyzed data in the neighborhood range of the to-be-analyzed data, obtains the abnormal confidence of each suspected abnormal data according to the persistence difference of each two adjacent suspected abnormal data and the similarity between the residual data in the local time range of the time corresponding to each suspected abnormal data and the temperature data sequence, and monitors the abnormal temperature data of the anchor cable in the high-frequency heating stage according to the abnormal confidence and the residual data of each suspected abnormal data, thereby improving the accuracy of the temperature monitoring result of the anchor cable in the high-frequency heating stage.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for real-time temperature monitoring during the production of high-frequency heated anchor cables. Background Technology

[0002] Anchor cables are devices used to secure structures to the ground or foundation, and are widely used in construction, bridges, tunnels, and other engineering projects. Anchor cables are typically made of high-strength steel or alloy materials. Temperature changes during the production process directly affect the material's properties, especially its hardness, strength, and ductility. Therefore, real-time monitoring of the anchor cable's own temperature during production is extremely important. Effective and accurate temperature monitoring can promptly detect temperature anomalies and allow for adjustments to equipment and processes, making the entire production process more stable and avoiding rework or scrap rates caused by temperature fluctuations, thus preventing unnecessary waste during production.

[0003] The anchor cable production process involves three stages: high-frequency heating, spray cooling (quenching), and tempering. Among these, high-frequency induction heating is an advanced method for achieving high quality, providing a precise and stable starting point for the entire anchor cable production process. Currently, industrial production often utilizes sensors to acquire the anchor cable's own temperature data, obtaining a temperature curve over time. The temperature curve is then decomposed using STL (Standard Tracing Tool) to obtain residual components. The upper and lower limits of the residual components are obtained using a box plot method. Temperature data between the upper and lower limits are considered normal data within the acceptable range, while temperature data outside these limits are judged as abnormal temperature data, thus enabling real-time temperature monitoring of the anchor cable during high-frequency heating.

[0004] However, since high-frequency induction heating is essentially a continuous thermal process, temperature changes are affected by the inertial characteristics of heat conduction and dissipation, as well as the combined influence of process operation and equipment status. When deviations occur in the process or equipment, abnormal temperature data often do not immediately manifest as pulse-type values. That is, temperature anomalies do not necessarily appear as sudden changes, but rather as slow shifts and trend accumulations in the temperature data. Therefore, when these temperature data are decomposed by STL and remain in the residual components, they will fall within the acceptable range. In other words, the traditional method has certain limitations in detecting temperature anomalies in anchor cables during the high-frequency heating stage, and may miss some real temperature anomaly data within the acceptable range, resulting in inaccurate monitoring results of temperature during the anchor cable production process.

[0005] Therefore, how to more comprehensively identify abnormal temperature data of anchor cables during the high-frequency heating stage and improve the accuracy of temperature monitoring results of anchor cables during the high-frequency heating stage has become an urgent problem to be solved. Summary of the Invention

[0006] In view of this, embodiments of the present invention provide a method for real-time temperature monitoring during the production process of high-frequency heated anchor cables, in order to solve the problem of how to more comprehensively identify abnormal temperature data of anchor cables during the high-frequency heating stage and improve the accuracy of temperature monitoring results of anchor cables during the high-frequency heating stage.

[0007] This invention provides a method for real-time temperature monitoring during the production of high-frequency heated anchor cables, the method comprising the following steps:

[0008] Temperature monitoring data of the target anchor cable at each moment during the high-frequency heating stage are obtained to form a temperature data sequence. The temperature data sequence is decomposed by STL to obtain the residual data at each moment. The data to be analyzed is obtained from the temperature data sequence based on the residual data at each moment using the box plot method.

[0009] Based on the persistence of temperature monitoring data in the neighborhood range of each data to be analyzed in the temperature data sequence, obtain the persistence characteristic value of each data to be analyzed, and obtain at least two suspected abnormal data based on the persistence characteristic value of each data to be analyzed;

[0010] Based on the difference in the persistence characteristic values ​​of each pair of adjacent suspected anomalous data, the persistence stability of each suspected anomalous data is obtained. Based on the similarity between the residual data of each suspected anomalous data and the temperature data sequence within its local time range, the matching degree of the residual data corresponding to each suspected anomalous data is obtained.

[0011] By combining the persistence and stability of each suspected abnormal data point with the matching degree of the residual data corresponding to each suspected abnormal data point, the abnormal confidence level of each suspected abnormal data point is obtained. Based on the abnormal confidence level of each suspected abnormal data point and the residual data at each time point, the abnormal temperature data of the target anchor cable during the high-frequency heating stage is monitored.

[0012] Preferably, the step of using the box plot method to obtain the data to be analyzed from the temperature data sequence based on the residual data at each time point includes:

[0013] The residual data at all times are combined into a residual sequence. A box plot is constructed based on the residual sequence to obtain the upper and lower limits. In the box plot, the time corresponding to the residual data located between the upper and lower limits is recorded as the time to be analyzed. In the temperature data sequence, the temperature monitoring data corresponding to the time to be analyzed is recorded as the data to be analyzed.

[0014] Preferably, the step of obtaining the persistence feature value of each data point to be analyzed based on the persistence of temperature monitoring data within its neighborhood range in the temperature data sequence includes:

[0015] For any data to be analyzed, in the temperature data sequence, if the number of temperature monitoring data preceding any data to be analyzed is greater than or equal to a preset number, then the preset number of temperature monitoring data preceding any data to be analyzed in the temperature data sequence are obtained to form the left neighbor temperature data sequence of any data to be analyzed. The second-order difference sequence of the left neighbor temperature data sequence is obtained, and all data in the second-order difference sequence of the left neighbor temperature data sequence are accumulated to obtain the left neighbor temperature change characteristic value of any data to be analyzed.

