METHOD AND SYSTEM FOR PREDICTING THE TEMPERATURE OF AN OVEN, DEVICE AND SUPPORT

Abstract

The application provides a method for predicting the temperature of a furnace, which includes: a historical dataset is obtained using monitored furnace data; a first training set and a first test set are acquired based on the historical dataset; a model is initially trained and tested; the model is trained and tested again using a second pre-acquired training set and a second pre-acquired test set; and an optimal temperature prediction model is discarded based on an output error from the second test set; and a predicted temperature in the furnace temperature region within a prediction time period is obtained based on furnace data for a specific temperature region within a predefined early prediction time period and the optimal temperature prediction model. Figure 1

Description

Title of the invention: METHOD AND SYSTEM FOR PREDICTING THE TEMPERATURE OF AN OVEN, DEVICE AND SUPPORT Technical field

[0001] The present application relates to the technical field of temperature prediction, for example a method for predicting the temperature of a furnace, a system for predicting the temperature of a furnace, a device and a support.

[0002] BACKGROUND

[0003] An industrial furnace is a device made of refractory material intended for calcining materials or sintering product articles. Industrial furnaces are primarily and widely used in the machinery industry, the metallurgical industry, the petroleum industry, the coal gas industry, and similar industries. The creation and development of industrial furnaces play an important role in human progress.

[0004] In the prior art, the main components of a conventional industrial furnace include the furnace masonry, the furnace flue gas evacuation system, the furnace preheater, the furnace combustion apparatus, and other similar components. In a conventional industrial furnace, to accelerate the sintering rate of a product and shorten the sintering time, a method is generally employed to increase or decrease the fuel supply or to modify the power output to control the temperature inside the furnace by an operator. However, due to the instability of this artificial control, it is inevitable that the power adjustment will be too large or too small during the process of an artificial operation, causing the furnace temperature to rise too quickly or too late and thus affecting the sintering of the product.

[0005] Once the raw materials for the cathode material are mixed, they are placed in the furnace to begin sintering. The raw materials pass through a temperature-increasing region, a heat-conservation region, and a temperature-decreasing region at a fixed rate within the furnace. Each temperature region must be maintained at its corresponding temperature for sintering. Under normal operating conditions, the temperature of each temperature region fluctuates within a stable range, but there are instances of larger temperature fluctuations, which cause a significant difference between the actual sintering temperature of the furnace and the set temperature, thus affecting the material's performance and increasing the cost.

[0006] SUMMARY

[0007] The present application provides a method for predicting the temperature of a furnace to predict a future temperature of a furnace.

[0008] In order to achieve the above objective, in a first aspect, the present application provides a method for predicting the temperature of a furnace. The method comprises the following operations.

[0009] The furnace data for any temperature region of the furnace in a defined time range of the temperature region corresponding to any temperature region of the furnace are acquired, and a historical data set is formed, where the furnace data include temperature data and power data.

[0010] A first training set and a first test set are acquired according to the historical data set, and multiple initial temperature prediction models selected in advance are trained using the first training set, to obtain multiple first temperature prediction models.

[0011] Multiple first temperature prediction models are tested using the first test set to output multiple first test set output errors, and multiple second temperature prediction models are discarded from the multiple first temperature prediction models according to the multiple first test set output errors and a predefined range of a first test error; the multiple second temperature prediction models are trained using a second training set acquired in advance, to obtain multiple third temperature prediction models;Multiple third temperature prediction models are tested against a second set of pre-acquired tests, to deliver multiple output errors from the second set of tests, and a third temperature prediction model corresponding to a minimum value of the multiple output errors from the second set of tests is used as the optimal temperature prediction model for any temperature region of the furnace; and the furnace data to be predicted from any temperature region of the furnace, collected in a predefined time period of the early prediction stage, are entered into the optimal temperature prediction model corresponding to any temperature region of the furnace, in order to obtain a predicted temperature in the temperature region of the furnace in a prediction time period.

