CN112430727B - A kind of continuous annealing furnace furnace temperature early warning method and system - Google Patents

Abstract

The method comprises the steps of collecting data, generating samples according to different collection time periods, checking the data quality of all the samples to generate a sample set, extracting the characteristics of the sample set, judging the stability and the degradation trend of the sample set, and predicting the furnace temperature change condition of each furnace section in the annealing furnace based on a BP neural network algorithm. The method combines the traditional alarm mode with the SPC process control statistics and neural network prediction algorithm mode, can predict the health degree of the furnace while realizing the ultralimit alarm, and predicts the furnace temperature by adopting the BP neural network algorithm, thereby providing a brand-new temperature health monitoring and early warning method for the continuous annealing furnace of the hot galvanizing unit. The early warning system using the method can assist operators in giving an alarm in advance and finding fault points in time when problems occur, so as to strive for early finding and early intervention.

Description

A kind of continuous annealing furnace furnace temperature early warning method and system

technical field

The invention relates to the field of cold rolling technology, in particular to a furnace temperature (annealing furnace temperature) early warning method and system for a continuous annealing furnace (hot-dip galvanizing unit annealing furnace), which is suitable for automatic collection and storage of furnace data in a horizontal continuous annealing furnace, And through the BP neural network algorithm and the SPC process control graph theoretical model, the furnace steady state is early-warned and monitored.

Background technique

Generally, a continuous annealing furnace is a device in which a heat treatment process is performed to raise or lower the temperature of a room or low temperature metal strip according to a preset temperature scheme to obtain desired material properties. The continuous annealing furnace is mainly divided into a heating section and a cooling section. The heating section is divided into a preheating section, a heating section and a soaking section, and the cooling section is divided into a slow cooling section. ), Rapid Cooling Section, Over Aging Section (OAS) and Final Cooling Section. The temperature control of the annealing furnace requires precise heating, heat preservation and cooling control of the strip according to the process requirements and the set curve. In this process, it is also necessary to ensure the uniformity of the furnace temperature of the annealing furnace, which directly determines the quality of the product.

The Chinese patent with the patent number of CN201710828269.2 in the prior art discloses a method and system for controlling the temperature of an annealing furnace. Determine whether the actual required heating load of each heating control zone in the M heating control zones is less than the preset load threshold value, and set the actual required heating load of each heating control zone in the M heating control zones less than the preset load threshold value. The corresponding gas flow controllers are all maintained at the first constant flow rate, and control each burner in the same heating control zone to ignite intermittently at the same time.

The Chinese patent with the patent number of CN201310109424.7 in the prior art discloses a temperature control method for a roller hearth annealing furnace, which includes six steps. The first step is to calculate the target of the product heating section outlet corresponding to a given heating curve of the process. temperature; the second step, calculating the billet temperature; the third step, calculating the remaining heating time from the billet to the end of the heating section; the fourth step, performing feedforward control on the heating temperature, taking the billet temperature calculated in step two as the starting point, using step three Calculate the remaining heating time t of the billet, take the process curve as the furnace temperature input, predict the temperature T of the billet reaching the end of the heating section according to the given time step Δt, and then obtain the deviation between the predicted temperature and the target temperature, and then obtain the billet. The feedforward temperature control amount of the section where it is located; the fifth step, use the calculation result of the billet temperature tracking model at the outlet of the heating section and the deviation from the target temperature of the billet at that place to determine the feedback control amount of the heating temperature; sixth step, to The heating section of the roller hearth furnace carries out "feedforward + feedback" temperature setting control.

In the above-mentioned prior art, monitoring the furnace temperature according to the process curve can only provide an over-limit alarm, but cannot provide an early warning and a deterioration trend reminder, and the prior art can no longer meet the production line with high production requirements.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a furnace temperature early warning method and system for a continuous annealing furnace (annealing furnace for hot-dip galvanizing units), which combines traditional alarm methods with SPC process control statistics and neural network prediction algorithm methods to provide continuous hot-dip galvanizing units. The annealing furnace provides a new method of temperature health monitoring and early warning.