[0016] In a temperature data sequence, if the number of temperature monitoring data following any data to be analyzed is greater than or equal to the preset number, then the preset number of temperature monitoring data following any data to be analyzed in the temperature data sequence are obtained to form the right neighbor temperature data sequence of any data to be analyzed. The second-order difference sequence of the right neighbor temperature data sequence is obtained, and all data in the second-order difference sequence of the right neighbor temperature data sequence are accumulated to obtain the right neighbor temperature change characteristic value of any data to be analyzed.

[0017] Based on the temperature change characteristic values ​​of the left and right neighbors of any data to be analyzed, obtain the persistence characteristic value of any data to be analyzed.

[0018] Preferably, obtaining the persistence feature value of any data to be analyzed based on the temperature change feature values ​​of its left and right neighbors includes:

[0019] If any of the data to be analyzed has only a left neighbor temperature change characteristic value, then the negative of the left neighbor temperature change characteristic value is used as the independent variable of an exponential function with the natural constant as the base, to obtain the persistence characteristic value of any of the data to be analyzed.

[0020] If any of the data to be analyzed has only a right neighbor temperature change characteristic value, then the negative of the right neighbor temperature change characteristic value is used as the independent variable of an exponential function with the natural constant as the base, to obtain the persistence characteristic value of any of the data to be analyzed.

[0021] If any data to be analyzed has both a left neighbor temperature change characteristic value and a right neighbor temperature change characteristic value, then the minimum value is obtained from the left neighbor temperature change characteristic value and the right neighbor temperature change characteristic value. The opposite of the minimum value is used as the independent variable of an exponential function with the natural constant as the base to obtain the persistence characteristic value of any data to be analyzed.

[0022] Preferably, the step of obtaining at least two suspected abnormal data points based on the persistence characteristic value of each data point to be analyzed includes:

[0023] For any data to be analyzed, if the persistent characteristic value of the data to be analyzed is greater than or equal to a preset persistent characteristic value threshold, then the data to be analyzed is recorded as suspected abnormal data.

[0024] Preferably, the step of obtaining the persistence stability of each suspected abnormal data based on the difference in persistence feature values ​​between every two adjacent suspected abnormal data includes:

[0025] For any suspected abnormal data, in the temperature data sequence, if the suspected abnormal data only has a left neighbor temperature change feature value, or the minimum of the left neighbor temperature change feature value and the right neighbor temperature change feature value of the suspected abnormal data is the left neighbor temperature change feature value, then the time corresponding to the suspected abnormal data is subtracted by a preset maximum unit time span to obtain the target time. All temperature monitoring data between the time corresponding to the suspected abnormal data and the target time in the temperature data sequence are combined into a neighborhood data sequence.

[0026] If any suspected abnormal data has only a right neighbor temperature change feature value, or if the minimum of the left and right neighbor temperature change feature values ​​of any suspected abnormal data is the right neighbor temperature change feature value, then the sum between the time corresponding to any suspected abnormal data and the preset maximum unit time span is calculated to obtain the target time. All temperature monitoring data between the time corresponding to any suspected abnormal data and the target time in the temperature data sequence are combined into a neighborhood data sequence.

[0027] In the neighborhood data sequence, calculate the absolute value of the difference between the persistent feature values ​​between every two adjacent suspected anomalous data, calculate the sum of all the absolute values ​​of the difference, and use the negative of the sum as the independent variable of an exponential function with the natural constant as the base to obtain the first stability of any suspected anomalous data.

[0028] Calculate the average time interval between every two adjacent suspected abnormal data in the neighborhood data sequence, use the negative of the average time interval as the independent variable of an exponential function with the natural constant as the base, obtain the distribution characteristic value of the suspected abnormal data in the neighborhood data sequence, calculate the proportion of the number of all suspected abnormal data in the neighborhood data sequence to the number of all temperature monitoring data in the neighborhood data sequence, and use the mean between the proportion and the distribution characteristic value as the second stability of any suspected abnormal data.

[0029] The mean between the first and second stability levels of any suspected anomalous data is calculated to obtain the persistent stability level of any suspected anomalous data.

[0030] Preferably, the step of obtaining the matching degree of the residual data corresponding to each suspected anomaly data based on the similarity between the residual data and the temperature data sequence within its local time range at the time corresponding to each suspected anomaly data includes:

[0031] For any suspected anomalous data, obtain the target time corresponding to the suspected anomalous data, and form a residual sequence of the time corresponding to the suspected anomalous data within its local time range by using the residual data of all times between the time corresponding to the suspected anomalous data and the target time. Calculate the DTW distance between the residual sequence and the temperature data sequence, and use the negative of the DTW distance as the independent variable of an exponential function with the natural constant as the base to obtain the matching degree of the residual data corresponding to the suspected anomalous data.

[0032] Preferably, the step of obtaining the anomaly confidence level of each suspected anomaly data by combining the persistence and stability of each suspected anomaly data with the matching degree of the residual data corresponding to each suspected anomaly data includes:

[0033] For any suspected anomalous data, calculate the average between the persistence and stability of the suspected anomalous data and the matching degree corresponding to the suspected anomalous data to obtain the anomalous confidence level of the suspected anomalous data.

