CN105108097A - Hybrid model and hybrid-model-based continuous casting breakout prediction method - Google Patents
Hybrid model and hybrid-model-based continuous casting breakout prediction method Download PDFInfo
- Publication number
- CN105108097A CN105108097A CN201510447779.6A CN201510447779A CN105108097A CN 105108097 A CN105108097 A CN 105108097A CN 201510447779 A CN201510447779 A CN 201510447779A CN 105108097 A CN105108097 A CN 105108097A
- Authority
- CN
- China
- Prior art keywords
- model
- thermocouple
- temperature
- abnormal
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B22—CASTING; POWDER METALLURGY
- B22D—CASTING OF METALS; CASTING OF OTHER SUBSTANCES BY THE SAME PROCESSES OR DEVICES
- B22D11/00—Continuous casting of metals, i.e. casting in indefinite lengths
- B22D11/16—Controlling or regulating processes or operations
- B22D11/18—Controlling or regulating processes or operations for pouring
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Mechanical Engineering (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Continuous Casting (AREA)
Abstract
The invention discloses a hybrid model and a hybrid-model-based continuous casting breakout prediction method and belongs to the technical field of monitoring of metallurgical continuous casting. The hybrid model comprises a single couple sequence model based on a GA (Genetic Algorithm)-BP (Back Propagation) neural network and a group couple space model based on logical judgment. The prediction method comprises the following steps: (1) monitoring the temperature of a crystallizer and saving data; (2) inputting the data to the single couple sequence model, judging whether the time-varying temperature of each thermocouple conforms to the temperature variation waveform during sticking, and saving a judgment result in an array Y (i, j, t); (3) when the array Y (i, j, t) is within the range of [theta min, theta max], marking that the thermocouple is abnormal, and calculating the number of abnormal thermocouples in the ith row and the (i-1)th row to be respectively m and n; and (4) comparing m+n with the threshold values of the numbers of sticking-alarming and sticking-warning thermocouples for sticking judgment. The hybrid model and the hybrid-model-based continuous casting bleed-out prediction method achieve the goal of improving the sticking breakout recognition precision.
Description
Technical field
The invention belongs to monitoring technique field in metallurgical continuous casting, more particularly, relate to a kind of mixed model and the continuous casting breakout prediction method based on mixed model.
Background technology
Continuous casting bleed-out is the pernicious quality accident that the cc billet surface quality development of defects such as bonding or crackle produces to a certain extent, conticaster can be caused to stop production, affect the production schedule of casting process continuity and whole steel-making, and damage equipment, affect operating rate and the output of casting machine, decrease the recovery rate of metal, cause huge economic loss.In actual production process, sticker type breakout occurrence frequency is the highest, accounts for 70% ~ 80% of all kinds of bleed-out event.Especially with the development of the efficient slab continuous casting technology of modernization; not only to cast crack sensitivity steel grade; casting steel grade wide ranges; casting difficulty is large; and the raising of pulling rate causes the problems such as more complicated Billet, friction and lubrication; initial solidification shell is cooled and solidify, stability that covering slag flows into and uniformity significantly decline, in crystallizer, strand bonding phenomenon increases, and then the sticker type breakout problem caused is very outstanding.Sticker type breakout is main bleed-out form, studies and solves sticker type breakout to ensureing that continuous casting production direct motion is significant with raising slab quality.
From the seventies in last century, develop the detection method of a variety of sticker type breakout sign both at home and abroad, most effective method is thermocouple thermometry, general principle is by burying some thermocouples on copper plate of crystallizer underground, detect copper plate of crystallizer different parts temperature variations, utilize copper plate temperature situation of change monitor the localized heat transfer situation of crystallizer inside in real time and identify strand rupture location and mobile message thereof.At present, breakout prediction method based on thermocouple temperature measurement mainly contains two classes, one class is by Logic judgment model, extract suitable logical condition after carrying out qualitative and quantitative analysis according to bleed-out Analysis on Mechanism and bleed-out data and carry out breakout prediction, its principle is according to each electric thermo-couple temperature amplitude of variation, rate temperature change, arranges the threshold values multilevel iudge of the parameters such as thermoelectric, variations in temperature time delay and setting up and down, makes the warning of bleed-out degree.Logic judgment model depends on concrete technology and equipment parameter etc., and the selection of model parameter needs a large amount of manpowers and time test, adaptivity and poor robustness, the rate of false alarm that frequent appearance is higher, report by mistake frequently and can affect the quality of strand and the high-efficiency reform of casting machine equally, and reduce wrong report and can increase and fail to report.Another kind of is by intellectual technology (such as neutral net, SVMs isotype recognizer) sticker type breakout is forecast, be characterized in having very strong adaptivity, self-learning capability, fault-tolerance and robustness, complicated nonlinear problem can better be processed, can further improve the accuracy of breakout prediction, become current study hotspot.Model of mind belongs to black-box model, and its deficiency is undue dependence data, as neural network model training must rely on enough effective samples, if the incomplete or inaccurate generalization ability that all can affect network of sample data.Go into operation the initial stage at conticaster, owing to lacking enough effective data, Logic judgment model must be relied on forecast and avoid bleed-out.