To achieve the above object, the present invention adopts the following technical solutions:

A first aspect of the present application provides a continuous annealing furnace temperature warning method, including:

The process parameters of the process points set in each furnace section in the annealing furnace are collected in real time, and the set process points include oxygen content, hydrogen content, dew point of the furnace atmosphere, furnace pressure, target plate temperature, roller tension, One or more, preferably at least two or more, of coil number, furnace temperature, coil specification, unit speed, plate temperature, and exhaust gas;

The collected data is transmitted to the SPC system, and the SPC system groups the collected data according to the category of the process point, and each group independently sets its sample collection time period, and the data of each sample collection time period is based on its The preset sample collection method corresponding to the process point category is processed to form a single sample;

Data quality inspection is carried out on all samples, only the data in the normal operation period of the unit is retained, the data in the fault period is excluded, and a few missing data in the normal operation period are interpolated to form a sample set;

Perform feature extraction on the sample set, and calculate the average and standard deviation of the samples within a preset time;

Statistical judgment is made on the sample set according to the first discrimination rule. If the judgment result is abnormal overrun, an overrun alarm will be triggered; otherwise, the sample set will be statistically judged according to the second discrimination rule. If the judgment result is abnormal fluctuation, it indicates that The annealing furnace has a deterioration trend, and if an unstable situation occurs, a deterioration trend warning is triggered;

Based on the BP neural network algorithm, the furnace temperature change of each furnace section in the annealing furnace is predicted.

Preferably, the preset sample collection method includes one or more of the average value of the sample collection period and the maximum value of the sample collection period.

Preferably, the roller tension at each set process point includes the tension of the uncoiler, the tension of the cleaning section, the tension of the inlet looper, the tension of the preheating section of the annealing furnace, the inlet tension of the heating section of the annealing furnace, the outlet tension of the heating section of the annealing furnace, Tension of soaking section of annealing furnace, tension of slow cooling section of annealing furnace, tension of fast cooling section of annealing furnace, tension of zinc pot section, tension of central looper, entrance tension of skin passer, exit tension of skin passer, tension of tension leveler, post-treatment Several or more of segment tension, exit looper tension, disc shear segment tension, and coiler tension.

Preferably, the steel coil specification includes one or more of the thickness, width, coil number, and steel type of the steel coil.

Preferably, the exhaust gas includes one or more of particulate matter, SO ₂ , NO _X , O ₂ , and exhaust gas temperature, where x is the atomic ratio of O to N in NOx.

Preferably, before the process parameters in the furnace are collected in real time, the early warning method further includes: setting a standard value of each process point of each furnace section, and the standard value is the health degree of the corresponding process point. Reference range for evaluation.

Preferably, the first discrimination rule includes: if the average value of the samples in a preset time exceeds its preset abnormal threshold, it is determined as an abnormal point; if the abnormal point in a preset time period before the current moment is abnormal If the number exceeds the preset value, it will be judged as an over-limit abnormality.

More preferably, the preset abnormal threshold value can be calculated and obtained according to the respective distribution patterns of the equipment operating state signal values in the healthy state. For example, the motor torque signal generally obeys a normal distribution, and the mean value plus three standard deviations is taken as the abnormal threshold value. .

Preferably, the second discrimination rule includes: using a self-defined control limit value, including a first control limit, a second control limit and a third control limit, and the first control limit is the standard deviation of the sample within a preset time. 3 times, the second control limit is 2 times the standard deviation, and the third control limit is 1 time the standard deviation; data trend analysis is carried out according to the following rules, as long as any one of the rules is met, it is determined that the fluctuation is abnormal;

Rule 1: A sample falls outside the preset first control limit;

Rule 2: consecutive K samples fall on the same side of the centerline;

Criterion 3: consecutive K samples are incremented or decremented;

Rule 4: Adjacent samples in consecutive K samples alternate up and down;

Criterion 5: K samples in consecutive K+1 samples fall outside the second control limit on the same side of the centerline;

Criterion 6: K samples in consecutive K+1 samples fall outside the third control limit on the same side of the centerline;

Criterion 7: consecutive K samples fall within the third control limit on the same side of the centerline;

Criterion 8: consecutive K samples fall on both sides of the centerline and none of them are within the third control limit;

Among them, K is a positive integer.

Preferably, based on the BP neural network algorithm, predict the furnace temperature change of each furnace section in the annealing furnace, including:

Select a sample training set from the sample set;

Build a BP neural network model. The BP neural network model includes an input layer, a hidden layer and an output layer. The neurons between adjacent layers are fully connected, and there is no connection between the neurons in each layer. Set the model parameters, including: setting the input Layer input signal selection, set the number of hidden layers and the number of nodes in each layer, set the number of nodes in the output layer, learning rate, transfer function, momentum factor, maximum training times and minimum precision;

Use the existing sample training set to train the BP neural network model, including: proceeding in the forward propagation direction, from the input layer to the hidden layer to the output layer, to obtain the output value of each neuron, until the final output layer is obtained If the output value of the output layer does not match the expected output value, proceed in the direction of reverse propagation. According to the error between the actual output of the output layer and the expected output value, adjust the neuron according to the set learning rate. The connection weights between the two; forward output calculation and reverse weight modification are alternately performed until the error of the network output is less than the preset minimum accuracy, or until the preset maximum training times are performed to determine the current furnace temperature change Predictive models of trends;

According to the BP neural network model that has been trained, the sample set is input into the input layer of the BP neural network model, and the predicted value of the current furnace temperature change trend is obtained through the processing of the neural network.

More preferably, the selection range of the learning rate of the BP neural network model is between 0.01-0.8, the selection range of the momentum factor is between 0-1, and is larger than the learning rate.

Further, the learning rate of the BP neural network model is preferably 0.06, and the momentum factor is preferably 0.95.

More preferably, the transfer function of the BP neural network model is

A second aspect of the present application provides a continuous annealing furnace temperature warning system, including:

The data acquisition module is used to collect the process parameters of the process points set in each furnace section in the annealing furnace in real time, and the set process points include oxygen content, hydrogen content, dew point of the furnace atmosphere, furnace pressure, target plate One or more of temperature, roll tension, coil number, furnace temperature, coil specification, unit speed, plate temperature, exhaust gas; preferably at least two or more;

The sample generation module is used to group the collected data according to the category of the process point, each group independently sets its sample collection time period, and the data of each sample collection time period is preset according to its process point category. The sample collection method is processed to form a single sample;

The data quality inspection module is used to check the data quality of all samples. Only the data in the normal operation period of the unit is retained, the data in the fault period is excluded, and a few missing data in the normal operation period are interpolated to form a sample set;

The feature extraction module is used to perform feature extraction on the sample set, and calculate the average and standard deviation of the samples within a preset time;

The over-limit judgment alarm module is used to perform statistical judgment on the sample set according to the first judgment rule, and if the judgment result is an over-limit abnormality, an over-limit alarm is triggered;

The deterioration trend judgment and early warning module is used to perform statistical judgment on the sample set according to the second discrimination rule when the sample set has no abnormality exceeding the limit. If the judgment result is abnormal fluctuation, it indicates that the annealing furnace has a deterioration trend and an unstable situation occurs , then trigger the deterioration trend warning;

The furnace temperature prediction module based on the BP neural network algorithm is used to predict the furnace temperature change of each furnace section in the annealing furnace based on the BP neural network algorithm.

Preferably, the roller tension at each set process point includes the tension of the uncoiler, the tension of the cleaning section, the tension of the inlet looper, the tension of the preheating section of the annealing furnace, the inlet tension of the heating section of the annealing furnace, the outlet tension of the heating section of the annealing furnace, Tension of soaking section of annealing furnace, tension of slow cooling section of annealing furnace, tension of fast cooling section of annealing furnace, tension of zinc pot section, tension of central looper, entrance tension of skin passer, exit tension of skin passer, tension of tension leveler, post-treatment Several or more of segment tension, exit looper tension, disc shear segment tension, and coiler tension.

Preferably, the steel coil specification includes one or more of the thickness, width, coil number, and steel type of the steel coil.

Preferably, the exhaust gas includes one or more of particulate matter, SO ₂ , NO _X , O ₂ , and exhaust gas temperature, where x is the atomic ratio of O to N in NOx.

Preferably, the furnace temperature prediction module based on BP neural network algorithm includes:

The BP neural network model building module is used to set the selection of the input signal of the input layer, the number of layers of the hidden layer and the number of nodes of each layer, the number of nodes of the output layer, the learning rate, the transfer function, the momentum factor, and the maximum number of training times. and minimum precision;

The BP neural network model training module is used to train the BP neural network model by using the existing sample training set to obtain the prediction model of the current furnace temperature change trend;

The furnace temperature change trend prediction module is used to input the sample set into the input layer of the BP neural network model according to the BP neural network model that has been trained. After processing by the neural network, the predicted value of the current furnace temperature change trend is obtained.