[0034] Preferably, the step of monitoring the abnormal temperature data of the target anchor cable during the high-frequency heating stage based on the abnormal confidence level of each suspected abnormal data and the residual data at each time point includes:

[0035] Suspected abnormal data with an abnormal confidence level greater than or equal to a preset abnormal confidence threshold are recorded as abnormal temperature data;

[0036] The residual data of the target anchor cable at all times during the high-frequency heating stage are combined into a residual sequence. A box plot is constructed based on the residual sequence to obtain the upper and lower limits. In the box plot, the temperature monitoring data corresponding to the residual data outside the upper and lower limits is recorded as abnormal temperature data.

[0037] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows:

[0038] This invention acquires temperature monitoring data of the target anchor cable at each moment during the high-frequency heating stage, forming a temperature data sequence. The temperature data sequence is then decomposed using STL to obtain residual data at each moment. Using a box plot method, the data to be analyzed is obtained from the temperature data sequence based on the residual data at each moment. Based on the persistence of temperature monitoring data within the neighborhood of each data point in the temperature data sequence, a persistence characteristic value is obtained for each data point. Based on the persistence characteristic value of each data point, at least two suspected anomalies are obtained. Based on the difference in persistence characteristic values ​​between any two adjacent suspected anomalies, the persistence stability of each suspected anomaly is obtained. Based on the similarity between the residual data at the moment corresponding to each suspected anomaly and the temperature data sequence within its local time range, the matching degree of the residual data corresponding to each suspected anomaly is obtained. Combining the persistence stability of each suspected anomaly and the matching degree of the residual data corresponding to each suspected anomaly, the anomaly confidence level of each suspected anomaly is obtained. Based on the anomaly confidence level of each suspected anomaly and the residual data at each moment, abnormal temperature data of the target anchor cable during the high-frequency heating stage is monitored. Specifically, using the box plot method, data to be analyzed is obtained from the temperature data sequence based on the residual data at each time point for further analysis, thus more comprehensively identifying abnormal temperature data of the target anchor cable during the high-frequency heating stage. Based on the difference in the persistence characteristic values ​​of two adjacent suspected abnormal data points, and the similarity between the residual data and the temperature data sequence within the local time range corresponding to each suspected abnormal data point, the anomaly confidence level of each suspected abnormal data point is comprehensively obtained. This determines whether each suspected abnormal data point is a genuine anomaly caused by process or equipment deviations during the high-frequency heating process, or an anomaly caused by normal process dynamics such as short-term process switching or equipment start-up and shutdown. This allows for the monitoring of genuine abnormal data caused by process or equipment deviations during the high-frequency heating process, which is recorded as abnormal temperature data, improving the accuracy of temperature monitoring results for the target anchor cable during the high-frequency heating stage. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention, 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a flowchart of a method for real-time temperature monitoring during the production process of high-frequency heated anchor cables, provided by the present invention. Detailed Implementation

[0041] Embodiments of this disclosure are described in detail below, with examples of these embodiments illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this disclosure, and should not be construed as limiting it.

[0042] It should be noted that the terms "first," "second," etc., used in this disclosure and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure.

[0043] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0044] See Figure 1 This is a flowchart of a method for real-time temperature monitoring during the production of high-frequency heated anchor cables, provided by the present invention. Figure 1 As shown, the method may include:

[0045] Step S101: Obtain temperature monitoring data of the target anchor cable at each moment during the high-frequency heating stage, form a temperature data sequence, perform STL decomposition on the temperature data sequence to obtain residual data at each moment, and use the box plot method to obtain the data to be analyzed from the temperature data sequence based on the residual data at each moment.

[0046] Any anchor cable in the high-frequency heating stage is designated as the target anchor cable. Using a non-contact infrared thermometer (e.g., a single-point infrared thermometer), with a sampling frequency of 50Hz, temperature monitoring data of the target anchor cable at each moment during the high-frequency heating stage is acquired, forming a temperature data sequence. The sampling frequency setting is not limited and can be set by the implementer according to the scenario.

[0047] Using the STL time series decomposition algorithm, the temperature data sequence is decomposed to obtain the residual data at each time point, forming a residual sequence. The STL time series decomposition algorithm is an existing technology and will not be described in detail here. Then, a box plot is constructed based on the residual sequence to obtain the upper and lower limits. The general steps are as follows: (1) Sort the data in the residual sequence in ascending order to obtain the residual sequence; (2) Calculate the first quartile Q1, the second quartile Q2, and the third quartile Q3 of the residual sequence; (3) Calculate the interquartile range IQR = Q3 - Q1 based on the first quartile Q1 and the third quartile Q3; (4) Calculate Q3 + 1.5 × IQR as the upper limit, i.e. the upper boundary of the box plot, and calculate Q1 - 1.5 × IQR as the lower limit, i.e. the lower boundary of the box plot.

[0048] In box plots, data outside the upper and lower limits are considered outliers, or abnormal values, while data between the upper and lower limits are considered normal data within the acceptable range. Therefore, in the traditional method, the temperature monitoring data at the time corresponding to the residual data outside the upper and lower limits is directly marked as abnormal temperature data, thereby completing the real-time temperature monitoring of the target anchor cable during the high-frequency heating stage.