About the method adopting neutral net to forecast sticker type breakout, in prior art, existing related art scheme is open, as patent publication No.: CN101850410A, publication date: on October 6th, 2010, invention and created name is: a kind of continuous casting breakout prediction method based on neutral net, this application case discloses a kind of continuous casting breakout prediction method based on neutral net, and the method comprises, step 1: the temperature data of the on-the-spot thermocouple of online acquisition continuous casting also stores this temperature data; Step 2: pretreatment is carried out to described temperature data; Step 3: single even sequential network breakout prediction model will be input to through the described pretreated temperature data gathered from any one thermocouple, and the output valve of single even sequential network breakout prediction model and maximum discrimination threshold are compared, if the output valve of this list even sequential network breakout prediction model is greater than maximum discrimination threshold, then forecast that bleed-out can occur; Meanwhile, genetic algorithm is used to initialize the connection weights and threshold of this list even sequential network breakout prediction model.The method can improve the recognition effect and the forecast precision that continuous casting are cohered to bleed-out process, thus decreases rate of false alarm and rate of failing to report.But, the weak point of this application case is: from single even temporal model to the structure of the even spatial model of group, rely on nerual network technique completely, although the intellectual technologies such as neutral net have obvious advantage in dynamic waveform pattern-recognition, utilize the even spatial model of its foundation group and improper, maximum discrimination threshold is greater than by the output valve of single even sequential network breakout prediction model in the method step 3, just forecast bleed-out can occur, and it is unreasonable, because often there is larger temperature fluctuation in the single electric thermo-couple temperature of actual production process, be easy to sticking temperature pattern close, and cause false alarm, the input of the group related in the method even spatial network breakout prediction model is respectively simultaneously from meeting the thermocouple and a lower row left side corresponding with it that judge further, in, the output valve obtained after the temperature data of right three thermocouples is input to single even temporal model, it can thus be appreciated that electric thermo-couple temperature Spatial Variation when boning the propagation of V-type tearing port do not considered by the even spatial model of this group, and only select four described thermocouples to judge the two-dimentional dissemination of bonding, in actual production process when the situation such as occur multiple thermocouple fault and temperature fluctuation is larger, the even spatial model of this group is easy to generation and fails to report and false alarm, during its practical application, robustness will be deteriorated.
In sum, how overcoming the existing weak point forecast sticker type breakout by Logic judgment model or neural network model, is the technical barrier needing solution in prior art badly.
Summary of the invention
1. invent the technical problem that will solve
Instant invention overcomes existing weak point sticker type breakout forecast by Logic judgment model or neural network model, provide a kind of mixed model and the continuous casting breakout prediction method based on mixed model, achieve the target improving sticker type breakout accuracy of identification.
2. technical scheme
For achieving the above object, technical scheme provided by the invention is:
Mixed model of the present invention, mainly comprises following two parts: single even temporal model and the even spatial model of group;
(1), single even temporal model;
The structure of single even temporal model comprises: the determination of mode input variable, the determination of model output variable, the pretreatment of data, the foundation of GA-BP neutral net;
(2), even spatial model is organized;
1) after, utilizing single even temporal model changing pattern identifying in time to electric thermo-couple temperatures all on crystallizer, Output rusults is saved in three-dimensional array Y (i, j, t), wherein, Y (i, j, t) represents that the i-th row j row thermocouple is at the even temporal model recognition result of the list of t;
2), when Y (i, j, t) is at threshold range [θ
min, θ
max] interior time, think that this thermocouple TC (i, j) variations in temperature meets sticking temperature pattern, mark this thermocouple abnormal;
3), then check the Y (i, j, t) of all thermocouples of the i-th row, add up at threshold range [θ
min, θ
max] in abnormal thermocouple number be m, check the Y (i-1, j, t) of all thermocouples of the i-th-1 row simultaneously, add up at threshold range [θ
min, θ
max] in abnormal thermocouple number be n, wherein i is greater than 1;
4) if m and n is all more than or equal to 2, then check whether 10 seconds menisci capable (the first row) abnormal thermocouple numbers are more than or equal to 2 in the past, if met, then utilizing abnormal thermocouple sum m+n to warn thermocouple number threshold values to compare with strike-alarm with boning respectively, carrying out the judgement of strike-alarm and bonding warning.
As further improvement of the present invention, the determination of mode input variable is as follows;
30 the temperature sampling points selecting single thermocouple continuous in time are as mode input variable;
Wherein, the collection period of temperature sampling point is 1 second.
As further improvement of the present invention, the determination of model output variable is as follows;
Model output variable is single thermocouple strike-alarm signal, and model output variable is determined by the degree of closeness of variations in temperature waveform pattern corresponding to above-mentioned 30 temperature sampling points and sticking temperature pattern;
Wherein, model output variable is the number between-1 ~ 2, and when the variations in temperature waveform pattern that 30 temperature sampling points are corresponding is identical with sticking temperature pattern, model output variable is labeled as 1; When the temperature variation curve held stationary that 30 temperature sampling points are corresponding, model output variable is labeled as 0.
As further improvement of the present invention, preparation and the pretreatment of data are as follows;
From historical data, extract sample, adopt formula (1) to normalize between [-1,1] by above-mentioned effective sample;
In formula (1), x' is the sample data after normalization, and x is the sample data before normalization, x
maxthe maximum of sample data before normalization, x
minit is the minimum of a value of sample data before normalization;
Sample data after above normalization is divided into two parts, and wherein a part is as training sample, and another part is as test sample book.
As further improvement of the present invention, the foundation of GA-BP neural network model is as follows;
BP neural network input layer nodes is 30, representative model input variable; Output layer nodes is 1, representative model output variable; Determine a discrimination threshold scope according to the result of network training, just think when model output variable does not exceed the threshold range preset and sticking temperature pattern detected; Training network adopts 3 layers of BP network, and hidden layer excitation function uses S type tangent transfer function, and output layer uses linear transfer function, and training process adopts LM optimized algorithm; Selected the number of hidden nodes is 12, obtains the BP neural network model that structure is 30 × 12 × 1;
Wherein: BP neural network learning process comprises information forward-propagating and error back propagation, according to given training sample input and output vector unceasing study and the connection weights and threshold adjusted between neuron, network is made constantly to approach mapping relations between sample input and output; The maximum frequency of training of BP neutral net is set to 2000, and learning rate is 0.05, and performance error is 0.0001;
Adopt genetic algorithm optimization BP neural network, 1), initialization of population set up GA-BP neural network model, the basic step of genetic algorithm optimization BP neural network comprises:, carries out real coding to all weights and threshold of BP neutral net, produce individual, population scale and evolutionary generation are set; 2), calculating the fitness function of each individuality according to formula (2), take minimum of a value as optimum;
In formula (2), n is network output layer nodes; y
ibe the actual output of i-th node; Y'
iit is the desired output of i-th node; 3), carry out genetic algorithm selection, crossover and mutation operation, produce individual population of new generation; 4), evaluate population of new generation, judge whether evolutionary generation reaches requirement or whether network error satisfies condition, obtain the corresponding individuality of current population adaptive optimal control angle value if met; Wherein, Population in Genetic Algorithms scale is set to 50, and crossover probability is 0.7, and mutation probability is 0.06, and evolutionary generation is 200.