[0012] Furthermore, the operation of acquiring the first training set and the first test set according to the historical data set includes: selecting furnace data for any temperature region of the furnace within a defined abnormal temperature range corresponding to the temperature region any of the furnace according to the historical dataset, and the use of the furnace data as the historical dataset to be processed; performing preprocessing on the historical dataset to be processed to obtain a preprocessed historical dataset, where the preprocessing includes data cleaning processing; and acquiring the first training set and the first test set according to the preprocessed historical dataset.

[0013] In addition, the data cleaning process includes: in the case where historical abnormal jump data of the furnace exist in the historical data set to be processed, replacing the historical abnormal jump data of the furnace with a median value of all the historical furnace data in the historical data set to be processed until the historical abnormal jump data of the furnace no longer exists in the historical data set to be processed, in which the historical abnormal jump data of the furnace are temperature data between which a difference value and adjacent historical furnace data exceed a predefined difference value.

[0014] In addition, each of the second training set and the second test set includes historical furnace data for any temperature region of the furnace in the predefined early prediction stage time period, the predefined early prediction stage time period includes a normal early prediction stage time period and an early prediction stage failure time period.

[0015] In addition, the initial temperature prediction models include a long- and short-term memory (LSTM) model, a recurrent neural network (RNN) model, a closed recurrent unit (GRU) model, a convolutional neural network (CNN) model, and a graph neural network (GNN) model.

[0016] In one embodiment, the method further comprises: acquiring a plurality of statistical modules according to the oven data for any temperature region of the oven, and obtaining a temperature distribution characteristic of any temperature region of the oven according to the plurality of statistical modules; and obtaining a temperature variation trend of any temperature region of the oven according to the temperature distribution characteristic.

[0017] In addition, the statistical modules include a mean, a variance, a deviation and a kurtosis.

[0018] In a second aspect, the present application further provides a system for predicting the temperature of a furnace. The system comprises a data acquisition module, a first filtering module, a second filtering module, and a temperature prediction module.

[0019] The data acquisition module is configured to acquire furnace data for any temperature region of the furnace within a defined time range of the temperature region corresponding to any temperature region of the furnace, and to form a historical data set; where the furnace data includes temperature data and power data.

[0020] The first filtering module is configured to acquire a first training set and a first test set according to the historical data set, and train multiple pre-selected initial temperature prediction models using the first training set, in order to obtain multiple first temperature prediction models; test the multiple first temperature prediction models using the first test set to output multiple first test set output errors, and discard multiple second temperature prediction models from the multiple first temperature prediction models according to the multiple first test set output errors and a predefined range of a first test error.

[0021] The second filtering module is configured to train multiple second temperature prediction models using a second training set acquired in advance, in order to obtain multiple third temperature prediction models; test the multiple third temperature prediction models according to a second test set acquired in advance, in order to output multiple output errors from the second test set, and use a third temperature prediction model corresponding to a minimum value of the multiple output errors from the second test set as the optimal temperature prediction model for any temperature region of the furnace.

[0022] The temperature prediction module is configured to input the data from the furnace to be predicted from any temperature region of the furnace collected in a predefined time period of early prediction stage into the optimal temperature prediction model corresponding to any temperature region of the furnace, to obtain a prediction temperature in the temperature region of the furnace in a prediction time period.

[0023] In a third aspect, the present application further provides a computer device. The computer device comprises a memory, a processor, and a computer program stored in the memory and capable of running on the processor. The computer program, when executed by the processor, implements the steps of the process described above.

[0024] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores A computer program. The computer program, when executed by a processor, implements the steps of the process described above.

[0025] The present application provides a method for predicting the temperature of a furnace, a system for predicting the temperature of a furnace, a device and a support.The process includes the following: a temperature region of the furnace is monitored to acquire furnace data including temperature and power data; a historical dataset is formed using the furnace data; a first training set and a first test set are acquired from the historical dataset; a prediction model is trained and initially tested using the first training set and the first test set, respectively; the prediction model is primarily filtered; second temperature prediction models are acquired; to filter the second temperature prediction models, the primarily filtered prediction model is trained and tested again using a second training set and the first test set, respectively; and an optimal temperature prediction model is discarded.In this way, an accurate prediction of each temperature region of the furnace is obtained. According to this application, by accurately predicting the future temperature of the furnace, the temperature variation trend of each temperature region is predicted in advance, and the influence of large temperature fluctuations on the materials is reduced. Brief description of the drawings.

[0026] [Fig. 1] is a flowchart of a method for predicting the temperature of the oven according to an embodiment of the present application;

[0027] [Fig.2] is a schematic diagram of a time prediction of a process predicting the temperature of an oven according to an embodiment of the present application;

[0028] [Fig.3] is a functional diagram of a temperature prediction system for a oven according to a method of implementation of this application; and

[0029] [Fig.4] is a diagram of an internal structure of a computer device according to a method of carrying out this request. DETAILED DESCRIPTION

[0030] In order to clarify the objective, the technical solution, and the beneficial effects of this application, the present application will be described below in conjunction with the accompanying drawings and embodiments. The embodiments described below are apparently part of the embodiments of this application and are intended to be purely illustrative, but are not intended to limit the scope of this application. Based on the embodiments of this request, all other embodiments obtained by a person skilled in the art without creative work fall within the scope of protection of the present invention.

[0031] In one embodiment, as shown in [Fig. 1], an embodiment of the present application provides a method for predicting the temperature of a furnace. The method for predicting the temperature of a furnace comprises the steps described below.

[0032] In SI 1, furnace data for any temperature region of the furnace within a defined time range of the temperature region corresponding to that temperature region are acquired, and a historical data set is formed, where the furnace data includes temperature data and power data, and the defined time range of the temperature region includes a normal defined temperature range and an abnormal defined temperature range. The duration of the normal defined temperature range can be defined by a person skilled in the art depending on the specific case, which is not specifically limited in the embodiments of this application.

[0033] In this embodiment, the temperature region measurement point position number data is acquired by the factor, and the temperature and power data are acquired at 1-minute intervals within the normal temperature defined time range and the abnormal temperature defined time range. In other embodiments, other collection methods and other collection intervals may be used to collect the temperature and power data from the temperature region, and the details are not described here.

[0034] An example of this embodiment is the following.

[0035] For different temperatures defined in different temperature regions, temperature and power data from the same temperature region (such as a 4u temperature region of multiple furnaces) of different furnaces within the defined abnormal temperature time range are selected in this embodiment. Optionally, the defined abnormal temperature time range is set to 6 days before and after the temperature anomaly, as shown in Table 1.

[0036] Table 1 Schedule for acquiring values ​​from the furnace region

[0037] [Tables 1] Oven number - Temperature region - Start time - End time - Oven #6 4u 08-04-2022 14-04-2022 - Oven #5 4u 10-04-2022 16-04-2022 Oven n°7 4u 05-15-2022 05-21-2022 Oven n°2 4u 05-17-2022 05-23-2022 Oven n°l 4u 05-21-2022 05-27-2022 Oven n°8 4u 06-08-2022 06-14-2022

[0038] The detection of oven temperature data may also not be limited to temperature and power data only, and other oven data may be added if necessary to make the prediction result more accurate.

[0039] In S12, a first training set and a first test set are acquired according to the historical data set, and multiple initial temperature prediction models selected in advance are trained using the first training set, to obtain multiple first temperature prediction models.

[0040] In this embodiment, the step of acquiring the first training set and the first test set according to the historical data set includes: selecting the furnace data for any temperature region of the furnace within a defined abnormal temperature time range corresponding to the any temperature region of the furnace according to the historical data set, and using the furnace data as the historical data set to be processed, where, optionally, in this embodiment, the defined abnormal temperature time range is set to 6 days before and after the temperature anomaly, and the defined abnormal temperature time range can be specifically set by a person skilled in the art according to the actual conditions, which are not limited in this application;the performance of preprocessing on the historical dataset to be processed to obtain a preprocessed historical dataset, where the preprocessing includes data cleaning; and the acquisition of the first training set and the first test set according to the preprocessed historical dataset.