[0049] Because high-frequency induction heating is a gradual thermal process, temperature changes are controlled by physical laws such as heat conduction, convection, and radiation. Anomalies do not always manifest as instantaneous abrupt changes, but rather as a gradual deviation. Temperature monitoring data exhibiting this trend, not conforming to the overall data trend, are retained in the residual components after being decomposed by the STL time series decomposition algorithm. However, since extreme changes do not occur, these residuals are distributed within an acceptable range, i.e., between the upper and lower limits in the box plot. Therefore, traditional methods for detecting temperature anomalies in anchor cables during the high-frequency heating stage have certain limitations, potentially missing some genuine temperature anomalies within the acceptable range, leading to inaccurate temperature monitoring results during anchor cable production.

[0050] Therefore, in this embodiment of the invention, the temperature monitoring data at the time corresponding to the residual data located between the upper and lower limits in the box plot is recorded as the data to be analyzed. Further analysis of the data to be analyzed is performed to more comprehensively identify abnormal temperature data of the target anchor cable during the high-frequency heating stage, thereby improving the accuracy of temperature monitoring results for the target anchor cable during the high-frequency heating stage.

[0051] Step S102: Based on the persistence of temperature monitoring data in the neighborhood range of each data to be analyzed in the temperature data sequence, obtain the persistence feature value of each data to be analyzed, and obtain at least two suspected abnormal data based on the persistence feature value of each data to be analyzed.

[0052] Considering that high-frequency induction heating is essentially a continuous thermal process, temperature changes are subject to physical laws such as heat conduction, convection, and radiation. When deviations occur in the process or equipment used to produce the target anchor cable, the anomaly often does not immediately manifest as a pulse-like value, but rather as a slow shift and a gradual accumulation in the temperature data. Therefore, in this embodiment of the invention, based on the persistence of temperature monitoring data within the neighborhood range of each data point to be analyzed in step S101, the persistence characteristics of each data point to be analyzed are analyzed to obtain a persistence characteristic value for each data point to be analyzed. This is used to obtain suspected abnormal data with a high probability of anomaly. The stronger the persistence characteristic of the data point to be analyzed, that is, the larger the persistence characteristic value of the data point to be analyzed, the greater the probability that it belongs to abnormal temperature data when the process or equipment malfunctions.

[0053] Taking the s-th data point to be analyzed as an example, the method for obtaining the persistent feature value of the s-th data point to be analyzed is as follows:

[0054] In the temperature data sequence, if the number of temperature monitoring data points preceding the s-th data point to be analyzed is greater than or equal to a preset number, then the preset number of temperature monitoring data points preceding the s-th data point to be analyzed are acquired to form the left neighbor temperature data sequence of the s-th data point to be analyzed. The preset number needs to be set according to the production process and operating equipment of the target anchor cable, as well as the sampling frequency of the non-contact infrared thermometer. In this embodiment, the preset number is set to 6, but this is not limited. The implementer can set it according to the scenario. The second-order difference sequence of the left neighbor temperature data sequence is obtained, and all data in the second-order difference sequence of the left neighbor temperature data sequence are accumulated to obtain the left neighbor temperature change characteristic value of the s-th data point to be analyzed, denoted as […]. The smaller the data in the second-order difference sequence of the left-neighbor temperature data sequence, the more consistent the degree of change between each two adjacent temperature monitoring data in the left-neighbor temperature data sequence of the s-th data to be analyzed, and thus... The smaller;

[0055] In a temperature data sequence, if the number of temperature monitoring data points following the s-th data point to be analyzed is greater than or equal to 6, then the 6 temperature monitoring data points following the s-th data point to be analyzed are obtained to form the right neighbor temperature data sequence of the s-th data point to be analyzed. The second-order difference sequence of the right neighbor temperature data sequence is obtained, and all data points in the second-order difference sequence of the right neighbor temperature data sequence are summed to obtain the right neighbor temperature change characteristic value of the s-th data point to be analyzed, denoted as . The smaller the data in the second-order difference sequence of the right-neighbor temperature data sequence, the more consistent the degree of change between each two adjacent temperature monitoring data in the right-neighbor temperature data sequence of the s-th data to be analyzed, and thus... The smaller;

[0056] If the s-th data point to be analyzed only has temperature change characteristic values ​​in its left neighborhood, then the negative of these left neighborhood temperature change characteristic values ​​is used as the independent variable of an exponential function with the natural constant as its base, to obtain the persistence characteristic value of the s-th data point to be analyzed, denoted as . ,Right now Where exp() represents an exponential function with base to the natural constant. This represents the characteristic value of the temperature change in the left neighborhood of the s-th data point to be analyzed. The smaller the value, the more consistent the gradient of change between any two adjacent temperature monitoring data points in the left neighbor temperature data sequence of the s-th data point to be analyzed. In other words, the smaller the difference in the temperature monitoring data changes of the s-th data point to be analyzed within its neighborhood, and thus... The larger the value, the stronger the persistent characteristics of the s-th data point to be analyzed;

[0057] If the s-th data point to be analyzed only has a right-neighbor temperature change characteristic value, then the negative of the right-neighbor temperature change characteristic value is used as the independent variable of an exponential function with the natural constant as the base, to obtain the persistence characteristic value of the s-th data point to be analyzed, denoted as . ,Right now Where exp() represents an exponential function with base to the natural constant. This represents the characteristic value of the temperature change in the right neighborhood of the s-th data point to be analyzed. The smaller the value, the more consistent the gradient of change between any two adjacent temperature monitoring data points in the right neighbor temperature data sequence of the s-th data point to be analyzed. In other words, the smaller the difference in the temperature monitoring data changes of the s-th data point to be analyzed within its neighborhood, and thus... The larger the value, the stronger the persistent characteristics of the s-th data point to be analyzed;

[0058] If the s-th data point to be analyzed simultaneously possesses temperature change characteristic values ​​in both its left and right neighborhoods, then the minimum value among these two values ​​is obtained. The negative of this minimum value is used as the independent variable of an exponential function with the natural constant as its base, yielding the persistence characteristic value of the s-th data point to be analyzed, denoted as... ,Right now Where exp() represents an exponential function with base to the natural constant, and min() represents a minimum function. This represents the characteristic value of the temperature change in the left neighborhood of the s-th data point to be analyzed. This represents the characteristic value of temperature change in the right neighborhood of the s-th data point to be analyzed.