As further improvement of the present invention, described threshold range [θ
min, θ
max] be [0.6,1.3].
As further improvement of the present invention, strike-alarm thermocouple number threshold values is 6, and bonding warning thermocouple number threshold values is 3; When abnormal thermocouple sum m+n is more than or equal to 3, send bonding warning; When abnormal thermocouple sum m+n is more than or equal to 6, send strike-alarm.
Continuous casting breakout prediction method based on mixed model of the present invention, comprises the steps:
Step (1), in copper plate of crystallizer, arrange many row's high density thermocouples, monitoring mould temperature situation of change, and gather and store on-the-spot all electric thermo-couple temperature real time datas, be saved in three-dimensional array T (i, j, t); Wherein, T (i, j, t) represents the temperature value of the i-th row j row thermocouple in t;
Step (2), by single for the input of all thermo-electric couple temperature datas even temporal model, in single even temporal model, the time series data of each electric thermo-couple temperature, after the conversion and data processing of shift register, input GA-BP neural network model calculates, and judges variations in temperature waveform when whether the change of each electric thermo-couple temperature in time series meets bonding, judged result is saved in three-dimensional array Y (i, j, t) in;
Step (3) if Y (i, j, t) setting threshold range [θ
min, θ
max] interior time, then think that current thermocouple TC (i, j) variations in temperature meets sticking temperature pattern, mark this thermocouple abnormal, so then carry out the judgement organizing even spatial model, calculate current thermocouple and be expert at and lastrow exception thermocouple number;
Step (4), the abnormal thermocouple sum that exported by the even spatial model of group, respectively with strike-alarm with bond and warn thermocouple number threshold values to compare, carry out strike-alarm and the judgement of warning that bonds.
3. beneficial effect
Adopt technical scheme provided by the invention, compared with prior art, there is following remarkable result:
(1), the present invention considers the advantage of artificial intelligence technology on wave pattern cognition, when boning with strand, in copper plate of crystallizer, the spatial and temporal variation of electric thermo-couple temperature is foundation, adopt the even temporal model of GA-BP neural network list, be used for identifying variations in temperature waveform when whether the change of single electric thermo-couple temperature in time series meets bonding, belong to dynamic waveform pattern recognition problem.Wherein, genetic algorithm optimization BP neural network, determines BP neutral net best initial weights and threshold value by ability of searching optimum, improves single even temporal model to the accuracy of identification of sticking temperature waveform pattern.And consider the two-dimentional dissemination of bonding V-type tearing port, the basis of single even temporal model establishes the even spatial model of group by effective logic rules, judge whether neighboring thermocouple has the propagation of bonding, improve the accuracy of identification of sticker type breakout, especially can reduce failing to report and false alarm when multiple thermocouple fault or larger electric thermo-couple temperature fluctuate in actual production process.
(2), by mixed model that the present invention proposes, make full use of the advantage of GA-BP neutral net in wave pattern cognition, and the effective logic rules that are coupled judge, the space-time not only achieving single even summation group idol judges, and overcome the deficiency that simple Logic judgment model parameter determines the guidance of difficult or inaccurate and simple model of mind shortage technique, reach good breakout prediction performance, promptly and accurately can quote whole bonding, avoid sticker type breakout accident, and false alarm frequency is down to floor level.
Accompanying drawing explanation
Fig. 1 is that in embodiment 1, schematic diagram arranged by copper plate of crystallizer thermocouple, mm;
Fig. 2 is the flow chart based on the continuous casting breakout prediction method of mixed model in embodiment 1;
Fig. 3 is sticking temperature pattern in embodiment 1;
Fig. 4 is BP neural network topology structure and learning process schematic diagram in embodiment 1;
Fig. 5 is BP neutral net and GA-BP neutral net test result figure in embodiment 1.
Detailed description of the invention
The present invention proposes a kind of mixed model and the continuous casting breakout prediction method based on mixed model, be intended to solve this technical barrier of sticker type breakout in sheet billet continuous casting process.The spatial and temporal variation of electric thermo-couple temperature in copper plate of crystallizer when the present invention mainly bonds based on strand, first genetic algorithm optimization BP neural network (namely setting up GA-BP neural network model) is adopted, set up single even temporal model, single electric thermo-couple temperature time dependent dynamic waveform during identification bonding, then the even spatial model of logic rules foundation group is adopted, differentiate whether vertical and horizontal neighboring thermocouple has sticking temperature waveform, identify the two-dimentional dissemination of bonding, form GA-BP neutral net and Logic judgment mixed model thus.Utilize genetic algorithm to determine BP neutral net best weight value and threshold value in the present invention, improve the accuracy of identification of single even temporal model.The continuous casting breakout prediction method based on mixed model that the present invention proposes, can quote strand timely and accurately and all bond, avoid sticker type breakout accident, and reduces the probability of bonding wrong report.
For understanding content of the present invention further, the present invention is described in detail in conjunction with the accompanying drawings and embodiments.
Embodiment 1
Conticaster in the present embodiment adopts High Efficiency Slab Caster, and two machines two flow, and slab section is 230 × (900 ~ 2150) mm
2, pulling rate is 0.80 ~ 2.03m/min, and adopt Combined vertical crystallizer, length of mould is 900mm, and crystallizer width and thickness are according to slab cross section regulation.As shown in Figure 1, many row's high density thermocouples are buried underground in copper plate of crystallizer, fixation side (outer arc) and wide of active side (inner arc) are respectively installed 6 row 12 and are arranged totally 72 thermocouples, left side and right side leptoprosopy are respectively installed 6 row 2 and are arranged totally 12 thermocouples, altogether install 168 thermocouples in copper plate of crystallizer.This conticaster is due to casting steel grade wide ranges, and crack sensitivity steel grade of often casting, casting difficulty is large, pulling rate is higher simultaneously, therefore sticker type breakout problem is very outstanding, single current breakout ratio reaches 0.0392%, sticker type breakout accounts for about 72% of whole bleed-out, it is main bleed-out form, so reducing sticker type breakout is the key reducing breakout ratio, and sticker type breakout is forecast timely and accurately to be the reliable guarantee reducing sticker type breakout, adopt the continuous casting breakout prediction method based on mixed model of the present embodiment to reach above-mentioned target.
