A method and system for controlling the dew point of a furnace zone in an annealing furnace

Abstract

The present application relates to the technical field of annealing furnace dew point control, and specifically to a furnace zone dew point control method and system for an annealing furnace, the method comprising: collecting dew point data, temperature data and furnace pressure data; obtaining the fluctuation difference coefficient and speed difference coefficient of each dew point subsequence; calculating the data invariance index of the dew point sequence; obtaining the data change rate of the dew point sequence; obtaining the front proportion of the temperature sequence based on the difference between the mutation points in the temperature sequence and the dew point sequence; and then obtaining the temperature response coefficient between the temperature sequence and the dew point sequence; obtaining the rising subsequence and falling subsequence of each corresponding dew point peak; calculating the temperature response coefficient of the dew point sequence, and then combining the neural network to control the furnace zone dew point of the annealing furnace. The present application solves the problem that the existing control technology is difficult to cope with complex furnace section changes and sudden process parameter fluctuations, and realizes precise control of the dew point.

Description

Technical Field

[0001] The present application relates to the technical field of annealing furnace dew point control, and in particular to a furnace zone dew point control method and system for an annealing furnace. Background Art

[0002] Annealing is a common heat treatment process in the fields of metal processing and semiconductor manufacturing. During the annealing process, the material is heated to a certain temperature and then slowly cooled to change its internal structure, thereby achieving the purpose of improving the material properties. In the annealing furnace, the role of water vapor is to decarburize. The water vapor content in the furnace is characterized in the form of dew point (Note: similar to humidity, but dew point is temperature), and the water vapor content in the furnace can be controlled by adjusting the dew point. A higher dew point will cause the humidity in the furnace to be too high, which will cause problems such as oxidation of the material surface; a lower dew point may cause the humidity in the furnace to be too low, the surface of the material is too dry, and microcracks or other defects on the surface of the material. Therefore, in order to ensure the quality and efficiency of the annealing process, accurate control of the dew point in the furnace area is crucial.

[0003] Traditional annealing furnace dew point control methods include PID control method and fuzzy control method. PID control method is to control the temperature in the furnace by adjusting the heating power and wind speed, thereby indirectly controlling the dew point temperature; fuzzy control is to adjust the heating power and wind speed according to empirical rules and fuzzy reasoning to achieve the control of the dew point temperature. However, since the annealing furnace is composed of multiple furnace sections such as preheating section, heating section, soaking section, cooling section, equalization section, and outlet section, the requirements for dew point in different furnace sections are not consistent. The existing control algorithm is not flexible enough to respond to sudden changes in process parameters, and cannot adapt to the requirements for dew point in different furnace sections and different process parameters, and the control effect is not good. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a furnace zone dew point control method and system for an annealing furnace. The technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present application provides a method for controlling a furnace zone dew point of an annealing furnace, the method comprising the following steps:

[0006] Collect dew point data, temperature data and furnace pressure data of the heating furnace section in the annealing furnace, and construct a dew point sequence, a temperature sequence and a furnace pressure sequence;

[0007] Based on the discreteness of the local dew point data, the average of the differences in the discreteness of all the local dew point data, and the difference between the average of the differences between the local dew point data and the average of the differences of other local dew point data, the fluctuation difference coefficient and the speed change difference coefficient of each dew point subsequence are obtained;

[0008] Based on the concentrated distribution of the frequency of dew point data in the dew point sequence, the data invariance index of the dew point sequence is obtained; based on the values ​​of the fluctuation difference coefficient, the speed difference coefficient and the data invariance index, the data change rate of the dew point sequence is obtained;

[0009] Based on the difference between the mutation points in the temperature sequence and the dew point sequence, the top proportion of the temperature sequence is obtained; based on the top proportion, the discrete degree of the difference between the mutation points in the temperature sequence and the dew point sequence, and the trend distance between the temperature sequence and the dew point sequence, the temperature influence index between the temperature sequence and the dew point sequence is obtained;

[0010] Based on the proximity relationship between the peaks and valleys of the temperature sequence and the dew point sequence, the ascending subsequence and the descending subsequence of each peak are obtained;

[0011] Based on the ratio of the lengths of the rising subsequence and the falling subsequence of the peak value, the difference between the peak values ​​in the temperature sequence and the dew point sequence and the mean value of the temperature sequence, and the temperature influence index, the temperature response coefficient of the dew point sequence is obtained; based on the data change rate and the temperature response coefficient, the dew point of the furnace zone of the annealing furnace is controlled in combination with a neural network.