[0059] Similarly, the persistence feature value of each piece of data to be analyzed is obtained. Further, based on the production scenario and actual production process of the target anchor cable, a preset persistence feature value threshold is set. In this embodiment, the preset persistence feature value threshold is set to 0.7. This is not a limitation; implementers can set it according to the scenario. Data to be analyzed with persistence feature values ​​less than 0.7 are considered to lack high persistence characteristics and are therefore marked as normal temperature data; while data with persistence feature values ​​greater than or equal to 0.7 exhibit high persistence characteristics and are more likely to be abnormal temperature data, so they are marked as suspected abnormal data.

[0060] At this point, all suspected abnormal data of the target anchor cable during the high-frequency heating stage were obtained.

[0061] Step S103: Based on the difference in the persistence characteristic values ​​of each pair of adjacent suspected abnormal data, obtain the persistence stability of each suspected abnormal data. Based on the similarity between the residual data and the temperature data sequence within the local time range of each suspected abnormal data, obtain the matching degree of the residual data corresponding to each suspected abnormal data.

[0062] The suspected abnormal data obtained through step S102 is based on the continuous characteristics of temperature monitoring data within the neighborhood range of each data to be analyzed. However, during the high-frequency heating stage of the target anchor cable, there are normal process dynamics such as short-term process switching or equipment start-up and shutdown (e.g., PID or fuzzy control to fine-tune power in real time, thermal overheating of the gap between adjacent anchor cables, etc.). The temperature monitoring data under these normal process dynamics also show continuous characteristics, but these data do not belong to abnormal data. Therefore, it is still necessary to continue to analyze whether the suspected abnormal data is caused by deviations in the process or equipment during the high-frequency heating of the target anchor cable, or by normal process dynamics.

[0063] For deviations in the process or equipment during the high-frequency heating of the target anchor cable, such as local losses in the high-frequency induction heating coil or electromagnetic-thermal coupling delays in the high-frequency induction heating system, the temperature monitoring data caused by such anomalies will not exhibit occasional fluctuations, but rather continuous and stable anomalies with weak discontinuity. That is, the difference in the persistent characteristic value between two adjacent suspected anomaly data points is small, and the time interval between two adjacent suspected anomaly data points is short. In contrast, the continuous characteristics of temperature monitoring data caused by normal process dynamics such as real-time power fine-tuning by PID or fuzzy control and thermal overheating at the junction of adjacent anchor cables will vary significantly, exhibiting strong discontinuity. That is, the difference in the persistent characteristic value between two adjacent suspected anomaly data points is large, and the time interval between two adjacent suspected anomaly data points is long. Therefore, in this embodiment of the invention, the degree of continuous stability of each suspected anomaly data point is obtained based on the difference in the persistent characteristic value between every two adjacent suspected anomaly data points, which is used to reflect the probability that each suspected anomaly data point represents a true anomaly due to a deviation in the process or equipment.

[0064] Taking the u-th suspected anomalous data as an example, the formula for calculating the persistence stability of the u-th suspected anomalous data is as follows:

[0065] In the temperature data sequence, if the u-th suspected abnormal data only has a left neighbor temperature change feature value, or if the minimum of the left neighbor temperature change feature value and the right neighbor temperature change feature value of any suspected abnormal data is the left neighbor temperature change feature value, then the time corresponding to the u-th suspected abnormal data is subtracted by the preset maximum unit time span to obtain the target time. All temperature monitoring data between the time corresponding to the u-th suspected abnormal data and the target time are combined to form a neighborhood data sequence. The preset maximum unit time span is set to 6, which is not limited here and can be set by the implementer according to the specific scenario.

[0066] If the u-th suspected abnormal data only has a right neighbor temperature change feature value, or if the minimum of the left neighbor temperature change feature value and the right neighbor temperature change feature value of the u-th suspected abnormal data is the right neighbor temperature change feature value, then calculate the sum between the time corresponding to any suspected abnormal data and the preset maximum unit time span to obtain the target time, and form a neighborhood data sequence from the time corresponding to the u-th suspected abnormal data in the temperature data sequence and the target time.

[0067] For the acquisition of the neighborhood data sequence, that is: if the calculation of the persistence feature value of the u-th suspected abnormal data in step S102 depends on the temperature data sequence of the left neighborhood of the u-th suspected abnormal data, then the u-th suspected abnormal data and the 6 temperature monitoring data before it in the temperature data sequence are combined to form the neighborhood data sequence; if the calculation of the persistence feature value of the u-th suspected abnormal data in step S102 depends on the temperature data sequence of the right neighborhood of the u-th suspected abnormal data, then the u-th suspected abnormal data and the 6 temperature monitoring data after it in the temperature data sequence are combined to form the neighborhood data sequence.

[0068] In the neighborhood data sequence, calculate the absolute value of the difference between the persistent feature values ​​between every two adjacent suspected anomalous data, calculate the sum of all the absolute values ​​of the difference, and use the negative of the sum as the independent variable of the exponential function with the natural constant as the base to obtain the first stability of the u-th suspected anomalous data.

