An online prediction method and system for metallurgical by-product gas production and consumption
Through the metallurgical by-product gas production consumption prediction method combined with deep learning and intelligent algorithms, the problems of low prediction accuracy and hysteresis in the existing technology are solved, and high-precision and low-hysteresis gas production consumption prediction are achieved, reducing the The gas discharge rate and improve the system's intelligent control capabilities.
Patent Information
- Application Number
- CN202210116792.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-02-07
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2042-02-07
AI Technical Summary
In the prior art, the prediction accuracy of metallurgical by-product gas production consumption is low, the hysteresis is large, and it cannot be predicted online, resulting in low gas utilization efficiency and high dispersion rate.
Deep learning algorithms and intelligent algorithms are used to establish a gas production consumption prediction model through equipment operation status parameters and production operation signals, obtain equipment feature curves and basic prediction curves, combine correlation weights and time intervals to predict, update feature state boundaries and Filtering is performed to improve prediction accuracy.
It realizes high-precision and low-hysteresis gas production consumption prediction, reduces gas discharge rate, improves operating efficiency and intelligent control capabilities of the gas system.
Smart Images

Figure CN114580712B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of coal gas prediction, and in particular to an online prediction method and system for the production and consumption of metallurgical by-product coal gas. Background Art
[0002] Blast furnace gas, coke oven gas and converter gas are by-product gases in the steel production process. Due to the complexity of the actual production process and the changeable operating conditions, it is extremely difficult to predict the gas production and consumption online, which makes it difficult to use the gas efficiently and leads to a high gas emission rate. Therefore, the online accurate prediction of by-product gas production and consumption is conducive to the rational use of by-product gas, reducing the gas emission rate, and effectively reducing the imbalance between gas supply and demand in the short term.
[0003] The existing coal gas production and consumption prediction technology has the defects of low prediction accuracy, large lag and inability to make online predictions due to the complex factors considered. Summary of the invention
[0004] 1. Technical issues to be resolved
[0005] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides an online prediction method and system for metallurgical by-product gas production and consumption, which solves the technical problems of low prediction accuracy, large hysteresis and inability to make online predictions.
[0006] (II) Technical solution
[0007] In order to achieve the above object, the main technical solutions adopted by the present invention include:
[0008] S10, for the gas production and consumption in the operating parameters of at least one group of equipment under each operating state collected in advance, perform data preprocessing on the gas production and consumption, and obtain a device characteristic curve under the operating state according to the preprocessed gas production and consumption under each operating state;
[0009] The operating parameters include process parameters and gas production and consumption;
[0010] S20. For each device in each group of devices, according to a pre-built deep learning algorithm, a correlation weight calculation is performed between the process parameters of each device and the gas production and consumption, and the correlation weight between the process parameters of each device and the gas production and consumption in each operating state and the reading time interval of each operating parameter are obtained;
[0011] S30, based on the gas production and consumption characteristics of each device and the pre-built intelligent algorithm, a prediction model for each device is established according to the correlation weight and the reading time interval, and the prediction model is used to predict the gas production and consumption of the device in each operating state;
[0012] S40, for each device, according to the prediction model and operating parameters of the device, obtaining a basic prediction curve of the device, wherein the basic prediction curve represents the relationship between the operating state of the device and the gas production and consumption;
[0013] S50. For each device, the basic prediction curve is combined with the corresponding device characteristic curve under the same operating state to obtain a prediction result of gas production and consumption under the operating state.
[0014] Optionally, before step S10, the method further includes:
[0015] Collecting operating parameters of each device in the at least one group of devices in each operating state, and storing them in a pre-built operating parameter database;
[0016] The gas production and consumption in the pre-collected operating parameters in step S10 is the gas production and consumption pre-stored in the operating parameter database.
[0017] Optionally, before step S40, a characteristic state space of each group of equipment is constructed to judge the operating state of the equipment in the prediction model;
[0018] The characteristic state space of each group of devices constructed is specifically:
[0019] For each device in a group of devices, determining the operating parameter range of each device in each operating state according to the operating parameters in the operating parameter database;
[0020] Based on the operating parameter ranges of each operating state, determining the characteristic state boundary of each device, and storing the boundary in a pre-built characteristic state library of each device;
[0021] The characteristic state libraries of each device are integrated to obtain the device characteristic state space of the group of devices.