As shown in Figure 2, the mixed model of the present embodiment mainly comprises following two parts: the even spatial model of the group that the even temporal model of the list based on GA-BP neutral net and logic-based judge.
(1), single even temporal model;
The structure of single even temporal model comprises: the foundation of the determination of mode input variable, the determination of model output variable, the preparation of data and pretreatment, GA-BP neutral net.The selection of mode input variable and model output variable is most important, directly affects predicting the outcome of the even temporal model of this list.According to history bonding or sticker type breakout actual measurement sample, the temperature anomaly change in time of the single thermocouple of bonding process was at about 30 seconds, the temperature acquisition cycle is 1 second, therefore select single thermocouple continuous in time 30 temperature sampling points as mode input variable (namely selecting the data of single thermocouple on 30 continuous print temperature acquisition cycles as mode input variable), can the typical waveform pattern of Complete Characterization bonding process single electric thermo-couple temperature change, the typical waveform pattern of the single electric thermo-couple temperature change of bonding process is referred to as sticking temperature pattern, specifically as shown in Figure 3.In the present embodiment, model output variable is single thermocouple strike-alarm signal, and model output variable is determined by the degree of closeness of variations in temperature waveform pattern corresponding to 30 temperature sampling points and sticking temperature pattern; Model output variable is the number between-1 ~ 2, and when the variations in temperature waveform pattern that 30 temperature sampling points are corresponding is identical with sticking temperature pattern, model output variable is labeled as 1; When the temperature variation curve held stationary that 30 temperature sampling points are corresponding, when namely variations in temperature is completely normal, model output variable is labeled as 0.
Preparation and the pretreatment of data are as follows, according to mode input variable and model output variable, sample is extracted from the historical datas such as bonding and normal condition, reject imperfect and temperature data that is apparent error, obtain 611 groups of effective samples (effective sample is 611 groups of single thermocouples, 30 temperature sampling points continuous in time) altogether, in 611 groups of effective samples, have 141 groups to be sticking temperature pattern sample; Adopt formula (1) that above-mentioned 611 groups of effective samples are normalized to [-1,1] between, sample data after above normalization is divided into two parts, 502 groups of sample datas are wherein selected to be used for training pattern, these 502 groups of sample datas are called training sample, have 131 groups to be sticking temperature pattern sample in training sample; 109 groups of remaining sample datas are used for test model, and these 109 groups of sample datas are called test sample book, have 30 groups to be sticking temperature pattern sample in test sample book.
In formula, x' is the sample data after normalization, and x is the sample data before normalization, x
maxthe maximum of sample data before normalization, x
minit is the minimum of a value of sample data before normalization.
The foundation of GA-BP neutral net is as follows, and from analyzing above, BP neural network input layer nodes is 30, represents 30 temperature sampling points (i.e. mode input variable) that single thermocouple is continuous in time; BP neutral net output layer nodes is 1, and Output rusults is single thermocouple strike-alarm signal (i.e. model output variable), to represent on current thermocouple the variations in temperature waveform pattern that 30 sampled points are corresponding and sticking temperature pattern degree of closeness.
When reality is tested, determine a discrimination threshold scope according to the result of network training, just think when model output variable does not exceed the discrimination threshold scope preset and sticking temperature pattern detected.Training network adopts 3 layers of BP network, and hidden layer excitation function uses S type tangent transfer function, and output layer uses linear transfer function, and training process adopts Levenberg-Marquardt (LM) optimized algorithm.Through trial repeatedly, selected the number of hidden nodes is 12, obtains the BP neural network model that structure is 30 × 12 × 1, as shown in Figure 4.BP neural network learning process comprises information forward-propagating and error back propagation, according to given training sample input and output vector unceasing study and the connection weights and threshold adjusted between neuron, network is made constantly to approach mapping relations between sample input and output.The maximum frequency of training of BP neutral net is set to 2000, and learning rate is 0.05, and performance error is 0.0001.
For improving the generalization ability of BP neutral net, 1., initialization of population adopt genetic algorithm (GA) Optimized BP Neural Network, basic step comprises:, carries out real coding to all weights and threshold of neutral net, produce individual, population scale and evolutionary generation are set; 2., calculating the fitness function of each individuality, as shown in formula (2), take minimum of a value as optimum; 3., carry out genetic algorithm selection, crossover and mutation operation, produce individual population of new generation; 4., evaluate population of new generation, judge whether evolutionary generation reaches requirement or whether network error satisfies condition, if met, to obtain current population adaptive optimal control angle value corresponding individual, namely corresponding optimum BP neural network weight and threshold value.Wherein, Population in Genetic Algorithms scale is set to 50, and crossover probability is 0.7, and mutation probability is 0.06, and evolutionary generation is 200.Utilize genetic algorithm optimization BP neural network, after the training and testing of effective sample, obtain optimum neural network structure, for single even temporal model to the identification of temperature pattern.
In formula, n is network output layer nodes; y
ifor the actual output of BP i-th node; Y'
iit is the desired output of i-th node.
Adopt above-mentioned 502 groups of training samples, set up BP neural network model and GA-BP neural network model respectively, and respectively above-mentioned 109 groups of test sample books (wherein 30 groups is sticking temperature pattern sample) are predicted, predict the outcome in table 1 and Fig. 5.As shown in Figure 5, the predicted value of GA-BPNN model (i.e. GA-BP neural network model) and desired value degree of closeness comparatively BPNN model (i.e. BP neural network model) are high, illustrate that GA Optimized BP Neural Network improves the generalization ability of network, simultaneously from the recognition result of GA-BPNN model to 30 groups of sticking temperature pattern samples, also can determine that the threshold range of single even sticking temperature pattern-recognition is that [0.6,1.3] are comparatively suitable.As shown in Table 1, GA-BPNN model is high compared with the accuracy of identification of BPNN model, also illustrate that to improve single even temporal model to the recognition effect of sticking temperature pattern by genetic algorithm (GA) Optimized BP Neural Network.