[0069] Calculate the average time interval between every two adjacent suspected abnormal data in the neighborhood data sequence, use the negative of the average time interval as the independent variable of an exponential function with the natural constant as the base, obtain the distribution characteristic value of the suspected abnormal data in the neighborhood data sequence, calculate the proportion of the number of all suspected abnormal data in the neighborhood data sequence to the number of all temperature monitoring data in the neighborhood data sequence, and use the mean between the proportion and the distribution characteristic value as the second stability of the u-th suspected abnormal data.

[0070] Calculate the mean between the first and second stability levels of the u-th suspected outlier data to obtain the persistence stability level of the u-th suspected outlier data.

[0071] In one implementation, the formula for calculating the persistence stability of the u-th suspected anomalous data is:

[0072]

[0073] in, This indicates the degree of persistence and stability of the u-th suspected outlier. This represents the number of all suspected anomalous data in the neighborhood data sequence of the u-th suspected anomalous data. This represents the (i+1)th suspected anomalous data point in the neighborhood data sequence of the u-th suspected anomalous data point. This represents the i-th suspected anomalous data point in the neighborhood data sequence of the u-th suspected anomalous data point. This represents the average time interval between any two adjacent suspected anomalies in the neighborhood data sequence of the u-th suspected anomaly. This represents the preset maximum unit time span, which is the number of data points in the neighborhood data sequence of the u-th suspected outlier. exp() represents an exponential function with the natural constant as the base. Represents the absolute value symbol.

[0074] It should be noted that, The smaller the value, the smaller the difference in persistent feature values ​​between two adjacent suspected anomaly data points, and the more stable the persistent features of each suspected anomaly data point within the neighborhood of the u-th suspected anomaly data point. The larger the value, the more likely the u-th suspected abnormal data is a real abnormal data caused by deviations in the process or equipment during the high-frequency heating of the target anchor cable; The smaller, The larger the value, the more numerous and concentrated the number of suspected anomalous data points are within the neighborhood of the u-th suspected anomalous data point. This also indicates a weaker discontinuity among the suspected anomalous data points within the neighborhood of the u-th suspected anomalous data point. The larger the value, the more likely the u-th suspected abnormal data is a real abnormal data point caused by deviations in the process or equipment during the high-frequency heating of the target anchor cable.

[0075] Thus, the persistence stability of the u-th suspected anomalous data point is obtained. Similarly, the persistence stability of each suspected anomalous data point is obtained.

[0076] Furthermore, considering that in the STL time series decomposition algorithm, the residual term itself belongs to the part of the original data that cannot be explained by trends and seasonality, and is usually a random disturbance or noise. If there is data in the temperature detection data sequence that originates from real anomalies, this part of the data cannot be decomposed by STL and will remain in the residual, exhibiting characteristics highly similar to the original data. Therefore, in this embodiment of the invention, based on the similarity between the residual data and the temperature data sequence within the local time range corresponding to each suspected anomaly data, the matching degree of the residual data corresponding to each suspected anomaly data is obtained. This is used to combine the persistence and stability of each suspected anomaly data to fully reflect the possibility that each suspected anomaly data belongs to a real anomaly caused by deviation in the process or equipment, thereby improving the accuracy of temperature monitoring results for anchor cables during the high-frequency heating stage.

[0077] Taking the u-th suspected outlier as an example, the specific steps to obtain the matching degree of the residual data corresponding to the u-th suspected outlier are as follows:

[0078] Obtain the target time corresponding to the u-th suspected anomaly data. Compile the residual data from all times between the u-th suspected anomaly data and the target time to form a residual sequence for the u-th suspected anomaly data within its local time range. Calculate the DTW distance between the residual sequence and the temperature data sequence. Use the negative of the DTW distance as the independent variable of an exponential function with the natural constant as the base to obtain the matching degree of the residual data corresponding to the u-th suspected anomaly data. The DTW distance is existing technology and will not be elaborated upon here.

[0079] In one implementation, the formula for calculating the matching degree of the residual data corresponding to the u-th suspected outlier is:

[0080]

[0081] in, This indicates the degree of matching between the residual data corresponding to the u-th suspected outlier. This represents the residual sequence of the time corresponding to the u-th suspected outlier within its local time range. Represents a temperature data sequence. Exp() represents the DTW distance between the residual sequence and the temperature data sequence at the time corresponding to the u-th suspected anomaly within its local time range, and exp() represents an exponential function with the natural constant as the base.

[0082] It should be noted that, The smaller the value, the more similar the residual sequence of the time corresponding to the u-th suspected outlier is to the temperature data sequence within its local time range, and thus... The larger the value, the higher the degree of matching between the residual sequence and the temperature data sequence of the time corresponding to the u-th suspected abnormal data within its local time range. The more likely the u-th suspected abnormal data is the real abnormal data caused by deviations in the process or equipment during the high-frequency heating of the target anchor cable.

[0083] Thus, the matching degree of the residual data corresponding to the u-th suspected outlier data is obtained. Similarly, the matching degree of the residual data corresponding to all suspected outlier data is obtained.

[0084] Step S104: Combining the persistence and stability of each suspected abnormal data and the matching degree of the residual data corresponding to each suspected abnormal data, obtain the abnormal confidence level of each suspected abnormal data. Based on the abnormal confidence level of each suspected abnormal data and the residual data at each time point, monitor the abnormal temperature data of the target anchor cable during the high-frequency heating stage.