Compared with the prior art, the technical scheme of the present invention has the following beneficial effects:

The present application provides a furnace temperature early warning method and system for a continuous annealing furnace, which combines the traditional alarm method with SPC process control statistics and BP neural network prediction algorithm method to generate a set of temperature prediction based on SPC control theory under steady state. A new way of warning. The technical solution of the present application can make trend prediction of furnace health degree, such as increasing, decreasing, chaotic, regular fluctuation, etc. while achieving over-limit alarm, and at the same time use BP neural network algorithm to predict furnace temperature. The early warning system using this method can assist the operator to give an alarm in advance and find the fault point in time when there is a problem, so as to strive for early detection and early intervention, so as to avoid missing the best processing time for the fault due to finding the fault point.

Description of drawings

The accompanying drawings constituting a part of the present application are used to provide further understanding of the present application, and the schematic embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation of the present application. In the attached image:

Fig. 1 is the schematic flow sheet of the continuous annealing furnace furnace temperature early warning method of the present application;

Fig. 2 is a kind of SPC model design schematic diagram of the present application;

Fig. 3 is the variation curve diagram of the annealing temperature 1 of the coil under the unsteady deterioration of the SPC of the annealing furnace of the embodiment of the present application (target temperature 830°C);

Fig. 4 is the change curve diagram of the annealing temperature 2 of the steel coil under the unsteady deterioration of the SPC of the annealing furnace of the embodiment of the present application (target temperature 830°C);

Fig. 5 is the variation curve of the annealing temperature of the coil under the steady-state health of the SPC of the annealing furnace of the embodiment of the present application (target temperature 810°C);

FIG. 6 is a graph showing the variation of the annealing temperature of the coil under the steady state health of the SPC of the annealing furnace according to the embodiment of the present application (target temperature 835° C.).

Detailed ways

In order to make the objectives, technical solutions and effects of the present invention clearer and clearer, the present invention will be described in further detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.

It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific sequence or sequence. It should be understood that data so used may be interchanged under appropriate circumstances. Furthermore, the terms "comprising" and "having", and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those expressly listed Rather, those steps or units may include other steps or units not expressly listed or inherent to these processes, methods, products or devices.

Referring to Figure 1, a method for early warning of a continuous annealing furnace temperature of the present application includes:

The process parameters of the process points set in each furnace section in the annealing furnace are collected in real time, and the set process points include oxygen content, hydrogen content, dew point of the furnace atmosphere, furnace pressure, target plate temperature, roller tension, One or more of coil number, furnace temperature, coil specification, unit speed, plate temperature, exhaust gas; preferably at least two or more;

The collected data is transmitted to the SPC system, and the SPC system groups the collected data according to the category of the process point, and each group independently sets its sample collection time period, and the data of each sample collection time period is based on its The preset sample collection method corresponding to the process point category is processed to form a single sample;

Data quality inspection is carried out on all samples, only the data in the normal operation period of the unit is retained, the data in the fault period is excluded, and a few missing data in the normal operation period are interpolated to form a sample set;

Perform feature extraction on the sample set, and calculate the average and standard deviation of the samples within a preset time;

Statistical judgment is made on the sample set according to the first discrimination rule. If the judgment result is abnormal overrun, an overrun alarm will be triggered; otherwise, the sample set will be statistically judged according to the second discrimination rule. If the judgment result is abnormal fluctuation, it indicates that The annealing furnace has a deterioration trend, and if an unstable situation occurs, a deterioration trend warning is triggered;

Based on the BP neural network algorithm, the furnace temperature change of each furnace section in the annealing furnace is predicted.

Example:

The nature of the early warning of furnace health is the early warning of important indicators of each furnace section. Therefore, the important parameters and reasonable ranges of each furnace section should be confirmed first. The specific content is shown in Table 1. In Table 1, the design of each furnace is mainly based on process parameters and quality-related factors. Furnace section health evaluation scheme.

Table 1 Health evaluation scheme of each furnace section

The electrical control unit of the cold rolling annealing furnace has two parts, namely DCS and PLC. Confirm the data address of the required process points and automatically collect this part of the process points. See Table 2, this part of the process points can include oxygen content, hydrogen content, dew point of furnace atmosphere, furnace pressure, target plate temperature, roll tension, coil number, furnace temperature, coil size, unit speed, plate Gentle exhaust.

Table 2 Important process points and their corresponding addresses

The SPC system is deployed on site and the collected data is transmitted to the SPC system. The main job of the SPC system is to adjust the SPC sample interval. If the sample interval is too short, it is easy to cause frequent false alarms. If the sample interval is too large, the warning will not work. advance. The SPC system of the present application groups the collected data according to the category of the process point, each group independently sets its sample collection time period, and the data of each sample collection time period is based on the prediction corresponding to the category of the process point. The designed sample collection method is processed to form a single sample. See Table 3.

Table 3 Schematic diagram of SPC collection interval

The purpose of data quality inspection is to detect whether the data quality of the input data meets the requirements of the algorithm. The inspection of data quality includes two aspects:

(1) Check whether there are missing values or outliers in the sample data. The detection of outliers usually adopts a rule-based method, such as whether the signal is within a certain value range, the judgment rules of PLC signals, etc.;

(2) Check whether the working conditions of the sample data meet the algorithm requirements, that is, whether it is the data during the normal operation of the unit, and it is necessary to use the flag bits in the PLC data to judge.

Next, feature extraction is performed on the sample set, and the average value and standard deviation of the samples within a preset time period are calculated, wherein the calculated average value is used for over-limit abnormality judgment, and the calculated standard deviation is used for fluctuation abnormality judgment.

Over-limit judgment alarm:

The SPC system will automatically calculate and set the abnormal threshold. The abnormal threshold can be calculated and obtained according to the respective distribution patterns of the equipment operating state signal values in the healthy state. For example, the motor torque signal generally obeys a normal distribution, and its mean plus three standard deviations is taken as the abnormal threshold. If the average value of the samples in the preset time exceeds its preset abnormal threshold, it is judged as an abnormal point; in order to reduce the false alarm rate, the number of abnormal points in a preset time period before the current moment can exceed the preset number of abnormal points. value, it is judged to be over-limit abnormality, and an over-limit alarm is triggered.

Deterioration trend judgment warning:

When the sample set is confirmed to be within the controllable range through the over-limit judgment, the deterioration trend will be judged, and the judgment will be based on the SPC rule.

SPC (Statistical Process Control) is a widely used process control method and a process control tool with the help of mathematical statistics. It analyzes and evaluates the production process, finds the signs of systematic factors in time according to the feedback information, and takes measures to eliminate its influence, so as to maintain the process in a controlled state only affected by random factors, so as to achieve the purpose of quality control.

The SPC rules (discrimination criteria) are as follows:

Criterion 1: 1 idea falls outside the A zone (the idea is out of control limits)

Rule 2: 9 consecutive points on the same side of the centerline

Rule 3: 6 points in a row increment or decrement

Rule 4: Adjacent points in a row of 14 points always alternate up and down

Criterion 5: 2 of 3 consecutive points fall outside Zone B on the same side of the centerline

Rule 6: 4 out of 5 consecutive points fall outside the C area on the same side of the centerline

Rule 7: 15 consecutive points fall within the C area on the same sides of the centerline

Rule 8: 8 consecutive points on either side of the centerline and no 1 in zone C

Among them, the C area is the preset third control limit, and its size is 1 times the standard deviation of the samples within the preset time; the B area is the preset second control limit, and its size is the standard deviation of the samples within the preset time. 2 times; the A area is the preset third control limit, and its size is 3 times the standard deviation of the samples within the preset time.

Referring to Figure 2, data trend analysis is performed according to the above rules. When the sample set conforms to any one of the above rules, it indicates that the health of the annealing furnace is deteriorating, and the non-steady state occurs, it is judged as abnormal fluctuation, and a deterioration trend warning is triggered.

The above-mentioned early warning method of the present application improves the traditional SPC, mainly in the following two points:

① The function of the SPC control chart is to reflect the stability of the production process, but the stability does not fully reflect the quality of the data, and stable output of wrong data is also a possibility. Therefore, the technical solution of the present application adds a function of setting upper and lower limit conditions, in order to ensure Re-analyze the trend of the data while the data is within the controllable range.

②The number of samples can be adjusted. Because the rate of change of the on-site data is different, it is necessary to adjust the number of samples collected for a single sample of the corresponding data according to the process conditions, so as to better reflect the on-site trend. For example, the temperature difference of the furnace temperature plate is averaged once every 30 minutes to make a sample, and the sample is analyzed by SPC XBar chart. If the continuous 6 points rise, it means that the temperature difference of the furnace temperature plate has an upward trend within 3 hours. Only by setting a reasonable sample can the real effect of SPC be exerted.

The present application can also predict the furnace temperature change of each furnace section in the annealing furnace based on the BP neural network algorithm.

BP neural network is a neural network learning algorithm. It is a hierarchical neural network composed of an input layer, a hidden layer, and an output layer, and the hidden layer can be extended to multiple layers. Each neuron between adjacent layers is fully connected, but there is no connection between each neuron in each layer, and the network learns in a teacher-supervised manner.

学习速率0.06，附加动量因子0.95，具体步骤如下：The learning process consists of two parts: the forward propagation of the signal and the reverse return of the error; in the forward propagation, the input samples are passed in from the input layer, processed layer by layer by each hidden layer, and transmitted to the output layer. If the output layer outputs If it is not in line with expectations, the error is used as an adjustment signal to return back layer by layer, and the connection weight matrix between neurons is processed to reduce the error. After repeated learning, the error is finally reduced to an acceptable range, and the final transfer function is selected as

The learning rate is 0.06, and the additional momentum factor is 0.95. The specific steps are as follows:

① Take a sample from the training sample set and input the information into the network.

②The actual output of the neural network is obtained after forward processing layer by layer through the connection between the nodes.

③ Calculate the error between the actual output of the network and the expected output.

④Reverse the error back to the previous layers layer by layer, and load the error signal onto the connection weight according to a certain principle, so that the connection weight of the entire neural network is transformed in the direction of reducing the error.

⑤ Repeat the above steps for each input-output sample pair in the training sample set, until the error of the entire training sample set is less than the preset accuracy, or until the preset number of times.

According to the BP neural network model that has been trained, the sample set is input into the input layer of the BP neural network model, and the predicted value of the current furnace temperature change trend is obtained through the processing of the neural network.

Table 4 is a prediction result table of using the BP neural network model obtained by the above training to predict the furnace temperature under unsteady deterioration and steady state health of the annealing furnace, respectively.

Table 4 Model algorithm result table

It should be noted that the data in Table 4 are the individual values of the samples. Due to the large number of samples, the points in Figures 3 to 6 are the mean values of the samples at a certain time.

Fig. 3 is a graph showing the change of the annealing temperature 1 of the steel coil under the unsteady deterioration of SPC in the annealing furnace, and Fig. 4 is a graph showing the change of the annealing temperature 2 of the steel coil under the unsteady deterioration of SPC in the annealing furnace, wherein, Fig. 3 The target temperature in Figure 4 is 830°C, and the target temperature in Figure 4 is 830°C. It can be seen from Table 4 combined with Figure 3 that the temperature of the rear panel is low. It can be seen from Table 4 combined with Figure 4 that the temperature of the rear panel is lower than 10°C.

Fig. 5 is the change curve of the annealing temperature of the coil under the steady state of the SPC in the annealing furnace, and Fig. 6 is the change curve of the annealing temperature of the steel coil under the steady state of the SPC in the annealing furnace. Among them, the target temperature in Fig. 5 was 810°C, and the target temperature in Figure 6 was 835°C. It can be seen from Table 4 and Figure 5 that the fluctuation of the rear panel is large, and the temperature of the rear panel is better. It can be seen from Table 4 and Figure 6 that the fluctuation of the rear plate is large, and the temperature of the rear plate is well controlled.

It can be seen from the above legend and data that the predicted trend is roughly the same as the actual plate temperature change trend, and the predicted value fits well with the measured value.

Based on the same inventive concept of the above embodiments, in another preferred embodiment, the present application also provides a furnace temperature warning system for a continuous annealing furnace, including:

The data acquisition module is used to collect the process parameters of the process points set in each furnace section in the annealing furnace in real time, and the set process points include oxygen content, hydrogen content, dew point of the furnace atmosphere, furnace pressure, target plate One or more of temperature, roll tension, coil number, furnace temperature, coil specification, unit speed, plate temperature, exhaust gas; preferably at least two or more;

The sample generation module is used to group the collected data according to the category of the process point, each group independently sets its sample collection time period, and the data of each sample collection time period is preset according to its process point category. The sample collection method is processed to form a single sample;

The data quality inspection module is used to check the data quality of all samples. Only the data in the normal operation period of the unit is retained, the data in the fault period is excluded, and a few missing data in the normal operation period are interpolated to form a sample set;

The feature extraction module is used to perform feature extraction on the sample set, and calculate the average and standard deviation of the samples within a preset time;

The over-limit judgment alarm module is used to perform statistical judgment on the sample set according to the first judgment rule, and if the judgment result is an over-limit abnormality, an over-limit alarm is triggered;

The deterioration trend judgment and early warning module is used to perform statistical judgment on the sample set according to the second discrimination rule when the sample set has no abnormality exceeding the limit. If the judgment result is abnormal fluctuation, it indicates that the annealing furnace has a deterioration trend and an unstable situation occurs , then trigger the deterioration trend warning;

The furnace temperature prediction module based on the BP neural network algorithm is used to predict the furnace temperature change of each furnace section in the annealing furnace based on the BP neural network algorithm.

In summary, the present application provides a continuous annealing furnace temperature warning method and system, which combines the traditional alarm method with the SPC process control statistics and BP neural network prediction algorithm method to generate a set of steady state based on SPC control theory. A new early warning method for temperature forecasting. The technical solution of the present application can make trend prediction of furnace health, such as increasing, decreasing, chaotic, regular fluctuation, etc., while achieving over-limit alarm, and at the same time, the BP neural network algorithm is used to predict furnace temperature. The early warning system using this method can assist the operator to give an alarm in advance and find the fault point in time when there is a problem, strive for early detection and early intervention, and avoid missing the best processing time for the fault due to finding the fault point. This application is applied to the design of the A08 continuous annealing furnace for Baosteel Baori Auto Sheet Cold Rolling and Hot-dip Galvanizing Unit. The trend of key process parameters in different furnace sections is analyzed through the SPC judgment principle, and the possible health hazards in the annealing furnace are judged, and predicted according to the BP neural network algorithm. The change of furnace temperature helps managers make decisions, avoids major failures, reduces the number of downtimes, and improves production efficiency.

The specific embodiments of the present invention have been described above in detail, but they are only used as examples, and the present invention is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications and substitutions to the present invention are also within the scope of the present invention. Therefore, equivalent changes and modifications made without departing from the spirit and scope of the present invention should be included within the scope of the present invention.

Claims (8)

1. The method for early warning the furnace temperature of the continuous annealing furnace is characterized by comprising the following steps of:

acquiring process parameters of process point locations set by each furnace section in the annealing furnace in real time, wherein the set process point locations comprise one or more of oxygen content, hydrogen content, dew point of atmosphere in the furnace, furnace pressure, target plate temperature, roller shaft tension, steel coil number, furnace temperature, steel coil specification, unit speed, plate temperature and waste gas;

the collected data are transmitted to an SPC system, the SPC system groups the collected data according to the category of the process point location, each group independently sets the sample collection time period, and the data in each sample collection time period are processed according to the preset sample collection method corresponding to the category of the process point location to form a single sample;

performing data quality inspection on all samples, only retaining data in a normal operation period of a unit, eliminating data in a fault period, and interpolating a few missing data in the normal operation period to form a sample set;

extracting characteristics of the sample set, and calculating the average value and the standard deviation of the samples in preset time;

carrying out statistical judgment on the sample set according to a first judgment rule, and triggering an overrun alarm if the judgment result is overrun abnormity; otherwise, carrying out statistical judgment on the sample set according to a second judgment rule, and if the judgment result is abnormal fluctuation, indicating that the annealing furnace has a degradation trend and an unstable condition appears, triggering degradation trend early warning;

predicting the furnace temperature change condition of each furnace section in the annealing furnace based on a BP neural network algorithm;

wherein the first discriminant rule includes: if the average value of the samples in the preset time exceeds a preset abnormal threshold value, judging the samples to be abnormal points; if the number of abnormal points in a preset time period before the current moment exceeds a preset value, judging that the overrun is abnormal;

the second rule includes: the method comprises the steps that self-defined control limit values are adopted and comprise a first control limit, a second control limit and a third control limit, the first control limit is 3 times of a standard deviation of a sample in preset time, the second control limit is 2 times of the standard deviation, and the third control limit is 1 time of the standard deviation; performing data trend analysis according to the following rules, and judging that the fluctuation is abnormal as long as any rule is met;

criterion 1: one sample falls outside a preset first control limit;

criterion 2: the continuous K samples fall on the same side of the center line;

criterion 3: consecutive K samples increment or decrement;

criterion 4: adjacent samples in the continuous K samples alternate up and down;

criterion 5: k samples of the K +1 consecutive samples fall outside a second control limit on the same side of the centerline;

criterion 6: k samples of the K +1 consecutive samples fall outside a third control limit on the same side of the centerline;

criterion 7: the continuous K samples fall within third control limits on the same two sides of the center line;

criterion 8: the continuous K samples fall on two sides of the central line, and none of the continuous K samples is within a third control limit;

wherein K is a positive integer.