[0012] Furthermore, the method for obtaining the fluctuation difference coefficient and the speed change difference coefficient is as follows:

[0013] Using a sequence segmentation algorithm to obtain each dew point subsequence of the dew point sequence;

[0014] Calculate the coefficient of variation of each dew point subsequence, calculate the average of the absolute values ​​of the differences between the coefficients of variation of each dew point subsequence and all other dew point subsequences as the average fluctuation difference of each dew point subsequence, and calculate the product of the coefficient of variation of each dew point subsequence and the average fluctuation difference as the fluctuation difference coefficient of each dew point subsequence;

[0015] The first-order difference sequence of each dew point subsequence is obtained, and the mean of the absolute values ​​of all elements in the first-order difference sequence is taken as the change speed of each dew point subsequence; the mean of the absolute values ​​of the differences between the change speed of each dew point subsequence and the change speeds of all other dew point subsequences is calculated as the change speed difference coefficient of each dew point subsequence.

[0016] Furthermore, the method for obtaining the data invariance index is:

[0017] The ratio of the number of occurrences of each element in the dew point sequence to the total number of elements in the dew point sequence is recorded as the frequency of occurrence of each element; the frequency of occurrence of different elements is used as the input of the clustering algorithm, and each cluster is output;

[0018] Calculate the mean of the occurrence frequency of all elements in each cluster, and take the largest mean among all clusters as the data invariant index of the dew point series.

[0019] Further, the formula of the speed change rate is: ; Where A is the data change rate of the dew point sequence; n is the number of dew point subsequences; is the fluctuation difference coefficient of the i-th dew point subsequence; is the speed difference coefficient of the i-th dew point subsequence; D is the data invariance index.

[0020] Furthermore, the method for obtaining the front ratio is:

[0021] The mutation point detection algorithm is used to obtain the mutation points in the temperature sequence and the dew point sequence, and the positions of the temperature mutation points in the temperature sequence and the dew point mutation points in the dew point sequence are respectively constructed in order from small to large to form a temperature mutation position sequence sequence and a dew point mutation position sequence sequence;

[0022] Calculate the difference between the elements in the temperature mutation sequence and the elements in the same position in the dew point mutation sequence, and count the number of differences less than 0 among all the differences. The ratio of the number of differences less than 0 to the number of all differences is recorded as the front proportion of the temperature sequence.

[0023] Furthermore, the method for obtaining the temperature impact index is:

[0024] The variance between all differences is recorded as the mutation asynchrony index between the temperature series and the dew point series;

[0025] The temperature sequence and dew point sequence are decomposed using a sequence decomposition algorithm, and the temperature trend sequence and dew point trend sequence are output.

[0026] Calculate the distance between the temperature trend sequence and the dew point trend sequence, and record it as the trend change coefficient between the temperature sequence and the dew point sequence;

[0027] The formula for the temperature impact index is: ; In the formula, F is the temperature influence index of the dew point sequence; G is the front proportion of the temperature sequence; H is the mutation asynchronous index between the temperature sequence and the dew point sequence; d is the trend change coefficient between the temperature sequence and the dew point sequence.

[0028] Further, the obtaining of the ascending subsequence and the descending subsequence of each peak value includes:

[0029] All peaks and valleys in the temperature sequence and dew point sequence are obtained through the peak and valley detection algorithm;

[0030] For each peak, obtain the valley values ​​of the two nearest neighbors before and after each peak, record the nearest neighbor valley value that is smaller than the position of each peak as the leading valley value of each peak, and record the nearest neighbor valley value that is larger than the position of each peak as the trailing valley value of each peak; record the subsequence composed of the data from the leading valley value to each peak as the ascending subsequence of each peak; record the subsequence composed of the data from each peak to the trailing valley value as the descending subsequence of each peak.

[0031] Further, the method for obtaining the temperature response coefficient is:

[0032] In the dew point sequence, find a peak value that is greater than the u-th peak value in the temperature sequence and has the smallest difference with the u-th peak value, and record this peak value as the u-th corresponding dew point peak value;

[0033] The data length ratio between the rising subsequence of the u-th peak in the temperature sequence and the rising subsequence of the u-th corresponding dew point peak in the dew point sequence is taken as the rising time ratio; the data length ratio between the falling subsequence of the u-th peak in the temperature sequence and the falling subsequence of the u-th corresponding dew point peak in the dew point sequence is taken as the falling time ratio; the average of the rising time ratio and the falling time ratio is taken as the corresponding time difference coefficient of the u-th peak;