[0041] The data cleaning process includes: in the case where historical abnormal furnace jump data exists in the historical dataset to be processed, replacing the historical abnormal furnace jump data with a median value of all the historical furnace data in the historical dataset to be processed until the historical abnormal furnace jump data no longer exists in the historical dataset to be processed, where the historical abnormal furnace jump data is temperature data between which a difference value and nearby historical furnace data exceed a predefined difference value. For example, the historical abnormal furnace jump data is temperature data between which a difference value and adjacent historical data from the furnace exceed a predefined difference value.

[0042] In this embodiment, the historical data acquired for processing are cleaned before training the prediction model, such as data that suddenly changed to 0, i.e., a value at a time when the data suddenly changed to 0. A replacement filling process is performed using a median from a sample population, so as to obtain a set of cleaned historical data and thus avoid error interference caused by the sudden change in the data. For example, when temperature data with a value of 0 exist in the furnace data, the temperature data with the value of 0 are replaced by a median value from all the temperature data until the temperature data with the value of 0 no longer exist in the furnace data.When power data with a value of 0 exists in the furnace data, the power data with a value of 0 is replaced by a median value from all power data until power data with a value of 0 no longer exists in the furnace data.

[0043] In S13, the first multiple temperature prediction models are tested using the first test set to output multiple output errors from the first test set, and multiple second temperature prediction models are discarded from the first multiple temperature prediction models according to the multiple output errors from the first test set and a predefined range of a first test error.

[0044] In S14, the multiple second temperature prediction models are trained using a second training set acquired in advance, to obtain multiple third temperature prediction models.

[0045] In S15, the multiple third temperature prediction models are tested according to a second test set acquired in advance, to deliver multiple output errors from the second test set, and a third temperature prediction model corresponding to a minimum value of the multiple output errors from the second test set is used as the optimal temperature prediction model for any temperature region of the furnace.

[0046] In S16, the data from the furnace to be predicted from any temperature region of the furnace, collected in a predefined time period of the early prediction stage, are entered into the optimal temperature prediction model corresponding to any temperature region of the furnace, to obtain a predicted temperature in the temperature region of the furnace in a prediction time period, where the data from the furnace to be predicted include temperature data to be predicted and power data to be predicted.

[0047] In this embodiment, the historical dataset comprises a first training set and a first test set, and the historical dataset generally includes data from the 6 days preceding the temperature anomaly time and data from the 6 days following the temperature anomaly time. Each of the second training set and the second test set includes historical furnace data for the furnace temperature region within the predefined early prediction stage time period, where the predefined early prediction stage time period includes a normal time period and a failure time period from an early stage of a time period at which a temperature is to be predicted, and the historical furnace data includes historical temperature data and historical power data.

[0048] According to this embodiment, the prediction model is trained and tested twice, so that the third temperature prediction model corresponding to the minimum output error value of the second test set is discarded to be the optimal temperature prediction model for the corresponding temperature region of the furnace. If a suitable dataset is selected, it is also possible to perform the model training and test filtering processes three or more times.

[0049] In this embodiment, the prediction model comprises a long- and short-term memory (LSTM) model, a recurrent neural network (RNN) model, a closed recurrent unit (GRU) model, a convolutional neural network (CNN) model, and a graph neural network (GNN) model. Optionally, in this embodiment, an LSTM-based method can be adopted to predict temperature data on a future occasion based on historical data.

[0050] For example, it is necessary to perform a 12-hour advance warning for a heating rod in order to provide sufficient material preparation time so that a temperature prediction for the following 12 hours can be made. Considering that there is also heating power data, the heating power data and the temperature region data of the current period are entered together into the optimal temperature prediction model as a feature to predict the furnace temperature data in the following 12 hours; that is, the temperature (t) and the power (p) in the time period from 1 to n are used to predict the temperature value in the opportunity interval from n+1 to k, as shown in [Fig. 2].

[0051] In actual modeling, a time step is first determined; for example, 24 hours is used as the time step to predict temperature data for the following 12 hours, and the time periods are randomly selected from the temperature range of the desired prediction temperature for prediction and inspection.

[0052] In other embodiments of the present application, other furnace temperature regions may be selected for prediction and testing under actual conditions. For example, data from 30 days prior to the repair time may be selected according to the maintenance fault log, and the data used for prediction may be discarded to train the initial temperature prediction model.

[0053] Optionally, in the present application, a trained model is defined for prediction in different temperature regions of different furnaces. The optimal temperature prediction models used for different temperature regions are different. When the temperature prediction in a temperature region is performed, the prediction model is retrained using a dataset prior to the prediction time, and the parameters of the optimal temperature prediction model are adjusted.

[0054] Once the heating device has been replaced in the oven, it is necessary to repeat the model training and filtering step and select an appropriate prediction model.

[0055] In this embodiment, once the predicted temperature in the furnace temperature region is obtained, the method further comprises the following: in the event that the predicted temperature in the furnace temperature region exceeds a predefined temperature threshold, a communication equipment address of a furnace-responsible person corresponding to any furnace is acquired; and prompt information is sent to the equipment address of the furnace-responsible person.

[0056] If the predicted temperature in the furnace temperature region exceeds a first predefined value, the communication equipment address of the person in charge of the furnace is called and a prompt voice is played. If the predicted temperature in the furnace temperature region exceeds a second predefined value, a short message is sent to the communication equipment address of the person in charge of the furnace as a prompt. The first predefined value is greater than the second predefined value.

[0057] In the event that the predicted temperature fluctuates considerably, a message is transmitted to the person responsible concerned so that they can manage the condition of greater temperature fluctuation in advance.

[0058] The responsible person concerned can define different alarm levels and adopt different reminder methods such as calls or text messages for different alarm levels, in order to implement countermeasures in a more targeted manner.

[0059] In one embodiment, the method for predicting the temperature of a furnace provided in this embodiment further comprises the following: multiple statistical modules are acquired from the furnace data for any temperature region of the furnace, and a temperature distribution characteristic of any temperature region of the furnace is obtained from the multiple statistical modules, where the multiple statistical modules include a mean, a variance, a deviation, and a kurtosis; and a temperature variation trend of any temperature region of the furnace is obtained from the temperature distribution characteristic.

[0060] In this embodiment, various statistical modules are used to analyze furnace data (temperature region data, power data) to obtain the temperature distribution characteristic of the furnace temperature region. The temperature within the furnace temperature region is analyzed using this temperature distribution characteristic; for example, the temperature distribution characteristic is analyzed to determine the temperature variation trend of the furnace temperature region. When an abnormal temperature is detected in the furnace temperature region, it indicates that the heating element must be replaced. In this embodiment, different statistical modules are used to analyze and calculate furnace data before and after the temperature region anomaly to obtain the temperature distribution characteristic.The temperature variation trend before and after the anomaly in the temperature region is obtained using the temperature distribution characteristic. The temperature variation trends before and after the anomaly in the same temperature region across different furnaces are compared and analyzed to determine the operating condition of the heating element or the accuracy of the heating element maintenance schedule based on the temperature variation trend before and after the anomaly. For example, when it is determined that the temperature changes significantly according to the temperature variation trends before and after the anomaly, this indicates a change in the stability of the heating element in the temperature region. Characteristics such as the quality of the heating element are assessed by a person skilled in the art based on the temperature variation trend before and after the anomaly.Assuming that the oven data is X = [% 7 x 2, ■■■, x „], x, denotes the ith oven data and n denotes the number of elements included in the oven data, the statistical value of X can be expressed as follows: .

[0061] [Math.l] average: u =

[0062] [Math.2] variance:a2 = | ££.-109 - u)2

[0063] [Math.3] difference: v3 = E (—)' = =

[0064] [Math.4] R-, he, -SlLOù-Ul* flattening = -f = 4 - 3. = —— - 3 (2¾^^)

[0065] The mean represents a central position of the data, and if X approximately follows a Gaussian distribution, then the data are essentially distributed on both sides of the mean. The variance is designated as the degree of deviation of the data variable from the mean. The variance is also a second-order central moment of the variable X.

[0066] The deviation characterizes the degree of asymmetry of the probability distribution density function curve with respect to the mean and measures the asymmetry of the probability distribution of random variables.

[0067] The kurtosis characterizes the state of the probability distribution density function curve at the mean, that is to say that if the kurtosis is less than 3, then the distribution is soft, and if the kurtosis is greater than 3, then the distribution is steep.

[0068] For example, this embodiment can generate a histogram of the temperature and feed frequency distribution of temperature region 4u of furnace No. 7 (7#4u) over a predetermined time. It has been analyzed that the temperature is intensely distributed in the vicinity of 700°C, and the power data of temperature region 4u of furnace No. 7 can be considered to approximately obey a Gaussian distribution. Therefore, a discriminant analysis can be performed on the temperature anomaly according to the characteristics of the temperature and power data in this embodiment.

[0069] In another embodiment, an autoregressive integrated moving average (ARIMA) prediction algorithm is also adopted for prediction. In this embodiment, the ARIMA prediction algorithm focuses mainly on the change in data at a certain time node in the future, and ignores the process data within it. More specifically, it focuses on the position of the data node after 12 hours, and not on how the data varies over 12 hours.

[0070] According to the method for predicting the temperature of a furnace provided in this application, the prediction of the future temperature of the furnace is obtained, the trend of temperature variation of the multiple temperature regions is predicted in advance, the technical personnel are helped to discover the abnormal temperature of the furnace in advance, determine in advance and take the corresponding measures to reduce the influence of large temperature fluctuations on the materials.

[0071] Based on the furnace temperature prediction method described above, an embodiment of the present application further provides a furnace temperature prediction system. As shown in [Fig. 3], the system comprises a data acquisition module 1, a first filtering module 2, a second filtering module 3, and a temperature prediction module 4. The data acquisition module 1 is configured to acquire furnace data for any temperature region of the furnace within a defined time range of the temperature region corresponding to that temperature region, and to form a historical dataset; where the furnace data includes temperature data and power data.The first filtering module 2 is configured to acquire a first training set and a first test set from the historical data set, and train multiple pre-selected initial temperature prediction models using the first training set, to obtain multiple first temperature prediction models; test the multiple first temperature prediction models using the first test set to output multiple first test set output errors, and discard multiple second temperature prediction models from the multiple first temperature prediction models according to the multiple first test set output errors and a predefined range of a first test error.The second filtering module 3 is configured to train multiple second temperature prediction models using a second pre-acquired training set, in order to obtain multiple third temperature prediction models; test the multiple third temperature prediction models according to a second pre-acquired test set, in order to output multiple output errors from the second test set, and use a third temperature prediction model corresponding to a minimum value of the multiple output errors from the second test set as the optimal temperature prediction model for any temperature region of the furnace. The temperature prediction module 4 is configured to... input the oven data to be predicted from any temperature region of the oven collected in a predefined early prediction stage time period into the optimal temperature prediction model corresponding to any temperature region of the oven, to obtain a predicted temperature in the oven temperature region in a prediction time period.

[0072] For specific definitions of the oven temperature prediction system, reference may be made to the definitions described above for the oven temperature prediction method, and details are not described here. Multiple modules in the system described above may be implemented wholly or partially by software, hardware, or a combination of software and hardware. The multiple modules described above may be integrated into or independent of the processor in the computing device as hardware, and may also be stored in the computing device's memory as software, so that the processor can invoke and execute operations corresponding to the multiple modules described above.

[0073] Figure 4 shows a schematic diagram of the internal structure of a computing device in one embodiment, and the computing device may be a terminal or a server. As shown in Figure 4, the computing device comprises a processor, memory, a network interface, a display, and an input device connected by a system bus. The processor of the computing device is configured to provide computing and control capabilities. The memory of the computing device comprises non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage media. The network interface of the computing device is configured to communicate with an external terminal via a network connection.The display screen of the computing device may be a liquid crystal display or an electronic ink display, and the input device of the computing device may be a touch layer applied to the display screen, or may be a key, trackball or touchpad disposed on a housing of the computing device, or may be an external keyboard, touchpad or mouse and other similar devices.

[0074] It should be understood by those skilled in the art that the structure shown in [Fig. 4] is a functional diagram of only a part of the structure associated with the diagrams in this application, and does not constitute a definition of the computer device to which the diagrams in this application are applied, and that the computer device may include more or fewer components than those shown in the designs, or may combine certain components, or have the same arrangement of components.

[0075] In summary, the present application provides a method for predicting the temperature of a furnace, a furnace temperature prediction system, a device, and a support. The method comprises the following: a historical dataset is obtained using monitored furnace data; a first training set and a first test set are acquired based on the historical dataset; a prediction model is initially trained and tested using the first training dataset and the first test set, respectively, and the prediction model is filtered primarily based on an output error from the first test set; the prediction model is trained and tested again using a second pre-acquired training set and a second pre-acquired test set, respectively, and an optimal temperature prediction model is discarded based on an output error from the second test set;and a predicted temperature in the furnace temperature region within a prediction time period is obtained based on furnace data for a furnace temperature region within a predefined time period of the early prediction stage and the optimal temperature prediction model. According to the present application, the future temperature of the furnace temperature region can be predicted; not only can the future temperature of the furnace temperature region be predicted with greater accuracy, but also the temperature variation trends of multiple temperature regions can be predicted in advance, and the amount of computation in the process of predicting the future temperature is reduced.

[0076] The various embodiments in this specification are described progressively. References may be made among these embodiments with respect to identical or similar parts, and the description of each embodiment focuses primarily on the differences between that embodiment and the other embodiments. For the system embodiment, since the system embodiment is fundamentally similar to the process embodiment, the description of the system embodiment is relatively simple. For the correlation between the system embodiment and the process embodiment, reference may be made to the partial description of the process embodiment. It should be noted that various technical features of the embodiments described above may be combined in any combination.For the sake of brevity, not all possible combinations of the various technical features in the embodiments described above are described. However, as long as the combinations of these technical features do not... are not inconsistent, they must be considered as falling within the scope of this specification.

[0077] The above embodiments express only several preferred embodiments of the present application, and their descriptions are more specific and detailed, but cannot be construed as limiting the scope of the present application. It should be emphasized that, for a person skilled in the art, several improvements and substitutions can be made without departing from the technical principles of the present application, and these improvements and substitutions should also be considered as part of the scope of protection of this application. Therefore, the scope of protection of this patent application should be based on the scope of protection of the appended claims.

Claims

1. Demands Method for predicting the temperature of a furnace, comprising: acquiring furnace data for any temperature region of the furnace within a defined time range of the temperature region corresponding to any temperature region of the furnace, and forming a historical data set, in which the furnace data includes temperature data and power data; the acquisition of a first training set and a first test set according to the historical data set, and the training of a plurality of initial temperature prediction models selected in advance using the first training set, to obtain a plurality of first temperature prediction models; testing the plurality of first temperature prediction models using the first test set to output a plurality of first test set output errors, and separating a plurality of second temperature prediction models from the plurality of first temperature prediction models according to the plurality of first test set output errors and a predefined range of a first test error; training the plurality of second temperature prediction models using a second training set acquired in advance, to obtain a plurality of third temperature prediction models; testing the plurality of third temperature prediction models according to a second pre-acquired test set, to output a plurality of output errors from the second test set, and using a third temperature prediction model corresponding to a minimum value of the plurality of output errors from the second test set as the optimal temperature prediction model for any temperature region of the furnace; and the input of data from the furnace to be predicted from any temperature region of the furnace collected within a predefined time period of the early prediction stage into the optimal temperature prediction model corresponding to the temperature region

2. any of the oven, to obtain a predicted temperature in the oven temperature region within a predicted time period; in which the acquisition of the first training set and the first test set according to the historical dataset includes: the selection of furnace data for any temperature region of the furnace within a defined abnormal temperature range corresponding to that temperature region of the furnace according to the historical dataset, and the use of the furnace data as the historical dataset to be processed; performing preprocessing on the historical dataset to be processed to obtain a preprocessed historical dataset, wherein the preprocessing includes data cleaning; and the acquisition of the first training set and the first test set based on the preprocessed historical dataset; and in which each of the second training set and the second test set includes historical furnace data for any temperature region of the furnace in the predefined early prediction stage time period, in which the predefined early prediction stage time period includes a normal early prediction stage time period and an early prediction stage failure time period. A method for predicting the temperature of an oven according to claim 1, wherein the data cleaning process comprises: In the case where historical furnace abnormal jump data exists in the historical dataset to be processed, the historical furnace abnormal jump data is replaced by a median value of all historical furnace data in the historical dataset to be processed until the historical furnace abnormal jump data no longer exists in the historical dataset to be processed, where the historical furnace abnormal jump data is temperature data between which a difference value and adjacent historical data of the furnace exceed a predefined difference value.

3. Method for predicting the temperature of a furnace according to claim 1, wherein an initial temperature prediction model comprises a long- and short-term memory (LSTM) model, a recurrent neural network (RNN) model, a closed recurrent unit (GRU) model, a convolutional neural network (CNN) model and a graph neural network (GNN) model.

4. A method for predicting the temperature of a furnace according to claim 1, further comprising: acquiring a plurality of statistical modules according to the furnace data for any temperature region of the furnace, and obtaining a temperature distribution characteristic of any temperature region of the furnace according to the plurality of statistical modules; and obtaining a temperature variation trend of any temperature region of the furnace according to the temperature distribution characteristic.

5. Method for predicting the temperature of a furnace according to claim 4, wherein the plurality of statistical modules includes a mean, a variance, a deviation and a kurtosis.

6. A furnace temperature prediction system, comprising: a data acquisition module, which is configured to acquire furnace data for any temperature region of the furnace within a defined time range of the temperature region corresponding to any temperature region of the furnace, and to form a historical dataset; wherein the furnace data includes temperature data and power data; a first filtering module, which is configured to acquire a first training set and a first test set from the historical dataset, and to train a plurality of pre-selected initial temperature prediction models using the first training set, in order to obtain a plurality of first temperature prediction models;test the plurality of first temperature prediction models using the first test set in order to output a plurality of first test set output errors, and discard; a plurality of second temperature prediction models from the plurality of first temperature prediction models according to the plurality of output errors of the first test set and a predefined range of a first test error; a second filtering module, which is configured to train the plurality of second temperature prediction models using a second pre-acquired training set, in order to obtain a plurality of third temperature prediction models; test the plurality of third temperature prediction models according to a second pre-acquired test set, in order to output a plurality of output errors from the second test set, and use a third temperature prediction model corresponding to a minimum value of the plurality of output errors from the second test set as the optimal temperature prediction model for any temperature region of the furnace; and a temperature prediction module, which is configured to input the data from the furnace to be predicted from any temperature region of the furnace collected in a predefined early prediction stage time period into the optimal temperature prediction model corresponding to any temperature region of the furnace, to obtain a prediction temperature in the temperature region of the furnace in a prediction time period; the first filtering module, which is further configured to select furnace data for any temperature region of the furnace within a defined abnormal temperature range corresponding to any temperature region of the furnace according to the historical dataset, and to use the furnace data as the historical dataset to be processed; to perform preprocessing on the historical dataset to be processed to obtain a preprocessed historical dataset, in which the preprocessing includes data cleaning; and to acquire the first training set and the first test set according to the preprocessed historical dataset; and in which each of the second training set and the second test set includes historical furnace data for any temperature region of the furnace in the predefined early prediction stage time period, in which the predefined early prediction stage time period includes a normal early prediction stage time period and an early prediction stage failure time period.

7. A computer device, comprising a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the method of any one of claims 1 to 5.

Images