2. The method for early warning the furnace temperature of the continuous annealing furnace according to claim 1, wherein the roll shaft tension of each set process point comprises several or more of uncoiler tension, cleaning section tension, inlet loop tension, annealing furnace preheating section tension, annealing furnace heating section inlet tension, annealing furnace heating section outlet tension, annealing furnace soaking section tension, annealing furnace slow cooling section tension, annealing furnace fast cooling section tension, zinc pot section tension, central loop tension, leveler inlet tension, leveler outlet tension, tension of a tension leveler, tension of a post-processing section, outlet loop tension, tension of a circle shear section and tension of a coiler.

3. The method for warning the furnace temperature of the continuous annealing furnace according to claim 1, wherein the steel coil specification comprises one or more of the thickness, the width, the coil number and the steel type of the steel coil.

4. The method for warning the temperature of the continuous annealing furnace according to claim 1, wherein the exhaust gas comprises particulate matters and SO₂、NO_X、O₂And one or more of the temperature of the waste gas, wherein x is the atomic number ratio of O to N in the NOx.

5. The method for warning the furnace temperature of the continuous annealing furnace according to claim 1, wherein before the real-time collection of the process parameters in the furnace fire, the method for warning the furnace temperature further comprises: and setting a standard value of each process point location of each furnace section, wherein the standard value is a reference range of health degree evaluation of the corresponding process point location.

6. The method for warning the furnace temperature of the continuous annealing furnace according to claim 1, wherein the step of predicting the furnace temperature change of each furnace section in the annealing furnace based on a BP neural network algorithm comprises the following steps:

selecting a sample training set from the sample set;

constructing a BP neural network model, wherein the BP neural network model comprises an input layer, a hidden layer and an output layer, all neurons in adjacent layers are in full connection, and all neurons in each layer are not in connection; setting model parameters, including: setting selection of input signals of an input layer, setting the number of layers of a hidden layer and the number of nodes of each layer, and setting the number of nodes, learning rate, transfer function, momentum factor, maximum training times and minimum precision of an output layer;

training a BP neural network model by using the existing sample training set, comprising the following steps: proceeding according to a forward propagation direction, obtaining the output value of each neuron from the direction from the input layer to the hidden layer to the output layer until obtaining the output value of the final output layer, if the output value of the output layer is not consistent with the expected output value, proceeding according to the backward propagation direction, and adjusting the connection weight between the neurons according to the set learning rate and the error between the actual output and the expected output value of the output layer; alternately performing forward output calculation and reverse weight modification until the error of network output is smaller than the preset minimum precision or the preset maximum training times to determine a current furnace temperature change trend prediction model;

and inputting the sample set into an input layer of the BP neural network model according to the trained BP neural network model, and obtaining a predicted value of the current furnace temperature change trend through the processing of the neural network.

7. The furnace temperature early warning method of the continuous annealing furnace according to claim 6, wherein the learning rate of the BP neural network model is selected from a range of 0.01-0.8; the selection range of the momentum factor is between 0 and 1 and is larger than the learning rate; the transfer function is

8. The utility model provides a continuous annealing stove furnace temperature early warning system which characterized in that includes:

the data acquisition module is used for acquiring process parameters of process point locations set by each furnace section in the annealing furnace in real time, wherein the set process point locations comprise one or more of oxygen content, hydrogen content, dew point of atmosphere in the furnace, furnace pressure, target plate temperature, roller shaft tension, steel coil number, furnace temperature, steel coil specification, unit speed, plate temperature and waste gas;

the sample generation module is used for grouping the acquired data according to the type of the process point location, independently setting the sample acquisition time period of each group, and processing the data of each sample acquisition time period according to a preset sample acquisition method corresponding to the type of the process point location to form a single sample;

the data quality inspection module is used for performing data quality inspection on all samples, only retaining data in a normal operation time period of the unit, eliminating data in a fault time period, and interpolating a few missing data in the normal operation time period to form a sample set;

the characteristic extraction module is used for extracting characteristics of the sample set and calculating the average value and the standard deviation of the samples in preset time;

the overrun judgment alarm module is used for carrying out statistical judgment on the sample set according to a first judgment rule, and triggering overrun alarm if the judgment result is overrun abnormity;

the degradation trend judgment and early warning module is used for carrying out statistical judgment on the sample set according to a second judgment rule under the condition that the sample set is not in overrun abnormality, and triggering degradation trend early warning if the judgment result is in fluctuation abnormality and indicates that the annealing furnace has a degradation trend and an unstable condition appears;

and the furnace temperature prediction module based on the BP neural network algorithm is used for predicting the furnace temperature change condition of each furnace section in the annealing furnace based on the BP neural network algorithm.

Images