[0034] In the temperature sequence, the difference between the u-th peak value and the mean value of the temperature sequence is recorded as the temperature rise coefficient of the u-th peak value; in the dew point sequence, the difference between the u-th corresponding dew point peak value and the mean value of the dew point sequence is recorded as the dew point rise coefficient of the u-th corresponding dew point peak value; the ratio of the temperature rise coefficient of the u-th peak value to the dew point rise coefficient of the u-th corresponding dew point peak value is recorded as the corresponding numerical difference coefficient of the u-th peak value;

[0035] The formula for the temperature response coefficient is: ; Where K is the temperature response coefficient of the dew point sequence; F is the temperature influence index of the dew point sequence, is the mean of the corresponding time difference coefficients of all peaks in the temperature series; is the mean of the corresponding numerical difference coefficients of all peaks in the temperature series.

[0036] Further, the control of the furnace zone dew point of the annealing furnace based on the data change rate and the temperature response coefficient in combination with a neural network includes:

[0037] The furnace pressure response coefficient between the dew point sequence and the furnace pressure sequence is obtained by the same method as the temperature response coefficient;

[0038] The sum of the temperature response coefficient and the furnace pressure response coefficient is recorded as the total impact index, the ratio of the temperature response coefficient to the total impact index is used as the temperature weight, and the ratio of the furnace pressure response coefficient to the total impact index is used as the furnace pressure weight;

[0039] The dew point sequence, temperature sequence, furnace pressure sequence, data change rate of the dew point sequence, temperature weight, and furnace pressure weight are used as inputs of the BP neural network, and the predicted dew point at the next moment is output. The dew point is input into the dew point controller, and the ratio between the dew point at the next moment and the dew point at the current moment is used as the adjustment coefficient. The product of the humidity value at the current moment and the adjustment coefficient is used as the set value of the humidity controller at the next moment.

[0040] In a second aspect, an embodiment of the present application further provides a furnace zone dew point control system for an annealing furnace, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above-mentioned methods when executing the computer program.

[0041] This application has at least the following beneficial effects:

[0042] This application evaluates the change frequency and fluctuation degree of data in the dew point sequence by constructing the data change rate, so as to better understand the law of dew point change; at the same time, it constructs the temperature influence index to preliminarily evaluate the impact of dew point on temperature; then, it constructs the temperature response coefficient to further evaluate the response time and degree of dew point to temperature changes.

[0043] This application makes an accurate prediction of the dew point at the next moment based on the degree of fluctuation of the dew point sequence itself and the degree of influence of the surrounding environment, and adjusts the water vapor content in the mixed gas according to the prediction result. Based on this method, this application can realize adaptive control of the dew point under different working conditions, solve the problem that the existing control technology is difficult to cope with complex furnace section changes and sudden process parameter fluctuations, and realize accurate control of the dew point. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0045] Figure 1 A flowchart of a method for controlling the dew point of a furnace zone of an annealing furnace provided in one embodiment of the present application;

[0046] Figure 2 This is a schematic diagram of the high frequency of occurrence of dew point elements;

[0047] Figure 3 This is a schematic diagram of the low frequency of occurrence of dew point elements. DETAILED DESCRIPTION

[0048] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the dew point control method and system for the furnace zone of an annealing furnace proposed in accordance with the present application, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0049] Unless defined otherwise, 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 belongs.

[0050] The following is a detailed description of a specific scheme of a furnace zone dew point control method and system for an annealing furnace provided by the present application in conjunction with the accompanying drawings.

[0051] See also Figure 1 , which shows a flow chart of a method for controlling the dew point of a furnace zone of an annealing furnace provided by an embodiment of the present application, the method comprising the following steps:

[0052] Step S1, collecting dew point data, temperature data and furnace pressure data of the heating furnace section in the annealing furnace, and constructing a dew point sequence, a temperature sequence and a furnace pressure sequence.

[0053] This embodiment takes the heating furnace section (RTF) in the annealing furnace as an example for analysis. The dew point meter, temperature sensor, and furnace pressure transmitter placed inside the heating furnace section are used to obtain the dew point data, temperature data, and furnace pressure data of the heating furnace section of the annealing furnace in real time. Multiple sensors collect data synchronously and in real time, with a collection time of 1 hour and a collection interval of 5 seconds.

[0054] The collected dew point data, temperature data and furnace pressure data are normalized. This embodiment adopts the maximum and minimum value normalization method. The implementer can choose other normalization methods according to the actual situation.

[0055] At this point, the preprocessed dew point data, temperature data and furnace pressure data are obtained.

[0056] The collected data is uploaded to the annealing furnace control system, which includes a data processing module and an industrial control module. The collected data is processed and predicted by the data processing module, and the annealing furnace is controlled by the industrial control module, which includes but is not limited to control units such as dew point controller, steam controller, moisture controller, dry gas controller, and mixed gas flow meter.