[0022] Optionally, before step S40, the method further includes:
[0023] According to the real-time operating status and operating parameters of each device, the gas production and consumption prediction result of the prediction model and / or the operating status of the device are corrected, specifically:
[0024] S3401, pre-determine the operation status of each device in the prediction model within the time period t2 according to the pre-set equipment operation plan information, and assign an initial value to the gas production and consumption prediction result according to the pre-determined equipment operation status;
[0025] S3402, matching the equipment operation parameters of time period t1 in the operation parameter database with the characteristic state vector of the equipment in the equipment characteristic state space, judging the equipment operation state of time period t2 in the prediction model, and judging whether the operation state of the equipment in the prediction model of time period t2 is consistent with the real-time equipment operation state based on the equipment operation state of the equipment in real-time production at time t, if not, updating the equipment operation state in the prediction model of time period t2;
[0026] The time t is the connecting point of the time periods t1 and t2; the time period t1 comes first;
[0027] S3403. According to the parameter reading time interval, obtain the value of each operating parameter in real-time production and the change of the operating state; according to the obtained value and the change of the operating state, update the original characteristic state boundary and / or the operating state of the equipment in the prediction model.
[0028] Optionally, S3403 specifically includes:
[0029] Based on the acquired real-time production equipment operation status, if the operation status changes, it is compared with the operation status of the equipment in the prediction model, and if they are different, the equipment operation status in the prediction model is updated;
[0030] If the operating state has not changed, based on the acquired operating state of actual production, the value of the operating parameter acquired in real time under the operating state is compared with the characteristic state boundary under the same state of the same equipment;
[0031] If g consecutive values of the operating parameters exceed the characteristic state boundary, the characteristic state boundary of the device in the characteristic state space is updated.
[0032] Optionally, the S40 is specifically:
[0033] According to the updated operating state of the equipment in the time period t2, based on the prediction model, the operating parameters under the operating state are calculated to obtain a basic prediction curve of gas production and consumption in the time period t2.
[0034] Optionally, data preprocessing is performed on the gas production and consumption, specifically:
[0035] By means of a smoothing filter, the gas production and consumption data are filtered, invalid values in the gas production and consumption data are eliminated, and missing gas production and consumption data are filled.
[0036] Optionally, the filtering process specifically includes:
[0037] Divide the time period from ta to t according to a preset time interval, obtain the gas production and consumption data at n moments, calculate the specific values of the gas production and consumption at the n moments according to a pre-constructed formula, and obtain the filtered production and consumption data at moment t;
[0038] The pre-built formula is:
[0039]
[0040] The L is the length of the sliding window; y(τ) is the specific value of the gas production and consumption corresponding to the n moments obtained.
[0041] Optionally, the pre-built deep learning algorithm is an XGboost algorithm.
[0042] Optionally, a metallurgical by-product gas online prediction system comprises:
[0043] at least one processor;
[0044] a memory communicatively coupled to the processor;
[0045] The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute any of the above-mentioned methods for online prediction of metallurgical by-product gas production and consumption.
[0046] (III) Beneficial effects
[0047] The online prediction method and system for metallurgical by-product gas production and consumption proposed in the present invention establish a gas production and consumption prediction model according to equipment operating status parameters, production operation signals and equipment operating status, predict gas production and consumption, give full play to the advantages of equipment operation signals, simplify the prediction calculation amount, reduce prediction lag, and improve the work efficiency of operators.
[0048] The online prediction method and system of the present invention predicts the gas production and consumption by analyzing the correlation weights between each equipment parameter and the prediction results, and taking the production plan of the equipment as an important factor affecting the gas prediction results. It cleverly combines traditional processes with intelligent algorithms, simplifies the prediction calculation amount, and improves the prediction accuracy.
[0049] The online prediction method and system of the present invention combine the basic prediction curve and the characteristic curve by mathematical means to obtain the predicted value of gas in a future period of time, thereby improving the prediction accuracy, and enhancing the self-learning, self-adaptation and self-updating capabilities of the method by updating the device characteristic state boundary and sliding filtering processing.
[0050] The online prediction method and system of the present invention, due to low hysteresis, small calculation amount and high prediction accuracy, realize the rational use of coal gas in the whole process, reduce the emission rate of coal gas, and are conducive to the systematic intelligent control of coal gas production and consumption. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] Figure 1 A schematic flow chart of an online prediction method for metallurgical by-product gas production and consumption provided in one embodiment of the present invention. DETAILED DESCRIPTION
[0052] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below in conjunction with the accompanying drawings through specific implementation methods. It is understood that the specific embodiments described below are only used to explain the relevant inventions, rather than to limit the invention. It should also be noted that the embodiments and features in the embodiments of this application can be combined with each other in the absence of conflict; for ease of description, only the parts related to the invention are shown in the accompanying drawings.
[0053] The prediction accuracy of coal gas production and consumption is affected by many links, including the determination of equipment operating status, the processing of sample data for prediction, and the establishment of prediction models. The processing method of each link will affect the prediction accuracy of coal gas production and consumption. Therefore, in order to make a high-precision prediction of coal gas production and consumption, an online prediction method for metallurgical by-product coal gas production and consumption is provided in an embodiment of the present invention.
[0054] like Figure 1 As shown, Figure 1 An online prediction method for metallurgical by-product gas production and consumption provided by an embodiment of the present invention is shown, which specifically includes the following steps:
[0055] S10, for the gas production and consumption in the operating parameters of at least one group of equipment under each operating state collected in advance, perform data preprocessing on the gas production and consumption, and obtain a device characteristic curve under the operating state according to the preprocessed gas production and consumption under each operating state;
[0056] The operating parameters may include process parameters and gas production and consumption, etc. In some other embodiments, they may also include other parameter types, such as production operation signals, etc., which are not limited here.
[0057] S20. For each device in each group of devices, according to a pre-built deep learning algorithm, a correlation weight calculation is performed between the process parameters of each device and the gas production and consumption, and the correlation weight between the process parameters and the gas production and consumption in each state of each device and the reading time interval of each operating parameter are obtained;
[0058] In this embodiment, the pre-built deep learning algorithm used is the XGboost algorithm, but in other embodiments, other algorithms that can achieve the same purpose may also be used, which is not limited here.
[0059] S30, based on the gas production and consumption characteristics of each device and the pre-built intelligent algorithm, a prediction model for each device is established according to the correlation weight and the reading time interval, and the prediction model is used to predict the gas production and consumption of the device in each operating state;
[0060] In this embodiment, the input conditions of the prediction model are the pretreated gas production and consumption data at time t and in the time period t1 before time t, process parameters and other required parameters, such as environmental parameters (ambient temperature, etc.). The specific types of parameters are mainly determined by the algorithm used and are not limited here.
[0061] The time t is any time during the operation of the equipment, which can be selected according to actual needs and is not limited here.
[0062] S40, for each device, according to the prediction model and operating parameters of the device, obtaining a basic prediction curve of the device, wherein the basic prediction curve represents the relationship between the operating state of the device and the gas production and consumption;
[0063] S50. For each device, the basic prediction curve is combined with the corresponding device characteristic curve under the same operating state to obtain a prediction result of gas production and consumption under the operating state.
[0064] The online prediction method for metallurgical by-product gas production and consumption provided in this embodiment combines the pre-collected operating parameters corresponding to the equipment in each operating state to obtain the equipment characteristic curve of the equipment in the smelting process, combines the deep learning algorithm to obtain the correlation weights between the process parameters and the gas production and consumption, obtains a reasonable reading time interval for reading the operating parameters, and then establishes a prediction model based on the correlation weights between the operating parameters and the gas production and consumption and the reading time interval of the parameters. By fitting the basic prediction curve of the prediction model and the equipment characteristic curve to obtain the predicted value of the gas production and consumption, the result of the gas production and consumption can be predicted, thereby improving the accuracy of predicting the gas production and consumption and reducing the hysteresis.
[0065] In some other embodiments, before step S10, the following steps may also be included:
[0066] The operating parameters of each device in the at least one group of devices in each operating state are collected and stored in a pre-built operating parameter database. In some other embodiments, the operating parameter database may also be referred to as a historical database.
[0067] In the step S10, the gas production and consumption in the pre-collected operating parameters may be the gas production and consumption pre-stored in the operating parameter database.
[0068] In actual application, the operating parameters collected by the group of equipment during real-time production can also be stored in the operating parameter database to increase the amount and breadth of database data, which is conducive to improving the prediction accuracy of gas production and consumption.
[0069] Furthermore, in actual applications, due to the uncontrollability of the actual production process, when the prediction model predicts the gas production and consumption, there is a difference between the operating status in the prediction model and the operating status of the equipment in real-time production, resulting in a deviation in the final gas production and consumption prediction result. Therefore, in actual applications, in order to ensure the accuracy of the prediction of gas production and consumption, before step S40, the gas production and consumption prediction results of the prediction model and / or the operating status of the equipment can be corrected according to the real-time production equipment and operating status and operating parameters to improve the prediction accuracy of the prediction model for gas production and consumption.
[0070] Prior to this, the characteristic state space of each group of equipment can also be constructed to judge the equipment operation status in the prediction model;
[0071] The characteristic state space of each group of devices constructed is specifically:
[0072] For each device in a group of devices, determining the operating parameter range of each device in each operating state according to the operating parameters in the operating parameter database;
[0073] Based on the operating parameter ranges of each operating state, determining the characteristic state boundary of each device, and storing the boundary in a pre-built characteristic state library of each device;
[0074] The characteristic state libraries of each device are integrated to obtain the device characteristic state space of the group of devices.
[0075] For example, in this embodiment, the upper and lower boundaries of the gas production and consumption of each device are determined according to the pre-acquired gas production and consumption value range and the different operating states of each device as the characteristic state boundaries of the device in the operating state.
[0076] Specifically, the prediction results of the gas production and consumption of the prediction model and / or the operating status of the equipment are corrected, including:
[0077] S3401, pre-determine the operation status of each device in the time period t2 according to the pre-set equipment operation plan information, and assign an initial value to the gas production and consumption prediction result according to the pre-determined equipment operation status;
[0078] In the actual production process, due to the production process mechanism, the gas production and consumption under some operating conditions can be directly confirmed. For example, if the operating state is maintenance state, the gas production and consumption must be 0, which does not need to be predicted by the prediction model. Therefore, the time period without gas production and consumption can be determined according to the set equipment operation plan information, and the gas production and consumption in this time period can be directly initialized to 0, which improves the data accuracy and reduces the lag of the prediction.
[0079] S3402, matching the equipment operation parameters of time period t1 in the operation parameter database with the characteristic state vector of the equipment in the equipment characteristic state space, judging the equipment operation state of time period t2 in the prediction model, and judging whether the operation state of the equipment in the prediction model of time period t2 is consistent with the real-time equipment operation state based on the equipment operation state of the equipment in real-time production at time t, if not, updating the equipment operation state in the prediction model of time period t2;
[0080] The time t is the connection point between the time periods t1 and t2; the time period t1 comes first; the time t is any time during the production process of the equipment, and the specific time of the time t is determined based on actual needs and is not limited here.
[0081] S3403. Obtain the value of each operating parameter of each device in real-time production and the change of the operating status according to the parameter reading time interval; update the original feature state boundary and / or update the operating status of the device in the prediction model according to the obtained value and the change of the operating status.
[0082] Furthermore, S3403 specifically includes:
[0083] Based on the acquired real-time production equipment operation status, if the operation status changes, it is compared with the operation status of the equipment in the prediction model, and if they are different, the equipment operation status in the prediction model is updated;
[0084] If the operating state has not changed, based on the acquired real-time production operating state, the value of the operating parameter acquired in real time under the operating state is compared with the characteristic state boundary under the same state of the same device.
[0085] Specifically, in this embodiment, when the values of g consecutive operating parameters obtained exceed the characteristic state boundary originally determined, and the operating state obtained in real time does not change and is consistent with the operating state of the device in the prediction model, the characteristic state boundary of the device in the characteristic state space is updated. If the operating state changes, the operating state of the device in the prediction model is corrected, and the above comparison is performed based on the operating parameter values corresponding to the changed operating state.
[0086] In step S40, for each device, a basic prediction curve of the device is obtained according to the prediction model and operating parameters of the device, and the basic prediction curve represents the relationship between the operating state of the device and the gas production and consumption. Specifically, according to the operating state of the device in the time period t2, based on the prediction model, the operating parameters under the operating state are calculated to obtain the basic prediction curve of the gas production and consumption in the time period t2.
[0087] Of course, after the operating status of the equipment in the prediction model is corrected and updated, the basic prediction curve of gas production and consumption should be obtained according to the operating status of the equipment in the time period t2 after the update.
[0088] In practical applications, since the data acquisition system may have errors when collecting data, the directly obtained gas production and consumption data is not accurate enough. Therefore, it is also necessary to perform data preprocessing on the obtained gas production and consumption data, which specifically includes:
[0089] The gas production and consumption data is filtered to eliminate invalid values in the gas production and consumption data, and to fill in missing gas production and consumption data to make the gas production and consumption data more accurate. Of course, the data acquisition system here only refers to the equipment or unit used to collect data, and can also be all other means of collecting data, which is not limited here.
[0090] In this embodiment, the filtering process is to filter the gas production and consumption data by means of a smoothing filter according to the time series characteristics of the gas production and consumption data;
[0091] Divide the time period from ta to t according to a preset time interval, obtain the gas production and consumption data at n moments, calculate the specific values of the gas production and consumption at the n moments according to a pre-constructed formula, and obtain the filtered production and consumption data at moment t;
[0092] The pre-built formula is:
[0093]
[0094] Wherein, L is the length of the sliding window; y(τ) is the specific value of the gas production and consumption corresponding to the obtained n moments.
[0095] In practical applications, the number of n is the length L of the sliding window.
[0096] The online prediction method for metallurgical by-product gas production and consumption provided by the embodiment of the present invention establishes equipment characteristic state space, combines production operation signals to determine the operating state of the equipment in the prediction model, pre-processes the operating parameter sample data for collection by filtering and other means, establishes a prediction model by analyzing the equipment operating parameters and the gas production and consumption characteristics of the equipment, improves the prediction accuracy by fitting characteristic curves and combining them with production plans, and achieves the purpose of reducing the lag of the prediction.
[0097] In order to better demonstrate the online prediction method for gas production and consumption proposed by the present invention, a more detailed description will be given below in conjunction with a specific embodiment.
[0098] In a metallurgical production line, the names of various equipment and their corresponding multiple operating states include:
[0099] Sintering machine, the operating status includes normal operation, reduced production, shutdown and maintenance status;
[0100] Vertical furnace, the operating status includes forward operation, reduced production, shutdown and maintenance status;
[0101] Coke oven, operating status includes initial coking stage, middle coking stage, final coking stage, shutdown and maintenance status;
[0102] Blast furnace, the operating status includes forward operation, wind stop, wind resumption, shutdown and maintenance status;
[0103] Hot blast furnace, operating status includes combustion period, air supply period, shutdown and maintenance status;
[0104] Converter, operating status includes payback period, non-payback period, shutdown and maintenance status;
[0105] Rotary kiln, the operating status includes forward operation, reduced production, shutdown and maintenance status;
[0106] The operating states of the heating furnace include normal operation, reduced production, waiting for material, shutdown and maintenance.
[0107] Of course, the above-mentioned devices and operating states are only for this embodiment. In some other embodiments, there are other devices and operating states. In actual applications, the devices and operating states should be determined according to actual needs, and no restrictions are made here.
[0108] In this embodiment, the data acquisition system monitors in advance and collects the operating parameters of each equipment in each operating state during the metallurgical process at a time interval b, stores the operating parameters collected through multiple monitoring in a pre-established operating parameter database, determines the operating parameter range and equipment characteristic state boundary of each equipment in each operating state, builds an equipment characteristic state library for each equipment, integrates the equipment characteristic state libraries of each equipment, and obtains the equipment characteristic state space.
[0109] Of course, the advance monitoring and collection of operating parameters of each device in each operating state during the metallurgical process may also include historical data that has been made public in other similar production processes.
[0110] Filtering the gas production and consumption data in the operating parameters, dividing the time period from ta to t according to a preset time interval b, obtaining the gas production and consumption data at the time corresponding to each time interval, taking the gas production and consumption data at the time corresponding to n time intervals as the basic data for sliding filtering and taking the average value according to a pre-constructed formula, and obtaining the filtered production and consumption data at time t;
[0111] The obtained gas production and consumption data at n moments are specifically data at moments tb, t-2b, ..., t-nb.
[0112] In this embodiment, the preferred range of a is 5 min to 24 h, and specifically selected here: a is 5 min; the preferred range of b is 1 s to 5 min, and preferably 10 s; n is preferably 3, that is, 5 minutes before time t are divided into time intervals of 10 s, and the gas production and consumption data corresponding to t-10s, t-20s, and t-30s are selected and substituted into the formula as the basic data for sliding filtering and the average value is taken to obtain the production and consumption data after filtering at time t, and so on, to achieve the effect of sliding filtering and realize the preprocessing of the gas production and consumption data.
[0113] Of course, in practical applications, the above-mentioned values a, b, t, n, etc. do not have a fixed required range and should be divided and confirmed according to actual needs, and are not restricted here.
[0114] According to the pre-processed gas production and consumption data, the trend characteristics of the gas production and consumption data of each device in each operating state are determined, and the gas production and consumption curve with this trend characteristic in each operating state is the device characteristic curve in this operating state.
[0115] In this embodiment, for step S20, the XGboost deep learning algorithm is used to calculate the correlation weight X between the process parameters of each device and the gas production and consumption, and obtain the reading time interval of each operating parameter;
[0116] Of course, in some other embodiments, other deep algorithms that can achieve the same performance purpose are also used, such as a grey correlation analysis algorithm, a principal component analysis algorithm, etc., which are not limited here.
[0117] Specifically, if X≥A, then this type of parameter is the main factor affecting the prediction result, and the reading time interval of this type of parameter used for prediction model input is b. In this embodiment, the time interval b used is the time interval used when pre-collecting the operating parameter data.
[0118] If B≤X<A, then such parameters are minor factors affecting the prediction results, and the reading time interval of such parameters used for prediction model input is c.
[0119] In this embodiment, the time when the operating parameter data is not read is acquired using the principle of adjacent values, which is the same as the data at the previous time of the collection point.
[0120] If X<B, then such parameters are negligible factors affecting the prediction results and are not used as inputs to the prediction model.
[0121] Specifically, in this embodiment, A, B, b, and c are all pre-set values, and B <A, b <c. Preferably, A = 0.5, B = 0.3, b = 10s, and c = 1min. In some other applications, the values corresponding to A, B, b, and c should be divided according to actual needs and are not limited here.
[0122] A prediction model is established for step S30. In this embodiment, it specifically includes: equipment whose gas generation or consumption has periodic characteristics, for example, converters, hot blast furnaces, etc. are modeled using the long short-term memory network method; equipment whose gas generation or consumption does not have periodic characteristics, such as sintering machines, vertical furnaces, coke ovens, rotary kilns, heating furnaces, etc. are modeled using the differential autoregressive moving average method to obtain the corresponding prediction model.
[0123] Of course, in some other embodiments, other methods that can achieve the above-mentioned modeling are also used, such as BP neural network method, convolutional neural network method, etc., which are not limited here.
[0124] After obtaining the prediction model, any prediction model can be used to obtain the basic prediction curve of the gas production and consumption of the corresponding equipment in any time period. By fitting the basic prediction curve of a certain equipment in a time period and the equipment characteristic curve of the equipment in the corresponding time period, the gas production and consumption prediction result of the equipment in that time period can be obtained.
[0125] In this embodiment, the corrections made to the prediction model specifically include:
[0126] According to the pre-established production plan (monthly plan, ten-day plan, daily plan, hourly plan, etc.), the equipment operating status at time t and after time t is predetermined, and the gas production and consumption within the planned time period is assigned according to the equipment status.
[0127] The assigning of values to the gas production and consumption within the planned time period specifically includes:
[0128] For equipment that is in a special state during the time period from t+d to t+e, the gas production and consumption data during this time period are directly assigned values according to the process characteristics, and there is no need to predict the gas production and consumption during this time period through a prediction model.
[0129] Specifically, for equipment that is in shutdown, maintenance, etc. during the time period from t+d to t+e, and equipment that is in a special stage of production, the gas production and consumption data during this time period is directly assigned a value of 0. The special stage includes blast furnace wind stop period, hot blast furnace air supply period, converter non-recovery period, etc.
[0130] In practical applications, the operating status of a device corresponds one-to-one to the production operation signal during production, so the current operating status of the device can be determined by the production operation signal.
[0131] The gas production and consumption of each device collected in real time during the time period from ta to t is automatically matched with the characteristic state vector of each device in the equipment characteristic state space, and the equipment characteristic state is combined with the equipment production operation signal to determine whether it is necessary to update the operating status of the equipment in the model at time t and a period of time thereafter.
[0132] In some cases, the equipment operation status cannot be judged based solely on gas production and consumption. This is because the same gas production and consumption range may correspond to multiple possible operation states. For example, if the gas production and consumption value is 0 during a certain period of time, it may be that the equipment is in a shutdown state or maintenance state or in a special stage of production. Therefore, combining the production operation signal of the equipment at time t can accurately determine whether it is necessary to update the operation status of the equipment at time t and a period of time thereafter.
[0133] The production plan can only be used to determine the operating status of the equipment in part of the time period (t+d to t+e) within the predicted time period from t to t+f, and cannot determine the operating status of the equipment at all time points within the time period from t to t+f. Based on the production plan judgment and combined with the real-time production operation signal, the operating status of the equipment at time t and a period of time thereafter can be directly updated.
[0134] Where f is a preset value, and d <e<f。
[0135] Determine the parameter range of each device in each state obtained by real-time acquisition. If the g operating parameter values continuously acquired in a certain operating state of a certain device exceed the predetermined characteristic state boundary and the operation signal does not change, then update the original characteristic state boundary of the device in the characteristic state space. If the g operating parameter values continuously acquired in a certain operating state of a certain device exceed the predetermined characteristic state boundary and the operation signal changes, automatically match the characteristic state boundary of other operating states of the device and correct the operating state of the device. Regardless of whether the operating parameter value exceeds the predetermined characteristic state boundary, as long as the operation signal at time t changes, the operating state of the device in the prediction model is corrected in time.
[0136] t, d, e, f and g are all pre-set values, among which f is preferably in the range of 30 min to 24 h, and 1 h is selected in this embodiment; g is preferably in the range of 1 to 5, and 3 is preferably selected in this embodiment; t is any moment, d and e are less than f, that is, d and e<1 h, but in some other embodiments, t, d, e, f, g, etc. have no fixed range, and the specific values should be confirmed based on actual conditions, such as collection frequency, predicted business needs and server computing power, and are not limited here.
[0137] The prediction model obtains the prediction result of gas production and consumption in the time period, i.e., the basic prediction curve. According to the operating state of the equipment in the time period, the characteristic curve of the equipment is obtained. The basic prediction curve and the characteristic curve are combined by mathematical means (including but not limited to nonlinear equations) to finally obtain the predicted value of the gas volume in the time period from t to t+f.
[0138] In this implementation, the mathematical method selected is to take the average value, and take the average value of the specific values of gas production and consumption at the same time in the equipment characteristic curve and the basic prediction curve to obtain the final prediction result.
[0139] Furthermore, in practical applications, there is also an online prediction system for metallurgical by-product gas production and consumption, including:
[0140] at least one processor;
[0141] a memory communicatively coupled to the processor;
[0142] The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute any of the above-mentioned methods for online prediction of metallurgical by-product gas production and consumption.
[0143] The online prediction method for metallurgical by-product gas proposed in the present invention avoids the large lag caused by considering too many factors in the prior art, and also avoids the low prediction accuracy caused by considering too few factors. Through the equipment characteristic state space and production operation signals, the equipment operation state is determined, and then a gas production and consumption prediction model is established to predict the gas production and consumption. This can give full play to the advantages of equipment operation signals, reduce prediction lag, and improve the work efficiency of operators.
[0144] The present invention proposes an online prediction method and system for metallurgical by-product gas, which analyzes the correlation weights between various equipment parameters and the prediction results, and takes the production plan of the equipment as an important factor affecting the gas prediction results, thereby predicting the gas production and consumption, cleverly combining traditional processes with intelligent algorithms, simplifying the prediction calculation amount, and improving the prediction accuracy.
[0145] The online prediction method and system for metallurgical by-product gas proposed in the present invention combine the basic prediction curve with the characteristic curve by mathematical means to obtain the predicted value of gas in a certain period of time in the future, thereby improving the prediction accuracy, and enhancing the self-learning, self-adaptation and self-updating capabilities of the method by updating the equipment characteristic state boundary and sliding filtering processing.
[0146] The online prediction method and system for metallurgical by-product gas proposed in the present invention have low hysteresis, small calculation amount and high prediction accuracy, thus realizing the rational use of gas in the whole process, reducing the emission rate of gas, and promoting the development of intelligent management and control of gas system.
[0147] The online prediction method and system for metallurgical by-product gas proposed in the present invention can utilize enterprise production plans, integrate operation information, and mine historical data, thereby improving prediction accuracy and reducing the lag of calculation results, and is suitable for online prediction of gas systems.
[0148] Since the system described in the above embodiment of the present invention is a system used to implement the method of the above embodiment of the present invention, a person skilled in the art can understand the specific structure and deformation of the system based on the method described in the above embodiment of the present invention, and thus will not be described in detail here. All systems / devices used in the method of the above embodiment of the present invention belong to the scope of protection of the present invention.
[0149] The present invention is described with reference to the flowchart of the method, system and computer program product according to the embodiment of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, and the combination of the processes and / or boxes in the flowchart and / or block diagram can be implemented by computer program instructions.
[0150] In addition, it should be noted that, in the description of this specification, the description of the terms "one embodiment", "some embodiments", "embodiment", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.
[0151] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments after knowing the basic creative concept. Therefore, the claims should be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present invention.
[0152] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention should also include these modifications and variations.