The accuracy of identification of the single even temporal model of table 1
(2), even spatial model is organized;
As shown in Figure 2, after utilizing single even temporal model changing pattern identifying in time to electric thermo-couple temperatures all on crystallizer, Output rusults is saved in three-dimensional array Y (i, j, t), it represents that the i-th row j row thermocouple is the even temporal model recognition result (alarm signal) of the list of t.Here threshold range [the θ preset
min, θ
max] be [0.6,1.3] as the above analysis, when Y (i, j, t) is within the scope of this, think that this thermocouple TC (i, j) variations in temperature meets sticking temperature pattern, mark this thermocouple abnormal.So then carry out the judgement organizing even spatial model, calculate current thermocouple and be expert at and the abnormal thermocouple number of lastrow; Be specially the Y (i, j, t) of all thermocouples of inspection i-th row, add up at threshold range [θ
min, θ
max] in abnormal thermocouple number be m, check the Y (i-1, j, t) of the i-th row lastrow (i.e. the i-th-1 row) all thermocouples simultaneously, add up at threshold range [θ
min, θ
max] in abnormal thermocouple number be n, wherein i is greater than 1.If m and n is all more than or equal to 2, then need to check whether 10 seconds menisci capable (the first row) abnormal thermocouple numbers are more than or equal to 2 in the past, if met, then utilizing abnormal thermocouple sum (m+n) to warn thermocouple number threshold values to compare with strike-alarm with boning respectively, carrying out the judgement of strike-alarm and bonding warning.As from the foregoing, the even spatial model of the group that the present embodiment proposes judges whether neighboring thermocouple has the propagation of bonding by logic rules on the basis of single even temporal model, substantially increases the reliability that mixed model forecasts continuous casting bleed-out.
The continuous casting breakout prediction method based on mixed model of the present embodiment, comprise the steps: 1., in copper plate of crystallizer layout arranges high density thermocouple more, monitoring mould temperature situation of change, and gather and store on-the-spot all electric thermo-couple temperature real time datas, be saved in three-dimensional array T (i, j, t); 2., by the single even temporal model of all thermo-electric couple temperature data inputs, in single even temporal model, the time series data of each electric thermo-couple temperature, after the conversion and data processing of shift register, input GA-BP neural network model calculates, and judges variations in temperature waveform when whether the change of each electric thermo-couple temperature in time series meets bonding, judged result is saved in three-dimensional array Y (i, j, t) in; If 3. Y (i, j, t) is at the threshold range [θ of setting
min, θ
max] interior time, then think that current thermocouple TC (i, j) variations in temperature meets sticking temperature pattern, mark this thermocouple abnormal, so then carry out the judgement organizing even spatial model, calculate current thermocouple and be expert at and lastrow exception thermocouple number; 4. the abnormal thermocouple sum, by the even spatial model of group exported, respectively with strike-alarm with bond and warn thermocouple number threshold values to compare, carries out strike-alarm and the judgement of warning that bonds.Wherein, T (i, j, t) temperature value of the i-th row j row thermocouple in t is represented, Y (i, j, t) represent that the i-th row j row thermocouple is at the even temporal model recognition result of the list of t, reflects the degree of closeness of current thermocouple temperature pattern and sticking temperature pattern.
In the present embodiment, organize even spatial model and carried out testing (actual generation 14 bondings) to on-the-spot 97 heats of continuous casting, determining best strike-alarm thermocouple number threshold values is 6, and bonding warning thermocouple number threshold values is 3.And test result and existing Danieli system breakout prediction method are compared, as shown in table 2.Wherein, quote rate=true alarm times/(failing to report number of times+true alarm times), forecast accuracy=true alarm times/(failing to report number of times+true alarm times+wrong report number of times).As shown in Table 2, organize even spatial model and all can quote bonding, without failing to report, and to report number of times by mistake be 1, and forecast accuracy reaches 93.33%, is better than existing Danieli system breakout prediction method.It can thus be appreciated that the continuous casting breakout prediction method based on mixed model that the present embodiment proposes reaches good breakout prediction performance, can reducing wrong report and avoid failing to report, is a kind of effective breakout prediction method.
The test result of the even spatial model of table 2 group
The present invention considers the advantage of artificial intelligence technology on wave pattern cognition, when boning with strand, in copper plate of crystallizer, the spatial and temporal variation of electric thermo-couple temperature is foundation, adopt the even temporal model of GA-BP neural network list, be used for identifying variations in temperature waveform when whether the change of single electric thermo-couple temperature in time series meets bonding, belong to dynamic waveform pattern recognition problem.Wherein, genetic algorithm optimization BP neural network, determines BP neutral net best initial weights and threshold value by ability of searching optimum, improves single even temporal model to the accuracy of identification of sticking temperature waveform pattern.And consider the two-dimentional dissemination of bonding V-type tearing port, the basis of single even temporal model establishes the even spatial model of group by effective logic rules, judge whether neighboring thermocouple has the propagation of bonding, improve the accuracy of identification of sticker type breakout, especially can reduce failing to report and false alarm when multiple thermocouple fault or larger electric thermo-couple temperature fluctuate in actual production process; By the mixed model that the present invention proposes, make full use of the advantage of GA-BP neutral net in wave pattern cognition, and the effective logic rules that are coupled judge, the space-time not only achieving single even summation group idol judges, and overcome the deficiency that simple Logic judgment model parameter determines the guidance of difficult or inaccurate and simple model of mind shortage technique, reach good breakout prediction performance, promptly and accurately can quote whole bonding, avoid sticker type breakout accident, and false alarm frequency is down to floor level.
Schematically above be described the present invention and embodiment thereof, this description does not have restricted, and also just one of the embodiments of the present invention shown in accompanying drawing, actual structure is not limited thereto.So, if those of ordinary skill in the art enlightens by it, when not departing from the invention aim, designing the frame mode similar to this technical scheme and embodiment without creationary, all should protection scope of the present invention be belonged to.