[0057] The preprocessed dew point data, temperature data and furnace pressure data are sorted in the positive order of acquisition time to obtain the dew point sequence, temperature sequence and furnace pressure sequence respectively.

[0058] Step S2, by constructing a data change rate, evaluate the frequency of changes in the data in the dew point sequence; then construct a temperature influence index to preliminarily evaluate the impact of temperature on the dew point; and then construct a temperature response coefficient to further quantify the response of the dew point to temperature changes.

[0059] In the annealing process, the annealing furnace uses water vapor in the protective gas to reduce the carbon content to achieve the purpose of decarburization. The dew point of the humidified mixed gas is affected by the humidification temperature, and the water vapor in the mixed gas is also continuously consumed due to the participation in the decarburization reaction, so the dew point in the annealing furnace is constantly changing. In order to achieve accurate control of the dew point, it is first necessary to know the fluctuation of the dew point itself.

[0060] Since methods such as variance and standard deviation are used to evaluate the overall fluctuation of data, it is difficult to evaluate whether the local changes of data are fluctuating rapidly; therefore, in order to evaluate whether the local changes of the dew point series data are rapid, it is necessary to perform local processing on the dew point series.

[0061] The dew point sequence is used as the input of the BG (Bernaola Galvan) sequence segmentation algorithm, and the output of the BG segmentation algorithm is each dew point subsequence. The coefficient of variation of each dew point subsequence is calculated, and the average of the absolute values ​​of the differences between the coefficient of variation of each dew point subsequence and all other dew point subsequences is calculated as the average fluctuation difference of each dew point subsequence. The product of the coefficient of variation of each dew point subsequence and the average fluctuation difference is calculated as the fluctuation difference coefficient of each dew point subsequence.

[0062] The first-order difference sequence of each dew point subsequence is obtained, and the mean of the absolute values ​​of all elements in the first-order difference sequence is taken as the change speed of each dew point subsequence. The mean of the absolute value of the difference between the change speed of each dew point subsequence and the change speed of all other dew point subsequences is calculated as the speed difference coefficient of each dew point subsequence.

[0063] The ratio of the number of occurrences of each element in the dew point sequence to the total number of elements in the dew point sequence is recorded as the frequency of occurrence of each element; the frequency of occurrence of each element is used as the input of the DPC density peak clustering algorithm, and the cross-validation method is used to obtain the cutoff distance of the DPC clustering algorithm. The output of the DPC clustering algorithm is J clusters. The BG sequence segmentation algorithm, the coefficient of variation, the DPC clustering algorithm and the cross-validation method are well-known technologies and will not be described here. It should be noted that for the input of the DPC clustering algorithm, the frequency of occurrence of the same element is only input once.

[0064] Calculate the mean of the occurrence frequency of all elements in each cluster, and take the largest mean among all clusters as the data invariance index of the dew point sequence. The larger the data invariance index, the higher the occurrence frequency of the dew point element corresponding to the cluster with the largest element mean, which means that the same element appears more times in the dew point sequence, the smaller the degree of change of the elements in the dew point sequence, and the more consistent the elements are. Figure 2 As shown, Figure 2 The schematic diagram of the high frequency of dew point elements is shown in Figure 1, while the schematic diagram of the low frequency of dew point elements is shown in Figure 2. Figure 3 As shown. Figure 2 as well as Figure 3 In the figure, the horizontal axis is the clusters, and the vertical axis is the data invariance index. Figure 2 and Figure 3 It reflects the difference between normal clustering and abnormal clustering.

[0065] According to the above analysis, the data change rate of the dew point sequence is constructed, and the formula is: ; Where A is the data change rate of the dew point sequence; n is the number of dew point subsequences; is the fluctuation difference coefficient of the i-th dew point subsequence; is the speed difference coefficient of the i-th dew point subsequence; D is the data invariance index.

[0066] It should be noted that if the fluctuation degree between the i-th dew point subsequence and other subsequences is more consistent, and the coefficient of variation of the i-th dew point subsequence is smaller, it means that the data fluctuation within each dew point subsequence is relatively stable, so the fluctuation difference coefficient of the i-th dew point subsequence is If the data change degree between the i-th dew point subsequence and other subsequences is relatively consistent, and no drastic change occurs, then the change speed of each dew point subsequence is relatively consistent, so the speed difference coefficient of the i-th dew point subsequence is The greater the data stationarity of the dew point sequence, the higher the consistency of the internal elements, and the greater the data invariance index of the dew point sequence. Therefore, if the data change rate A of the dew point sequence is smaller, it means that the data fluctuation of the dew point sequence is smaller, and the possibility of frequent data changes is lower.