Claims
1. An online prediction method for metallurgical by-product gas production and consumption, characterized in that: The following steps are involved: S10, for the gas production and consumption in the operating parameters of at least one group of equipment under each operating state collected in advance, perform data preprocessing on the gas production and consumption, and obtain a device characteristic curve under the operating state according to the preprocessed gas production and consumption under each operating state; The operating parameters include process parameters and gas production and consumption; S20. For each device in each group of devices, according to a pre-built deep learning algorithm, a correlation weight calculation is performed between the process parameters of each device and the gas production and consumption, and the correlation weight between the process parameters of each device and the gas production and consumption in each operating state and the reading time interval of each operating parameter are obtained; S30, based on the gas production and consumption characteristics of each device and the pre-built intelligent algorithm, a prediction model for each device is established according to the correlation weight and the reading time interval, and the prediction model is used to predict the gas production and consumption of the device in each operating state; S40, for each device, according to the prediction model and operating parameters of the device, obtaining a basic prediction curve of the device, wherein the basic prediction curve represents the relationship between the operating state of the device and the gas production and consumption; S50, for each device, combining and calculating the basic prediction curve with the corresponding device characteristic curve under the same operating state to obtain a prediction result of gas production and consumption under the operating state; Before step S40, a characteristic state space of each group of equipment is constructed to judge the operating state of the equipment in the prediction model; The characteristic state space of each group of devices constructed is specifically: For each device in a group of devices, determining the operating parameter range of each device in each operating state according to the operating parameters in the operating parameter database; Based on the operating parameter ranges of each operating state, determining the characteristic state boundary of each device, and storing the boundary in a pre-built characteristic state library of each device; Integrate the characteristic state library of each device to obtain the device characteristic state space of the group of devices; Before step S40, the method further includes: According to the real-time operating status and operating parameters of each device, the gas production and consumption prediction result and / or the operating status of the device in the prediction model are corrected, specifically: S3401, pre-determine the operation status of each device in the prediction model within the time period t2 according to the pre-set equipment operation plan information, and assign an initial value to the gas production and consumption prediction result according to the pre-determined equipment operation status; S3402, matching the equipment operation parameters of time period t1 in the operation parameter database with the characteristic state vector of the equipment in the equipment characteristic state space, judging the equipment operation state of time period t2 in the prediction model, and judging whether the operation state of the equipment in the prediction model of time period t2 is consistent with the real-time equipment operation state based on the equipment operation state of the equipment in real-time production at time t, if not, updating the equipment operation state in the prediction model of time period t2; The time t is the connecting point of the time periods t1 and t2; the time period t1 comes first; S3403. According to the parameter reading time interval, obtain the value of each operating parameter in real-time production and the change of the operating state; according to the obtained value and the change of the operating state, update the original characteristic state boundary and / or the operating state of the equipment in the prediction model.
2. The online prediction method according to claim 1, characterized in that: Before step S10, the method further includes: Collecting the operating parameters of each device in the at least one group of devices in each operating state, and storing them in a pre-built operating parameter database; The gas production and consumption in the pre-collected operating parameters in step S10 is the gas production and consumption pre-stored in the operating parameter database.
3. The online prediction method according to claim 1, characterized in that: S3403 specifically includes: Based on the acquired real-time production equipment operation status, if the operation status changes, it is compared with the operation status of the equipment in the prediction model, and if they are different, the equipment operation status in the prediction model is updated; If the operating state has not changed, based on the acquired operating state of actual production, the value of the operating parameter acquired in real time under the operating state is compared with the characteristic state boundary under the same state of the same equipment; If g consecutive values of the operating parameters exceed the characteristic state boundary, the characteristic state boundary of the device in the characteristic state space is updated.
4. The online prediction method according to claim 3, characterized in that: The S40 is specifically: According to the updated operating state of the equipment in the time period t2, based on the prediction model, the operating parameters under the operating state are calculated to obtain a basic prediction curve of gas production and consumption in the time period t2.
5. The online prediction method according to claim 1, characterized in that: The gas production and consumption data is preprocessed, specifically: By means of a smoothing filter, the gas production and consumption data are filtered, invalid values in the gas production and consumption data are eliminated, and missing gas production and consumption data are filled.
6. The online prediction method according to claim 5, characterized in that: The filtering process specifically includes: According to the pre-set time interval The gas production and consumption data at n moments are obtained by dividing the time period into n to t moments, and the specific values of the gas production and consumption at the n moments are calculated according to a pre-constructed formula to obtain the filtered production and consumption data at moment t; The pre-built formula is: ; The L is the length of the sliding window; is the specific value of the gas production and consumption corresponding to the n moments obtained.
7. The online prediction method according to claim 1, characterized in that: The pre-built deep learning algorithm is the XGboost algorithm.
8. An online prediction system for metallurgical by-product gas production and consumption, characterized in that: include: at least one processor; a memory communicatively coupled to the processor; The memory stores program instructions executable by the processor, and the processor calls the program instructions to execute the online prediction method for metallurgical by-product gas as described in any one of claims 1-7.