Claims (8)
1. a mixed model, is characterized in that, mainly comprises following two parts: single even temporal model and the even spatial model of group;
(1), single even temporal model;
The structure of single even temporal model comprises: the determination of mode input variable, the determination of model output variable, the pretreatment of data, the foundation of GA-BP neutral net;
(2), even spatial model is organized;
1) after, utilizing single even temporal model changing pattern identifying in time to electric thermo-couple temperatures all on crystallizer, Output rusults is saved in three-dimensional array Y (i, j, t), wherein, Y (i, j, t) represents that the i-th row j row thermocouple is at the even temporal model recognition result of the list of t;
2), when Y (i, j, t) is at threshold range [θ
min, θ
max] interior time, think that this thermocouple TC (i, j) variations in temperature meets sticking temperature pattern, mark this thermocouple abnormal;
3), then check the Y (i, j, t) of all thermocouples of the i-th row, add up at threshold range [θ
min, θ
max] in abnormal thermocouple number be m, check the Y (i-1, j, t) of all thermocouples of the i-th-1 row simultaneously, add up at threshold range [θ
min, θ
max] in abnormal thermocouple number be n, wherein i is greater than 1;
4) if m and n is all more than or equal to 2, then check whether 10 seconds capable abnormal thermocouple numbers of menisci are more than or equal to 2 in the past, if met, then utilizing abnormal thermocouple sum m+n to warn thermocouple number threshold values to compare with strike-alarm with boning respectively, carrying out the judgement of strike-alarm and bonding warning.
2. a kind of mixed model according to claim 1, is characterized in that: the determination of mode input variable is as follows;
30 the temperature sampling points selecting single thermocouple continuous in time are as mode input variable;
Wherein, the collection period of temperature sampling point is 1 second.
3. a kind of mixed model according to claim 1, is characterized in that: the determination of model output variable is as follows;
Model output variable is single thermocouple strike-alarm signal, and model output variable is determined by the degree of closeness of variations in temperature waveform pattern corresponding to above-mentioned 30 temperature sampling points and sticking temperature pattern;
Wherein, model output variable is the number between-1 ~ 2, and when the variations in temperature waveform pattern that 30 temperature sampling points are corresponding is identical with sticking temperature pattern, model output variable is labeled as 1; When the temperature variation curve held stationary that 30 temperature sampling points are corresponding, model output variable is labeled as 0.
4. a kind of mixed model according to claim 1, is characterized in that: preparation and the pretreatment of data are as follows;
From historical data, extract sample, adopt formula (1) to normalize between [-1,1] by above-mentioned effective sample;
In formula (1), x' is the sample data after normalization, and x is the sample data before normalization, x
maxthe maximum of sample data before normalization, x
minit is the minimum of a value of sample data before normalization;
Sample data after above normalization is divided into two parts, and wherein a part is as training sample, and another part is as test sample book.
5. a kind of mixed model according to claim 1, is characterized in that: the foundation of GA-BP neural network model is as follows;
BP neural network input layer nodes is 30, representative model input variable; Output layer nodes is 1, representative model output variable; Determine a discrimination threshold scope according to the result of network training, just think when model output variable does not exceed the threshold range preset and sticking temperature pattern detected; Training network adopts 3 layers of BP network, and hidden layer excitation function uses S type tangent transfer function, and output layer uses linear transfer function, and training process adopts LM optimized algorithm; Selected the number of hidden nodes is 12, obtains the BP neural network model that structure is 30 × 12 × 1;
Wherein: BP neural network learning process comprises information forward-propagating and error back propagation, according to given training sample input and output vector unceasing study and the connection weights and threshold adjusted between neuron, network is made constantly to approach mapping relations between sample input and output; The maximum frequency of training of BP neutral net is set to 2000, and learning rate is 0.05, and performance error is 0.0001;
Adopt genetic algorithm optimization BP neural network, 1), initialization of population set up GA-BP neural network model, the basic step of genetic algorithm optimization BP neural network comprises:, carries out real coding to all weights and threshold of BP neutral net, produce individual, population scale and evolutionary generation are set; 2), calculating the fitness function of each individuality according to formula (2), take minimum of a value as optimum;
In formula (2), n is network output layer nodes; y
ibe the actual output of i-th node; Y'
iit is the desired output of i-th node; 3), carry out genetic algorithm selection, crossover and mutation operation, produce individual population of new generation; 4), evaluate population of new generation, judge whether evolutionary generation reaches requirement or whether network error satisfies condition, obtain the corresponding individuality of current population adaptive optimal control angle value if met; Wherein, Population in Genetic Algorithms scale is set to 50, and crossover probability is 0.7, and mutation probability is 0.06, and evolutionary generation is 200.
6. a kind of mixed model according to claim 1, is characterized in that: described threshold range [θ
min, θ
max] be [0.6,1.3].
7. a kind of mixed model according to claim 1 and 2, is characterized in that: strike-alarm thermocouple number threshold values is 6, and bonding warning thermocouple number threshold values is 3; When abnormal thermocouple sum m+n is more than or equal to 3, send bonding warning; When abnormal thermocouple sum m+n is more than or equal to 6, send strike-alarm.