[0067] If the fluctuation difference between the i-th dew point subsequence and other dew point subsequences is large, and the fluctuation degree itself is large, then the fluctuation difference coefficient of the i-th dew point subsequence is If the change speed between the i-th dew point subsequence and other dew point subsequences is more inconsistent, the speed difference coefficient of the i-th dew point subsequence is The larger the dew point sequence data is, the smaller the data invariance index is. Therefore, if the data change rate A is larger, the data fluctuation of the dew point sequence is larger and the internal elements are more inconsistent.

[0068] So far, the data change rate of the dew point sequence has been obtained, and the data fluctuation degree of the dew point sequence has been evaluated by the data change rate.

[0069] When the temperature rises, the mixed gas can accommodate more water vapor, and the dew point becomes relatively high. On the contrary, as the temperature decreases, the water vapor in the mixed gas gradually liquefies, the water vapor becomes less, and the dew point becomes relatively low. At the same time, under normal circumstances, the furnace pressure in the annealing furnace is relatively stable, and the impact on the dew point is small at this time. If the furnace pressure fluctuates, the dew point will also fluctuate violently. Since the dew point in the mixed gas is affected by temperature and furnace pressure, the dew point can be controlled more accurately by calculating the degree of influence between the dew point and temperature and furnace pressure.

[0070] Based on the above analysis, the PELT mutation point detection algorithm is used to obtain the mutation points in the temperature sequence and dew point sequence, and according to the position order of each temperature mutation point in the temperature sequence and the position order of each dew point mutation point in the dew point sequence, the temperature mutation position sequence and dew point mutation position sequence are constructed in ascending order.

[0071] Calculate the difference between the elements in the temperature mutation sequence and the elements in the same position in the dew point mutation sequence, and count the number of differences less than 0 among all the differences. The ratio of the number of differences less than 0 to the number of all differences is recorded as the front proportion G of the temperature sequence. The larger the front proportion G, the greater the possibility that the temperature has suddenly changed before the dew point has suddenly changed; conversely, the smaller the possibility that the temperature has suddenly changed.

[0072] Furthermore, the variance between all differences is recorded as the mutation asynchronous index H between the temperature series and the dew point series; the smaller the mutation asynchronous index H is, the more consistent the interval between each temperature mutation time and the dew point mutation time is, which means that the dew point is more affected by temperature; conversely, the less affected the dew point is by temperature.

[0073] It should be noted that: if the lengths of the temperature mutation sequence and the dew point mutation sequence are inconsistent, the end of the shorter sequence is padded with 0 to make its length consistent with the longer sequence.

[0074] The temperature series and dew point series are decomposed by the STL (Seasonal and Trend decomposition using Loess) decomposition algorithm. The output of the STL algorithm is the trend sequence of the temperature series and the dew point series, which are recorded as the temperature trend sequence and the dew point trend sequence respectively.

[0075] The DTW distance can be used to calculate the similarity between two sequences without being affected by the acquisition time; therefore, the DTW distance between the temperature trend sequence and the dew point trend sequence is obtained and recorded as the trend change coefficient d between the temperature sequence and the dew point sequence. The smaller the trend change coefficient d, the more consistent the trend change between the temperature sequence and the dew point sequence. The PELT mutation point detection algorithm, the STL decomposition algorithm, and the DTW distance are well-known technologies and will not be described here.

[0076] Furthermore, the temperature impact index is constructed, and the formula is: ; In the formula, F is the temperature influence index of the dew point sequence; G is the front proportion of the temperature sequence; H is the mutation asynchronous index between the temperature sequence and the dew point sequence; d is the trend change coefficient between the temperature sequence and the dew point sequence.

[0077] It should be noted that if the forward proportion G is larger, it means that before each dew point sequence mutates, the possibility that the temperature sequence has mutated is greater; if the mutation asynchronous index H is smaller, it means that the time interval between each mutation of the temperature sequence and the dew point sequence is relatively consistent; if the trend change coefficient d is smaller, it means that the data trend changes of the temperature sequence and the dew point sequence are more consistent and the similarity is greater. Therefore, if the temperature influence index F is larger, it means that the dew point sequence is more affected by the temperature sequence, and when predicting the dew point, it is more necessary to pay attention to the change of temperature data.

[0078] If the leading proportion G is smaller and the mutation asynchronous index H is larger, it means that there is no obvious time difference between the mutation time of the dew point sequence and the temperature sequence, and the mutation of the temperature sequence will not cause the mutation of the dew point sequence; if the trend change coefficient d is larger, it means that the data trend difference between the dew point sequence and the temperature sequence is larger, and there is no similar data trend. Therefore, if the temperature influence index F is smaller, it means that the dew point sequence is less affected by the temperature sequence, and there is no need to pay attention to the change of temperature data when predicting the dew point.

[0079] All peaks and valleys in the temperature sequence and dew point sequence are obtained through the peak and valley detection algorithm, and the temperature sequence and dew point sequence are divided into multiple subsequences through each peak and each valley. In the temperature sequence, taking the u-th peak as an example, the valley values ​​of the two nearest neighbors before and after the u-th peak are obtained, and the nearest neighbor valley value that is less than the u-th peak position is recorded as the leading valley value, and the nearest neighbor valley value that is greater than the u-th peak position is recorded as the trailing valley value.

[0080] The subsequence consisting of the data between the preceding valley value and the u-th peak value is recorded as the ascending subsequence of the u-th peak value; the subsequence consisting of the data between the u-th peak value and the following valley value is recorded as the descending subsequence of the u-th peak value.

[0081] It should be noted that, in special cases, if the u-th peak is located at the beginning or end of the temperature sequence, and there is no valley before or after the u-th peak, the subsequence formed from the first element of the sequence to the u-th peak is recorded as the rising subsequence of the u-th peak, and the subsequence formed from the u-th peak to the last element of the sequence is recorded as the falling subsequence of the u-th peak; neither the rising subsequence nor the last subsequence contains peaks or valleys.

[0082] In the same way, the rising subsequences and falling subsequences of all peak values ​​in the temperature sequence and the dew point sequence are obtained.

[0083] In the dew point sequence, find a peak value that is greater than the u-th peak value in the temperature sequence and has the smallest difference with the u-th peak value, and record this peak value as the u-th corresponding dew point peak value. It should be noted that if the u-th peak value in the temperature sequence does not have a corresponding dew point peak value in the dew point sequence, the u-th peak value does not participate in the calculation.

[0084] The data length ratio between the rising subsequence of the u-th peak in the temperature sequence and the rising subsequence of the u-th corresponding dew point peak in the dew point sequence is the rising time ratio; the data length ratio between the falling subsequence of the u-th peak in the temperature sequence and the falling subsequence of the u-th corresponding dew point peak in the dew point sequence is the falling time ratio. The average of the rising time ratio and the falling time ratio is recorded as the corresponding time difference coefficient of the u-th peak.

[0085] In the temperature series, the difference between the u-th peak and the mean of the temperature series is recorded as the temperature rise coefficient of the u-th peak; in the dew point series, the difference between the u-th corresponding dew point peak and the mean of the dew point series is recorded as the dew point rise coefficient of the u-th corresponding dew point peak. The ratio of the temperature rise coefficient of the u-th peak to the dew point rise coefficient of the u-th corresponding dew point peak is recorded as the corresponding numerical difference coefficient of the u-th peak.

[0086] Furthermore, based on the above analysis, the temperature response coefficient is constructed, and the formula is: ; Where K is the temperature response coefficient of the dew point sequence; F is the temperature influence index of the dew point sequence, is the time difference mean, expressed as the mean of the corresponding time difference coefficients of all peaks in the temperature series; is the mean of the numerical differences, expressed as the mean of the corresponding numerical difference coefficients of all peaks in the temperature series.

[0087] It should be noted that the time difference mean Reflects the average difference in the time required for a temperature change to cause a dew point change. The larger the value, the longer it takes for the dew point to change after the temperature changes, and the weaker the impact; the mean value of the difference It reflects the average difference in the value of dew point change caused by temperature change. The larger the temperature response index is, the more likely it is that the dew point will change only when the temperature changes significantly. Therefore, if the temperature response index is smaller, it means that the effect of temperature on the dew point is weak, and when the temperature changes, the dew point needs longer to respond.

[0088] Step S3, based on the data change rate and the temperature response coefficient, the dew point of the furnace zone of the annealing furnace is controlled in combination with the neural network.

[0089] The furnace pressure response coefficient of the dew point sequence is calculated according to the dew point sequence and the furnace pressure sequence using the calculation method and principle of the temperature response coefficient. The sum of the temperature response coefficient and the furnace pressure response coefficient is recorded as the total influence index, the ratio of the temperature response coefficient to the total influence index is used as the temperature weight, and the ratio of the furnace pressure response coefficient to the total influence index is used as the furnace pressure weight.

[0090] The dew point sequence, temperature sequence, furnace pressure sequence, data change rate of the dew point sequence, temperature weight, and furnace pressure weight are used as the input of the BP neural network, and the input layer has a total of 6 input nodes; the output of the BP neural network is the predicted dew point at the next moment, and the output layer has only one output node; the optimization algorithm of the BP neural network is the SGD algorithm, and the loss function is MSE; the division of the training set and the test set is 7:3, the hidden layer is 5 layers, and the number of iterations is 60 times. The BP neural network is a well-known technology and will not be described here.

[0091] The predicted dew point at the next moment is obtained through the BP neural network and input into the dew point controller. The dew point controller records the ratio between the dew point at the next moment and the dew point at the current moment as the adjustment coefficient. At the same time, the dew point controller is connected to the moisture controller, and the product of the moisture value at the current moment and the adjustment coefficient is used as the set value of the moisture controller at the next moment, so that the dew point can be changed in real time according to the change of process parameters, thereby improving the control accuracy of the dew point. A furnace zone dew point control method and system for an annealing furnace is realized.

[0092] Based on the same inventive concept as the above method, an embodiment of the present application also provides a furnace zone dew point control system for an annealing furnace, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the steps of a furnace zone dew point control method for an annealing furnace are implemented in any one of the above-mentioned methods for dew point control of the furnace zone of an annealing furnace.

[0093] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0094] The various embodiments in the present application are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0095] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for controlling the dew point of an annealing furnace, characterized in that: The method comprises the following steps: Collect dew point data, temperature data and furnace pressure data of the heating furnace section in the annealing furnace, and construct a dew point sequence, a temperature sequence and a furnace pressure sequence; Based on the discreteness of the local dew point data, the average of the differences in the discreteness of all the local dew point data, and the difference between the average of the differences between the local dew point data and the average of the differences of other local dew point data, the fluctuation difference coefficient and the speed change difference coefficient of each dew point subsequence are obtained; Based on the concentrated distribution of the frequency of dew point data in the dew point sequence, the data invariance index of the dew point sequence is obtained; based on the values ​​of the fluctuation difference coefficient, the speed difference coefficient and the data invariance index, the data change rate of the dew point sequence is obtained; Based on the difference between the mutation points in the temperature sequence and the dew point sequence, the top proportion of the temperature sequence is obtained; based on the top proportion, the discrete degree of the difference between the mutation points in the temperature sequence and the dew point sequence, and the trend distance between the temperature sequence and the dew point sequence, the temperature influence index between the temperature sequence and the dew point sequence is obtained; Based on the proximity relationship between the peaks and valleys of the temperature sequence and the dew point sequence, the ascending subsequence and the descending subsequence of each peak are obtained; Based on the ratio of the lengths of the rising subsequence and the falling subsequence of the peak value, the difference between the peak values ​​in the temperature sequence and the dew point sequence and the mean value of the temperature sequence, and the temperature influence index, the temperature response coefficient of the dew point sequence is obtained; based on the data change rate and the temperature response coefficient, the dew point of the furnace zone of the annealing furnace is controlled in combination with the neural network; The method for obtaining the fluctuation difference coefficient and the speed change difference coefficient is as follows: Using a sequence segmentation algorithm to obtain each dew point subsequence of the dew point sequence; Calculate the coefficient of variation of each dew point subsequence, calculate the average of the absolute values ​​of the differences between the coefficients of variation of each dew point subsequence and all other dew point subsequences as the average fluctuation difference of each dew point subsequence, and calculate the product of the coefficient of variation of each dew point subsequence and the average fluctuation difference as the fluctuation difference coefficient of each dew point subsequence; Obtain the first-order difference sequence of each dew point subsequence, and take the average of the absolute values ​​of all elements in the first-order difference sequence as the change speed of each dew point subsequence; calculate the average of the absolute values ​​of the difference between the change speed of each dew point subsequence and the change speeds of all other dew point subsequences as the change speed difference coefficient of each dew point subsequence; The method for obtaining the data invariance index is: The ratio of the number of occurrences of each element in the dew point sequence to the total number of elements in the dew point sequence is recorded as the frequency of occurrence of each element; the frequency of occurrence of different elements is used as the input of the clustering algorithm, and each cluster is output; Calculate the mean of the occurrence frequency of all elements in each cluster, and take the largest mean among all clusters as the data invariant index of the dew point series; The formula for the data change rate is: ; Where A is the data change rate of the dew point sequence; n is the number of dew point subsequences; is the fluctuation difference coefficient of the i-th dew point subsequence; is the speed difference coefficient of the i-th dew point subsequence; D is the data invariance index; The method for obtaining the front ratio is: The mutation point detection algorithm is used to obtain the mutation points in the temperature sequence and the dew point sequence, and the positions of the temperature mutation points in the temperature sequence and the dew point mutation points in the dew point sequence are respectively constructed in order from small to large to form a temperature mutation position sequence sequence and a dew point mutation position sequence sequence; Calculate the difference between the elements in the temperature mutation sequence and the elements in the same position in the dew point mutation sequence, and count the number of differences less than 0 among all the differences. The ratio of the number of differences less than 0 to the number of all differences is recorded as the front proportion of the temperature sequence. The method for obtaining the temperature impact index is: The variance between all differences is recorded as the mutation asynchrony index between the temperature series and the dew point series; Use sequence decomposition algorithm to decompose temperature sequence and dew point sequence, and output temperature trend sequence and dew point trend sequence; Calculate the distance between the temperature trend sequence and the dew point trend sequence, and record it as the trend change coefficient between the temperature sequence and the dew point sequence; The formula for the temperature impact index is: ; In the formula, F is the temperature influence index of the dew point sequence; G is the front proportion of the temperature sequence; H is the mutation asynchronous index between the temperature sequence and the dew point sequence; d is the trend change coefficient between the temperature sequence and the dew point sequence; The step of obtaining the ascending subsequence and the descending subsequence of each peak value includes: All peaks and valleys in the temperature sequence and dew point sequence are obtained through the peak and valley detection algorithm; For each peak, obtain the valley values ​​of the two nearest neighbors before and after each peak, record the nearest neighbor valley value that is smaller than the position of each peak as the preceding valley value of each peak, and record the nearest neighbor valley value that is larger than the position of each peak as the following valley value of each peak; record the subsequence composed of data from the preceding valley value to each peak as the ascending subsequence of each peak; record the subsequence composed of data from each peak to the following valley value as the descending subsequence of each peak; The method for obtaining the temperature response coefficient is: In the dew point sequence, find a peak value that is greater than the u-th peak value in the temperature sequence and has the smallest difference with the u-th peak value, and record this peak value as the u-th corresponding dew point peak value; The data length ratio between the rising subsequence of the u-th peak in the temperature sequence and the rising subsequence of the u-th corresponding dew point peak in the dew point sequence is taken as the rising time ratio; the data length ratio between the falling subsequence of the u-th peak in the temperature sequence and the falling subsequence of the u-th corresponding dew point peak in the dew point sequence is taken as the falling time ratio; the average of the rising time ratio and the falling time ratio is taken as the corresponding time difference coefficient of the u-th peak; In the temperature sequence, the difference between the u-th peak value and the mean value of the temperature sequence is recorded as the temperature rise coefficient of the u-th peak value; in the dew point sequence, the difference between the u-th corresponding dew point peak value and the mean value of the dew point sequence is recorded as the dew point rise coefficient of the u-th corresponding dew point peak value; the ratio of the temperature rise coefficient of the u-th peak value to the dew point rise coefficient of the u-th corresponding dew point peak value is recorded as the corresponding numerical difference coefficient of the u-th peak value; The formula for the temperature response coefficient is: ; Where K is the temperature response coefficient of the dew point sequence; F is the temperature influence index of the dew point sequence, is the mean of the corresponding time difference coefficients of all peaks in the temperature series; is the mean of the corresponding numerical difference coefficients of all peak values ​​in the temperature series; The method of controlling the dew point of the furnace zone of the annealing furnace based on the data change rate and the temperature response coefficient in combination with a neural network includes: The furnace pressure response coefficient between the dew point sequence and the furnace pressure sequence is obtained by the same method as the temperature response coefficient; The sum of the temperature response coefficient and the furnace pressure response coefficient is recorded as the total impact index, the ratio of the temperature response coefficient to the total impact index is used as the temperature weight, and the ratio of the furnace pressure response coefficient to the total impact index is used as the furnace pressure weight; The dew point sequence, temperature sequence, furnace pressure sequence, data change rate of the dew point sequence, temperature weight, and furnace pressure weight are used as inputs of the BP neural network, and the predicted dew point at the next moment is output. The dew point is input into the dew point controller, and the ratio between the dew point at the next moment and the dew point at the current moment is used as the adjustment coefficient. The product of the humidity value at the current moment and the adjustment coefficient is used as the set value of the humidity controller at the next moment.

2. A furnace zone dew point control system for an annealing furnace, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method for controlling the dew point of a furnace zone of an annealing furnace as claimed in claim 1 are implemented.

Images