[0085] After obtaining the persistence stability of each suspected outlier and the matching degree of the corresponding residual data in step S103, the anomaly confidence level of each suspected outlier is obtained by combining the persistence stability of each suspected outlier and the matching degree of the corresponding residual data. Specifically:

[0086] Taking the u-th suspected anomalous data as an example, the mean between the persistence stability of the u-th suspected anomalous data and the matching degree corresponding to the u-th suspected anomalous data is calculated to obtain the anomalous confidence of the u-th suspected anomalous data.

[0087] In one implementation, the formula for calculating the anomaly confidence of the u-th suspected anomalous data is:

[0088]

[0089] in, This represents the anomaly confidence level of the u-th suspected anomalous data point. This indicates the degree of persistence and stability of the u-th suspected outlier. This indicates the degree of matching of the residual data corresponding to the u-th suspected outlier.

[0090] It should be noted that, The larger, The larger the value, the stronger the persistence of the u-th suspected anomaly, and the more stable the persistence characteristic. Furthermore, the higher the degree of matching between the residual sequence corresponding to the u-th suspected anomaly and the temperature data sequence within its local time range, the better. The larger the value, the greater the likelihood that the u-th suspected abnormal data is a real abnormal data caused by deviations in the process or equipment during the high-frequency heating of the target anchor cable.

[0091] Similarly, the anomaly confidence level of all suspected abnormal data is obtained. A preset anomaly confidence level threshold is set based on process operation differences and requirements. Suspected abnormal data with an anomaly confidence level greater than or equal to the preset threshold are considered abnormal temperature data. In this embodiment, the preset anomaly confidence level threshold is set to 0.75, but this is not a limitation; implementers can set it according to the scenario, that is, mark suspected abnormal data with an anomaly confidence level greater than or equal to 0.75 as abnormal temperature data. Simultaneously, the temperature monitoring data corresponding to the residual data located outside the upper and lower limits in the box plot obtained in step S101 are also marked as abnormal temperature data.

[0092] Thus, all abnormal temperature data caused by process or equipment deviations during the high-frequency heating of the target anchor cable have been obtained. Relevant personnel can adjust the heating power in a timely manner based on all abnormal temperature data to ensure the production quality of the target anchor cable and equipment safety. Adjusting the heating power based on abnormal temperature data is an existing technology and will not be elaborated here.

[0093] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for real-time temperature monitoring during the production of high-frequency heated anchor cables, characterized in that, The method for real-time temperature monitoring during the production of high-frequency heated anchor cables includes: Temperature monitoring data of the target anchor cable at each moment during the high-frequency heating stage are obtained to form a temperature data sequence. The temperature data sequence is decomposed by STL to obtain the residual data at each moment. The data to be analyzed is obtained from the temperature data sequence based on the residual data at each moment using the box plot method. The method of using box plots to obtain the data to be analyzed from the temperature data sequence based on the residual data at each time point includes: The residual data at all times are combined into a residual sequence. A box plot is constructed based on the residual sequence to obtain the upper and lower limits. In the box plot, the time corresponding to the residual data between the upper and lower limits is recorded as the time to be analyzed. In the temperature data sequence, the temperature monitoring data corresponding to the time to be analyzed is recorded as the data to be analyzed. Based on the persistence of temperature monitoring data in the neighborhood range of each data to be analyzed in the temperature data sequence, obtain the persistence characteristic value of each data to be analyzed, and obtain at least two suspected abnormal data based on the persistence characteristic value of each data to be analyzed; The step of obtaining the persistence feature value of each data point to be analyzed based on the persistence of temperature monitoring data within its neighborhood range in the temperature data sequence includes: For any data to be analyzed, in the temperature data sequence, if the number of temperature monitoring data preceding any data to be analyzed is greater than or equal to a preset number, then the preset number of temperature monitoring data preceding any data to be analyzed in the temperature data sequence are obtained to form the left neighbor temperature data sequence of any data to be analyzed. The second-order difference sequence of the left neighbor temperature data sequence is obtained, and all data in the second-order difference sequence of the left neighbor temperature data sequence are accumulated to obtain the left neighbor temperature change characteristic value of any data to be analyzed. In a temperature data sequence, if the number of temperature monitoring data following any data to be analyzed is greater than or equal to the preset number, then the preset number of temperature monitoring data following any data to be analyzed in the temperature data sequence are obtained to form the right neighbor temperature data sequence of any data to be analyzed. The second-order difference sequence of the right neighbor temperature data sequence is obtained, and all data in the second-order difference sequence of the right neighbor temperature data sequence are accumulated to obtain the right neighbor temperature change characteristic value of any data to be analyzed. Based on the temperature change characteristic values ​​of the left and right neighbors of any data to be analyzed, obtain the persistence characteristic value of any data to be analyzed. Based on the difference in the persistence characteristic values ​​of each pair of adjacent suspected anomalous data, the persistence stability of each suspected anomalous data is obtained. Based on the similarity between the residual data of each suspected anomalous data and the temperature data sequence within its local time range, the matching degree of the residual data corresponding to each suspected anomalous data is obtained. By combining the persistence and stability of each suspected abnormal data point with the matching degree of the residual data corresponding to each suspected abnormal data point, the abnormal confidence level of each suspected abnormal data point is obtained. Based on the abnormal confidence level of each suspected abnormal data point and the residual data at each time point, the abnormal temperature data of the target anchor cable during the high-frequency heating stage is monitored.

2. The method for real-time temperature monitoring during the production of high-frequency heated anchor cables according to claim 1, characterized in that, The step of obtaining the persistence feature value of any data to be analyzed based on the temperature change feature values ​​of its left and right neighbors includes: If any of the data to be analyzed has only a left neighbor temperature change characteristic value, then the negative of the left neighbor temperature change characteristic value is used as the independent variable of an exponential function with the natural constant as the base, to obtain the persistence characteristic value of any of the data to be analyzed. If any of the data to be analyzed has only a right neighbor temperature change characteristic value, then the negative of the right neighbor temperature change characteristic value is used as the independent variable of an exponential function with the natural constant as the base, to obtain the persistence characteristic value of any of the data to be analyzed. If any data to be analyzed has both a left neighbor temperature change characteristic value and a right neighbor temperature change characteristic value, then the minimum value is obtained from the left neighbor temperature change characteristic value and the right neighbor temperature change characteristic value. The opposite of the minimum value is used as the independent variable of an exponential function with the natural constant as the base to obtain the persistence characteristic value of any data to be analyzed.

3. The method for real-time temperature monitoring during the production of high-frequency heated anchor cables according to claim 1, characterized in that, Based on the persistent characteristic value of each piece of data to be analyzed, at least two suspected abnormal data points are obtained, including: For any data to be analyzed, if the persistent characteristic value of the data to be analyzed is greater than or equal to a preset persistent characteristic value threshold, then the data to be analyzed is recorded as suspected abnormal data.

4. The method for real-time temperature monitoring during the production of high-frequency heated anchor cables according to claim 2, characterized in that, The step of obtaining the persistence stability of each suspected abnormal data point based on the difference in persistence characteristic values ​​between every two adjacent suspected abnormal data points includes: For any suspected abnormal data, in the temperature data sequence, if the suspected abnormal data only has a left neighbor temperature change feature value, or the minimum of the left neighbor temperature change feature value and the right neighbor temperature change feature value of the suspected abnormal data is the left neighbor temperature change feature value, then the time corresponding to the suspected abnormal data is subtracted by a preset maximum unit time span to obtain the target time. All temperature monitoring data between the time corresponding to the suspected abnormal data and the target time in the temperature data sequence are combined into a neighborhood data sequence. If any suspected abnormal data has only a right neighbor temperature change feature value, or if the minimum of the left neighbor temperature change feature value and the right neighbor temperature change feature value of any suspected abnormal data is the right neighbor temperature change feature value, then the sum between the time corresponding to any suspected abnormal data and the preset maximum unit time span is calculated to obtain the target time. All temperature monitoring data between the time corresponding to any suspected abnormal data and the target time in the temperature data sequence are combined into a neighborhood data sequence. In the neighborhood data sequence, calculate the absolute value of the difference between the persistent feature values ​​between every two adjacent suspected anomalous data, calculate the sum of all the absolute values ​​of the difference, and use the negative of the sum as the independent variable of an exponential function with the natural constant as the base to obtain the first stability of any suspected anomalous data. Calculate the average time interval between every two adjacent suspected abnormal data in the neighborhood data sequence, use the negative of the average time interval as the independent variable of an exponential function with the natural constant as the base, obtain the distribution characteristic value of the suspected abnormal data in the neighborhood data sequence, calculate the proportion of the number of all suspected abnormal data in the neighborhood data sequence to the number of all temperature monitoring data in the neighborhood data sequence, and use the mean between the proportion and the distribution characteristic value as the second stability of any suspected abnormal data. The mean between the first and second stability levels of any suspected anomalous data is calculated to obtain the persistent stability level of any suspected anomalous data.

5. The method for real-time temperature monitoring during the production of high-frequency heated anchor cables according to claim 4, characterized in that, The step of obtaining the matching degree of the residual data corresponding to each suspected anomaly data point based on the similarity between the residual data and the temperature data sequence within its local time range at the time corresponding to each suspected anomaly data point includes: For any suspected anomalous data, obtain the target time corresponding to the suspected anomalous data, and form a residual sequence of the time corresponding to the suspected anomalous data within its local time range by using the residual data of all times between the time corresponding to the suspected anomalous data and the target time. Calculate the DTW distance between the residual sequence and the temperature data sequence, and use the negative of the DTW distance as the independent variable of an exponential function with the natural constant as the base to obtain the matching degree of the residual data corresponding to the suspected anomalous data.

6. The method for real-time temperature monitoring during the production of high-frequency heated anchor cables according to claim 1, characterized in that, The process of combining the persistence and stability of each suspected anomaly data point with the matching degree of the residual data corresponding to each suspected anomaly data point to obtain the anomaly confidence level of each suspected anomaly data point includes: For any suspected anomalous data, calculate the average between the persistence and stability of the suspected anomalous data and the matching degree corresponding to the suspected anomalous data to obtain the anomalous confidence level of the suspected anomalous data.

7. The method for real-time temperature monitoring during the production of high-frequency heated anchor cables according to claim 1, characterized in that, The abnormal temperature data of the target anchor cable during the high-frequency heating stage is monitored based on the anomaly confidence level of each suspected abnormal data and the residual data at each time point, including: Suspected abnormal data with an abnormal confidence level greater than or equal to a preset abnormal confidence threshold are recorded as abnormal temperature data; The residual data of the target anchor cable at all times during the high-frequency heating stage are combined into a residual sequence. A box plot is constructed based on the residual sequence to obtain the upper and lower limits. In the box plot, the temperature monitoring data corresponding to the residual data outside the upper and lower limits is recorded as abnormal temperature data.

Images