8., based on a continuous casting breakout prediction method for mixed model as claimed in claim 1, it is characterized in that: comprise the steps:
Step (1), in copper plate of crystallizer, arrange many row's high density thermocouples, monitoring mould temperature situation of change, and gather and store on-the-spot all electric thermo-couple temperature real time datas, be saved in three-dimensional array T (i, j, t); Wherein, T (i, j, t) represents the temperature value of the i-th row j row thermocouple in t;
Step (2), by single for the input of all thermo-electric couple temperature datas even temporal model, in single even temporal model, the time series data of each electric thermo-couple temperature, after the conversion and data processing of shift register, input GA-BP neural network model calculates, and judges variations in temperature waveform when whether the change of each electric thermo-couple temperature in time series meets bonding, judged result is saved in three-dimensional array Y (i, j, t) in;
Step (3) if Y (i, j, t) setting threshold range [θ
min, θ
max] interior time, then think that current thermocouple TC (i, j) variations in temperature meets sticking temperature pattern, mark this thermocouple abnormal, so then carry out the judgement organizing even spatial model, calculate current thermocouple and be expert at and lastrow exception thermocouple number;
Step (4), the abnormal thermocouple sum that exported by the even spatial model of group, respectively with strike-alarm with bond and warn thermocouple number threshold values to compare, carry out strike-alarm and the judgement of warning that bonds.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510447779.6A CN105108097B (en) | 2015-07-24 | 2015-07-24 | Hybrid model for continuous casting breakout prediction |
| CN201710206080.XA CN106980729B (en) | 2015-07-24 | 2015-07-24 | A kind of continuous casting breakout prediction method based on mixed model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510447779.6A CN105108097B (en) | 2015-07-24 | 2015-07-24 | Hybrid model for continuous casting breakout prediction |
Related Child Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710206080.XA Division CN106980729B (en) | 2015-07-24 | 2015-07-24 | A kind of continuous casting breakout prediction method based on mixed model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105108097A true CN105108097A (en) | 2015-12-02 |
| CN105108097B CN105108097B (en) | 2017-05-10 |
Family
ID=54656329
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710206080.XA Active CN106980729B (en) | 2015-07-24 | 2015-07-24 | A kind of continuous casting breakout prediction method based on mixed model |
| CN201510447779.6A Active CN105108097B (en) | 2015-07-24 | 2015-07-24 | Hybrid model for continuous casting breakout prediction |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710206080.XA Active CN106980729B (en) | 2015-07-24 | 2015-07-24 | A kind of continuous casting breakout prediction method based on mixed model |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN106980729B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105522133A (en) * | 2015-12-21 | 2016-04-27 | 安徽工业大学 | Judgment method of bonding behavior between continuous casting slab and crystallizer copper plate |
| CN108436050A (en) * | 2018-04-16 | 2018-08-24 | 大连理工大学 | A method of continuous cast mold bleed-out is forecast using space density clustering DBSCAN |
| CN108801320A (en) * | 2017-04-28 | 2018-11-13 | 中国石油天然气股份有限公司 | A kind of diagnostic method of natural gas metering system |
| CN110523941A (en) * | 2019-08-06 | 2019-12-03 | 北京数钰科技发展有限公司 | Continuous casting bonding steel leakage multistage risk control method and control device |
| CN111570748A (en) * | 2020-04-28 | 2020-08-25 | 中冶南方连铸技术工程有限责任公司 | Crystallizer bleed-out forecasting method based on image processing |
| CN113579190A (en) * | 2021-06-24 | 2021-11-02 | 邯郸钢铁集团有限责任公司 | Slab continuous casting breakout prediction method based on regional characteristics |
| CN113814369A (en) * | 2021-09-28 | 2021-12-21 | 大连理工大学 | Crystallizer bleed-out forecasting method based on eigenvector and Adaboost integrated model |
| CN115446276A (en) * | 2022-10-05 | 2022-12-09 | 大连理工大学 | Continuous casting breakout early warning method for recognizing V-shaped bonding characteristics of crystallizer copper plate based on convolutional neural network |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6591509B2 (en) * | 2017-11-06 | 2019-10-16 | 株式会社東芝 | Mold temperature abnormality sign detection device and program |
| CN108097905B (en) * | 2017-12-28 | 2020-02-28 | 北京首钢自动化信息技术有限公司 | Steel leakage early warning method |
| CN109396375B (en) * | 2018-12-11 | 2019-09-27 | 大连理工大学 | A kind of crystallizer bleedout prediction electric thermo method based on feature vector and hierarchical clustering |
| CN109365769B (en) * | 2018-12-18 | 2021-03-30 | 重庆邮电大学 | Crystallizer bleed-out forecasting method based on mixed model judgment |
| CN110918973B (en) * | 2019-12-11 | 2021-04-27 | 东北电力大学 | Crystallizer thermal image abnormal region marking method based on run |
| CN111112567B (en) * | 2019-12-16 | 2021-09-21 | 河钢股份有限公司 | Comprehensive judgment method for breakout prediction of slab caster |
| CN112926622B (en) * | 2021-01-22 | 2022-10-18 | 大连理工大学 | Crystallizer breakout prediction method for generating countermeasure network based on feature vector and SWGAN-GP |
| CN113935416B (en) * | 2021-10-15 | 2024-06-25 | 东北大学 | Continuous casting breakout prediction method of multi-state crystallizer |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101477374A (en) * | 2009-01-20 | 2009-07-08 | 东北大学 | Continuous casting bleed-out time sequence spacing combined diagnosis prediction method based on fuzzy neural network |
| CN101850410A (en) * | 2010-06-22 | 2010-10-06 | 攀钢集团钢铁钒钛股份有限公司 | Continuous casting breakout prediction method based on neural network |
| CN101934353A (en) * | 2009-06-30 | 2011-01-05 | 上海宝信软件股份有限公司 | Device and method for breakout prediction during slab continuous casting |
| CN102151814A (en) * | 2010-12-30 | 2011-08-17 | 中冶连铸技术工程股份有限公司 | Bonding alarm method and system during continuous casting production |
| CN102554171A (en) * | 2011-12-21 | 2012-07-11 | 天津钢铁集团有限公司 | Breakout prediction method for continuous casting |
| CN102699302A (en) * | 2012-07-10 | 2012-10-03 | 中冶赛迪电气技术有限公司 | Bleed-out forecasting system and forecasting method of slab continuous casting crystallizer |
| CN103878335A (en) * | 2014-02-10 | 2014-06-25 | 河北钢铁股份有限公司邯郸分公司 | Method for preventing crystallizer of continuous casting machine from sticking breakout |
| CN104484715A (en) * | 2014-11-28 | 2015-04-01 | 江苏大学 | Neural network and particle swarm optimization algorithm-based building energy consumption predicting method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5755782A (en) * | 1991-01-24 | 1998-05-26 | Autogenics | Stents for autologous tissue heart valve |
| CN101985166B (en) * | 2009-07-29 | 2012-04-25 | 宝山钢铁股份有限公司 | Continuous casting slab longitudinal-crack steel leakage prediction method |
-
2015
- 2015-07-24 CN CN201710206080.XA patent/CN106980729B/en active Active
- 2015-07-24 CN CN201510447779.6A patent/CN105108097B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101477374A (en) * | 2009-01-20 | 2009-07-08 | 东北大学 | Continuous casting bleed-out time sequence spacing combined diagnosis prediction method based on fuzzy neural network |
| CN101934353A (en) * | 2009-06-30 | 2011-01-05 | 上海宝信软件股份有限公司 | Device and method for breakout prediction during slab continuous casting |
| CN101850410A (en) * | 2010-06-22 | 2010-10-06 | 攀钢集团钢铁钒钛股份有限公司 | Continuous casting breakout prediction method based on neural network |
| CN102151814A (en) * | 2010-12-30 | 2011-08-17 | 中冶连铸技术工程股份有限公司 | Bonding alarm method and system during continuous casting production |
| CN102554171A (en) * | 2011-12-21 | 2012-07-11 | 天津钢铁集团有限公司 | Breakout prediction method for continuous casting |
| CN102699302A (en) * | 2012-07-10 | 2012-10-03 | 中冶赛迪电气技术有限公司 | Bleed-out forecasting system and forecasting method of slab continuous casting crystallizer |
| CN103878335A (en) * | 2014-02-10 | 2014-06-25 | 河北钢铁股份有限公司邯郸分公司 | Method for preventing crystallizer of continuous casting machine from sticking breakout |
| CN104484715A (en) * | 2014-11-28 | 2015-04-01 | 江苏大学 | Neural network and particle swarm optimization algorithm-based building energy consumption predicting method |
Non-Patent Citations (3)
| Title |
|---|
| 何飞: "板坯连铸结晶器漏钢预报系统的开发与实践", 《连铸》 * |
| 何飞等: "连铸结晶器漏钢预报模型研究进展", 《铸造技术》 * |
| 张本国: "薄板坯连铸神经网络漏钢预报系统研究", 《中国博士学位论文全文数据库 工程科技Ⅰ辑》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105522133B (en) * | 2015-12-21 | 2018-04-06 | 安徽工业大学 | A kind of continuous casting billet bonds the decision method of behavior with copper plate of crystallizer |
| CN105522133A (en) * | 2015-12-21 | 2016-04-27 | 安徽工业大学 | Judgment method of bonding behavior between continuous casting slab and crystallizer copper plate |
| CN108801320B (en) * | 2017-04-28 | 2020-12-01 | 中国石油天然气股份有限公司 | Diagnosis method of natural gas measurement system |
| CN108801320A (en) * | 2017-04-28 | 2018-11-13 | 中国石油天然气股份有限公司 | A kind of diagnostic method of natural gas metering system |
| CN108436050A (en) * | 2018-04-16 | 2018-08-24 | 大连理工大学 | A method of continuous cast mold bleed-out is forecast using space density clustering DBSCAN |
| CN108436050B (en) * | 2018-04-16 | 2019-06-21 | 大连理工大学 | A method of continuous cast mold bleed-out is forecast using space density clustering DBSCAN |
| CN110523941A (en) * | 2019-08-06 | 2019-12-03 | 北京数钰科技发展有限公司 | Continuous casting bonding steel leakage multistage risk control method and control device |
| CN111570748A (en) * | 2020-04-28 | 2020-08-25 | 中冶南方连铸技术工程有限责任公司 | Crystallizer bleed-out forecasting method based on image processing |
| CN111570748B (en) * | 2020-04-28 | 2021-08-06 | 中冶南方连铸技术工程有限责任公司 | Crystallizer bleed-out forecasting method based on image processing |
| CN113579190A (en) * | 2021-06-24 | 2021-11-02 | 邯郸钢铁集团有限责任公司 | Slab continuous casting breakout prediction method based on regional characteristics |
| CN113814369A (en) * | 2021-09-28 | 2021-12-21 | 大连理工大学 | Crystallizer bleed-out forecasting method based on eigenvector and Adaboost integrated model |
| CN113814369B (en) * | 2021-09-28 | 2022-11-18 | 大连理工大学 | Crystallizer bleed-out forecasting method based on eigenvector and Adaboost integrated model |
| CN115446276A (en) * | 2022-10-05 | 2022-12-09 | 大连理工大学 | Continuous casting breakout early warning method for recognizing V-shaped bonding characteristics of crystallizer copper plate based on convolutional neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105108097B (en) | 2017-05-10 |
| CN106980729B (en) | 2018-08-17 |
| CN106980729A (en) | 2017-07-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105108097A (en) | Hybrid model and hybrid-model-based continuous casting breakout prediction method | |
| He et al. | Mold breakout prediction in slab continuous casting based on combined method of GA-BP neural network and logic rules | |
| CN105328155B (en) | A kind of bleed-out visual feature forecasting procedure based on improvement neutral net | |
| CN109396375B (en) | A kind of crystallizer bleedout prediction electric thermo method based on feature vector and hierarchical clustering | |
| CN102705078B (en) | Diesel engine fault prediction method based on gray model | |
| CN111444940A (en) | Fault diagnosis method for critical parts of fan | |
| CN108436050B (en) | A method of continuous cast mold bleed-out is forecast using space density clustering DBSCAN | |
| CN102886504B (en) | Crystallizer bleed-out visual forecasting method based on machine vision | |
| CN102554171B (en) | Breakout prediction method for continuous casting | |
| CN107742008A (en) | A kind of fault early warning method of gearbox of wind turbine | |
| CN105759201A (en) | High voltage circuit breaker self-diagnosis method based on abnormal sample identification | |
| CN104267610B (en) | High-precision blast furnace smelting process abnormal data detection and repair method | |
| CN107844067A (en) | A kind of gate of hydropower station on-line condition monitoring control method and monitoring system | |
| CN104181196B (en) | A kind of continuous casting billet surface longitudinal crack online test method | |
| CN105388884A (en) | Alarm system for detecting leakage fault of heat supply network based on identification algorithm driven by data and method | |
| CN109902265A (en) | A kind of underground method for early warning based on hidden Markov model | |
| Duan et al. | Modeling of breakout prediction approach integrating feature dimension reduction with K-means clustering for slab continuous casting | |
| CN112434887A (en) | Water supply network risk prediction method combining network kernel density estimation and SVM | |
| CN114925867A (en) | Internet of things maintenance prediction system for key equipment in seamless steel tube production | |
| CN104697585A (en) | Cooling tower tower-barrel strength predicating control system and method thereof | |
| CN113935416B (en) | Continuous casting breakout prediction method of multi-state crystallizer | |
| CN118130984B (en) | Cable partial discharge fault real-time monitoring method based on data driving | |
| CN115446276B (en) | Continuous casting steel leakage early warning method based on convolutional neural network recognition crystallizer copper plate V-shaped bonding characteristics | |
| CN117688834B (en) | Method for early warning insulation overheat fault of coil of water-hydrogen turbine generator | |
| Li et al. | Sensor Failure Identification in Industrial Big Data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |