Multistage demand control method for fused magnesia smelting process

Abstract

The invention relates to the field of fused magnesite, and discloses a fused magnesite smelting process multistage demand control method, which comprises the following steps: S1, raw material treatment and mixing, S2, electric furnace charging and heating treatment, and starting of an electric furnace to gradually heat the raw materials to a temperature above a melting point, and S3, fused magnesite processing parameter control. The method comprises the following steps: S1, monitoring real-time pressure and chemical components in an electric furnace, S4, performing electric power multi-stage demand prediction control, and predicting a subsequent power utilization state, and S5, performing self-updating of a prediction control algorithm, and establishing a prediction algorithm self-optimization system. According to the electric power multi-stage prediction control, the future electric power demand can be accurately predicted according to historical data and real-time environment parameters, so that the electric power supply is adjusted in advance, the influence of electric power fluctuation on the fused magnesia smelting process is avoided, the stability and efficiency of the smelting process are ensured, and the production cost is reduced. And judging whether the electric furnace can be opened according to the operation state of the electric furnace.

Description

Technical Field

[0001] The invention relates to the technical field of fused magnesia, in particular to a multi-stage demand control method for a fused magnesia smelting process. Background Art

[0002] Fused magnesia is a material for industrial use, usually refers to fused magnesium oxide (MgO). It is prepared by arc melting method, has high purity and good chemical stability, and is often used in refractory materials, furnace lining materials in iron and steel making industries, and electric furnace lining applications. Fused magnesia is widely used in various high-temperature industrial environments due to its high fire resistance, high temperature resistance, and chemical corrosion resistance.

[0003] In the prior art, fused magnesia is usually smelted by electric arc smelting, which requires precise control of the demand for electric energy. The current and voltage of the electric furnace can be adjusted through multi-level demand control to achieve the electric energy input required for smelting, ensuring the arc stability inside the furnace, high smelting efficiency and optimal energy utilization.

[0004] However, in actual use, since multiple groups of electric furnaces will occupy more electricity inside the factory when they are working, if new electric furnaces are started at this time, the voltage inside the factory will fluctuate, thereby affecting the electric furnace that is processing fused magnesia and making its internal arc unable to provide a stable high-temperature environment for fused magnesia production, thereby affecting the smelting quality and production efficiency of fused magnesia. In addition, unstable power supply may also cause damage to equipment and increase maintenance costs. In view of this, we propose a multi-level demand control method for the fused magnesia smelting process. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a multi-stage demand control method for electric molten magnesia smelting process, which solves the problem that the prediction timeliness is short in the multi-stage demand control process of electric energy in the prior art, and it is impossible to effectively determine whether a new group of electric furnaces can be started.

[0006] To achieve the above objectives, the present invention is implemented by the following technical scheme: a multi-stage demand control method for fused magnesia smelting process, comprising the following steps:

[0007] S1: Raw material processing and mixing

[0008] Prepare the raw materials of the required components and weights in the fused magnesia processing process, and pre-treat the raw materials, including drying and pre-melting;

[0009] S2: Electric furnace charging and heating

[0010] Put the processed fused magnesia raw materials into an electric furnace, start the electric furnace to gradually heat the raw materials to above the melting point, and monitor the temperature of various places in the electric furnace through remote infrared monitoring equipment;

[0011] S3: Fused magnesia processing parameter control

[0012] Monitor the internal conditions of the electric furnace, including real-time pressure and chemical composition, and adjust the furnace parameters, including temperature, pressure, current, and voltage, through algorithm control equipment;

[0013] S4: Electricity multi-level demand forecasting control

[0014] Set the power warning range and issue warning alarms according to the warning range. At the same time, predict the subsequent power consumption status based on historical power consumption data as well as time, environment, and processing conditions;

[0015] S5: Predictive control algorithm self-update

[0016] A self-optimization system for the prediction algorithm is established, and the algorithm is self-optimized and updated based on the difference between the predicted power consumption status and the current actual power consumption status.

[0017] Preferably, the S1 raw material processing and mixing are mainly used to classify and proportion the raw materials. The raw material processing is not limited to drying and pre-melting, but also includes screening, crushing and grading of the raw materials.

[0018] Preferably, in S2, the raw materials placed inside are mainly heated by an electric furnace, and during the heating process, the temperature of each characteristic point area inside the electric furnace is detected by an infrared temperature sensing device, and a corresponding temperature change curve is generated according to the detection results.

[0019] Preferably, the characteristic point information in the temperature detection process mainly includes different depth areas inside the electric furnace, the end of the electric furnace close to the heat source, and the end of the electric furnace far from the heat source.

[0020] Preferably, in the fused magnesia processing parameter setting in S3, it is necessary to establish a temperature-time change curve and a pressure-time change curve that change with time according to the temperature and pressure required in the production process, and at the same time compare the multiple groups of actual temperature change curves collected in S2 with the reference temperature-time change curve to determine whether there is an uneven heating area inside the current equipment, and adjust the heating power by controlling the heating component.

[0021] Preferably, in the multi-level electricity demand forecasting control in S4, a grading mechanism will be established according to the electricity consumption situation, including primary control, secondary control and tertiary control, and different interval values ​​will be allocated according to the divided electricity consumption levels, and 15 minutes will be a group of demand cycles. At the same time, when the current electricity consumption status tends to be saturated, the start-up of the standby electric furnace will be stopped.

[0022] Preferably, in the multi-level electricity demand forecasting control in S4, a mathematical model will be established according to the current electricity consumption status, and historical data will be used as the basic data of the model forecasting control algorithm to predict the future electricity consumption status. At the same time, the forecasting algorithm will be supplementarily weighted according to the current regional temperature, air pressure and time data.

[0023] Preferably, in the multi-level electricity demand forecasting control in S4, the occasional electricity peak data that may occur in the actual electricity consumption process will be recorded, and a corresponding calculation allowable interval value will be established, that is, the allowable interval of electricity fluctuations within the interval range. When the electricity demand reaches the warning level but does not exceed the allowable fluctuation interval, the abnormal status alarm will not be triggered.

[0024] Preferably, the prediction algorithm update in S5 is mainly used to self-update and optimize the power multi-level demand state prediction algorithm established in S4. At the same time, during the self-update optimization parameter generation process, manual modification of specific parameters is supported.

[0025] Preferably, the prediction algorithm data and historical usage data updated in S5 will be synchronously stored in the data storage system. When the updated prediction algorithm has a large error and cannot be self-updated and optimized, the system will call the historical data version for data backtracking.

[0026] The present invention provides a multi-stage demand control method for fused magnesia smelting process. It has the following beneficial effects:

[0027] 1. The present invention can accurately predict future power demand based on historical data and real-time environmental parameters through the multi-level power prediction control, so as to adjust the power supply in advance, avoid the influence of power fluctuations on the molten magnesia smelting process, ensure the stability and efficiency of the smelting process, and determine whether the electric furnace can be started according to the operating status of the electric furnace.

[0028] 2. The present invention uses a self-updating algorithm to monitor the subsequent electricity demand algorithm of the electric furnace in real time, and continuously optimizes the prediction algorithm according to the actual operation of the electric furnace and historical data, thereby improving the accuracy of the prediction and further ensuring the stability and efficiency of the fused magnesia smelting process. At the same time, this self-updating mechanism also makes the entire control system more intelligent and can adaptively cope with various complex production environments.

[0029] 3. The present invention adopts remote infrared monitoring equipment to monitor the internal temperature of the electric furnace in real time, and adjusts the heating power in combination with the temperature change curve, which effectively solves the problem of uneven heating inside the electric furnace and further improves the smelting quality of fused magnesia. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of the multi-stage demand control method for the fused magnesia smelting process;

[0031] Figure 2 It is a time-power consumption line graph when the electric furnace is in operation in the prior art. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] Please see attached Figure 1 -Attached Figure 2 The embodiment of the present invention provides a multi-stage demand control method for a fused magnesia smelting process, comprising the following steps:

[0034] S1: Raw material processing and mixing

[0035] Prepare the raw materials of the required components and weights in the fused magnesia processing process, and pre-treat the raw materials, including drying and pre-melting;

[0036] S2: Electric furnace charging and heating

[0037] Put the processed fused magnesia raw materials into an electric furnace, start the electric furnace to gradually heat the raw materials to above the melting point, and monitor the temperature of various places in the electric furnace through remote infrared monitoring equipment;

[0038] S3: Fused magnesia processing parameter control

[0039] Monitor the internal conditions of the electric furnace, including real-time pressure and chemical composition, and adjust the furnace parameters, including temperature, pressure, current, and voltage, through algorithm control equipment;

[0040] S4: Electricity multi-level demand forecasting control

[0041] Set the power warning range and issue warning alarms according to the warning range. At the same time, predict the subsequent power consumption status based on historical power consumption data as well as time, environment, and processing conditions;

[0042] S5: Predictive control algorithm self-update

[0043] A self-optimization system for the prediction algorithm is established, and the algorithm is self-optimized and updated based on the difference between the predicted power consumption status and the current actual power consumption status.

[0044] The S1 raw material processing and mixing are mainly used to classify and proportion the raw materials. The raw material processing is not limited to drying and pre-melting, but also includes screening, crushing and grading of the raw materials.

[0045] In S2, the raw materials placed inside are mainly heated by an electric furnace. During the heating process, the temperature of each characteristic point area inside the electric furnace is detected by an infrared temperature sensing device, and a corresponding temperature change curve is generated according to the detection results.

[0046] The characteristic point information in the temperature detection process mainly includes different depth areas inside the electric furnace, the end of the electric furnace close to the heat source, and the end of the electric furnace far from the heat source. The collection of characteristic point information helps to more comprehensively understand the temperature distribution inside the electric furnace, thereby more accurately judging the heating state of the electric furnace, and providing strong support for subsequent multi-level power demand forecasting and control.

[0047] In the fused magnesia processing parameter setting in S3, it is necessary to establish a temperature-time change curve and a pressure-time change curve that change with time according to the temperature and pressure required in the production process. At the same time, the multiple groups of actual temperature change curves collected in S2 are compared with the reference temperature-time change curve to determine whether there is an uneven heating area inside the current equipment. If there is an uneven heating area, it is necessary to optimize the heating process by adjusting the heating parameters of the electric furnace, such as current, voltage or heating time, or to feed back abnormal signals to the control terminal for maintenance personnel to inspect and repair the equipment.

[0048] In the multi-level power demand forecasting control in S4, a grading mechanism will be established according to the power consumption situation, including normal control, early warning control and abnormal control, and different interval values ​​will be allocated according to the divided power consumption levels, with 15 minutes as a group of demand cycles. At the same time, when the current power consumption status tends to be saturated, the start-up of the standby electric furnace will be stopped.

[0049] The electric energy required from the start of the electric furnace to the completion of the processing is determined by Figure 2It can be seen that after the electric furnace is started, the initial heating process (section a) will consume more electricity, and after reaching the specified temperature (section b), it is only necessary to consume electricity steadily to maintain the temperature inside the electric furnace, and when the processing is about to be completed (section c), the consumed electricity will gradually decrease. Therefore, for large-scale electric furnace operation, the factory's electricity consumption status can be controlled according to the number of electric furnaces in sections a, b, and c, and a dynamic balance can be ensured between the number of electric furnaces in sections a and c, thereby avoiding a large number of electric furnaces in section a at the same time, causing power shortages and voltage instability in the factory, and effectively avoiding the waste of electricity caused by too many electric furnaces in section c.

[0050] In the process of building a prediction algorithm, it is necessary to first establish the corresponding neural network structure, including the following levels:

[0051] Input Layer:

[0052] The number of nodes in the input layer depends on the number of selected input variables. In the process of molten magnesia smelting, possible input variables include furnace temperature, current intensity, and material state in the furnace. Suppose the input variables are (X=[x_1,x_2,\ldots,x_n]), where (x_i) represents the (i)th input variable.

[0053] Hidden Layer:

[0054] There can be multiple hidden layers, each containing a number of nodes, and the number of nodes and layers in the hidden layer is determined by cross-validation or other model selection methods.

[0055] Output layer:

[0056] The output layer has only one node, which is used to predict the energy consumption. Assume that the output is (Y_{\text{pred}}), which is the predicted energy consumption.

[0057] The calculation process is as follows:

[0058] Calculation between input and hidden layer:

[0059] For the first hidden layer, its input is the weighted sum of the output of the input layer and processed by the activation function (\sigma):

[0060] [a^{(1)}=W^{(1)}X+b^{(1)}][h^{(1)}=\sigma(a^{(1)})]

[0061] Among them, (W^{(1)}) is the weight matrix from the input layer to the first hidden layer, (b^{(1)}) is the bias vector, and (\sigma) is the activation function (such as the sigmoid function or the ReLU function).

[0062] Calculation between hidden layers:

[0063] For the subsequent hidden layer (l), its input is the output of the previous layer:

[0064] [a^{(l)}=W^{(l)}h^{(l-1)}+b^{(l)}][h^{(l)}=\sigma(a^{(l)})]

[0065] Among them, (W^{(l)}) is the weight matrix from the (l-1)th layer to the (l)th layer, (b^{(l)}) is the bias vector, and (\sigma) is the activation function.

[0066] Calculation of the final output layer:

[0067] The output layer is the weighted sum of the outputs of the last hidden layer and processed by the activation function:

[0068] [a^{(L)}=W^{(L)}h^{(L-1)}+b^{(L)}][Y_{\text{pred}}=\sigma(a^{(L)})]

[0069] Where (W^{(L)}) is the weight matrix from the penultimate hidden layer to the output layer;

[0070] (b^{(L)}) is the bias vector;

[0071] (\sigma) is an appropriate output layer activation function, including a linear function or a function selected according to actual conditions.

[0072] Prediction process:

[0073] When training is complete, given new input data (X_{\text{future}}), the predicted value (Y_{\text{pred}}) can be calculated through forward propagation:

[0074] [Y_{\text{pred}}=f(X_{\text{future}};W,b)]

[0075] Among them, (f) represents the forward propagation process of the entire neural network;

[0076] (W) and (b) are the trained weights and biases.

[0077] In the multi-level electricity demand forecasting control in S4, a mathematical model will be established based on the current electricity consumption status, and historical data will be used as the basic data of the model forecasting control algorithm to predict the future electricity consumption status. At the same time, the forecasting algorithm will be supplemented with weighting based on the current regional temperature, air pressure and time data.

[0078] The algorithm used for time weighting is as follows:

[0079] [X_t=c+\phi_1X_{t-1}+\theta_1\epsilon_{t-1}+\epsilon_t];

[0080] Where (X_t) is the value of the time series at time (t);

[0081] (\phi_1) and (\theta_1) are the parameters of the model;

[0082] (\epsilon_t) is the noise error.

[0083] The algorithm used for environmental weighting is as follows:

[0084] [\hat{Y}_{t+1}=f(W\cdotX_t+b)];

[0085] Where (X_t) is the input feature vector;

[0086] (W) is the weight matrix;

[0087] (b) is the bias term;

[0088] Among them, the environmental weight can be achieved by adjusting (W).

[0089] In the multi-level power demand forecasting control in S4, the occasional power peak data that may occur in the actual power consumption process will be recorded, and the corresponding calculation allowable interval value will be established, that is, the power fluctuation allowable interval within the interval range. When the power demand reaches the warning level but does not exceed the fluctuation allowable interval, the abnormal status alarm will not be triggered.

[0090] The prediction algorithm update in S5 is mainly used to self-update and optimize the power multi-level demand state prediction algorithm established in S4. At the same time, during the self-update optimization parameter generation process, manual modification of specific parameters is supported.

[0091] Assuming we want to predict the power demand state (x_t), based on the observed historical data (y_t), the basic iterative steps of the algorithm are as follows:

[0092] initialization:

[0093] Initialize the parameter estimates (\hat{\theta}_0) and the error covariance matrix (P_0).

[0094] predict:

[0095] Prediction value (\hat{y}{t|t-1}=H_t\hat{\theta}{t-1})

[0096] Where (H_t) is the feature vector of historical data;

[0097] (\hat{\theta}_{t-1}) is the parameter estimate at the previous moment.

[0098] renew:

[0099] Calculate the prediction error (e_t=y_t-\hat{y}_{t|t-1});

[0100] Update parameter estimates: [\hat{\theta}t=\hat{\theta}{t-1}+K_te_t];

[0101] Where (K_t) is the gain matrix, which is calculated from the error covariance matrix (P_{t-1}) and the eigenvector (H_t): [K_t=P_{t-1}H_t^T(H_tP_{t-1}H_t^T+\lambdaI)^{-1}];

[0102] Among them, (\lambda) is a regularization parameter;

[0103] (I) is the identity matrix.

[0104] Update the error covariance matrix:

[0105] Update formula: [P_t=(I-K_tH_t)P_{t-1}]

[0106] The prediction algorithm data and historical usage data updated in S5 will be synchronously stored in the data storage system. When the updated prediction algorithm has a large error and cannot be self-updated and optimized, the system will call the historical data version for data backtracking to ensure the stability of the system and the accuracy of the prediction. In addition to storing the prediction algorithm data and historical usage data, a data backup and recovery mechanism is also established in the data storage system. Once the main data storage system fails or data is lost, the backup system can take over quickly to ensure the integrity and availability of the data.

[0107] For the update and optimization of the prediction algorithm, the system not only relies on historical usage data, but also combines real-time feedback and user behavior data. By continuously collecting and analyzing this data, the system can more accurately understand user needs and behavior patterns, thereby adjusting the prediction algorithm and improving the accuracy and practicality of the prediction results.

[0108] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-stage demand control method for fused magnesia smelting process, characterized in that: The following steps are involved: S1: Raw material processing and mixing Prepare the raw materials of the required components and weights in the fused magnesia processing process, and pre-treat the raw materials, including drying and pre-melting; S2: Electric furnace charging and heating Put the processed fused magnesia raw materials into an electric furnace, start the electric furnace to gradually heat the raw materials to above the melting point, and monitor the temperature of various places in the electric furnace through remote infrared monitoring equipment; S3: Fused magnesia processing parameter control Monitor the internal conditions of the electric furnace, including real-time pressure and chemical composition, and adjust the furnace parameters, including temperature, pressure, current, and voltage, through algorithm control equipment; S4: Electricity multi-level demand forecasting control Set the power warning range and issue warning alarms according to the warning range. At the same time, predict the subsequent power consumption status based on historical power consumption data as well as time, environment, and processing conditions; S5: Predictive control algorithm self-update A self-optimization system for the prediction algorithm is established, and the algorithm is self-optimized and updated based on the difference between the predicted power consumption status and the current actual power consumption status.

2. A multi-stage demand control method for fused magnesia smelting process according to claim 1, characterized in that: The S1 raw material processing and mixing are mainly used to classify and proportion the raw materials. The raw material processing is not limited to drying and pre-melting, but also includes screening, crushing and grading of the raw materials.

3. A multi-stage demand control method for fused magnesia smelting process according to claim 1, characterized in that: In S2, the raw materials placed inside are mainly heated by an electric furnace. During the heating process, the temperature of each characteristic point area inside the electric furnace is detected by an infrared temperature sensing device, and a corresponding temperature change curve is generated according to the detection results.

4. A multi-stage demand control method for fused magnesia smelting process according to claim 3, characterized in that: The characteristic point information in the temperature detection process mainly includes different depth areas inside the electric furnace, the end of the electric furnace close to the heat source, and the end of the electric furnace far from the heat source.

5. A multi-stage demand control method for fused magnesia smelting process according to claim 1, characterized in that: In the fused magnesia processing parameter setting in S3, it is necessary to establish a temperature-time change curve and a pressure-time change curve that change with time according to the temperature and pressure required in the production process. At the same time, the multiple groups of actual temperature change curves collected in S2 are compared with the reference temperature-time change curve to determine whether there is an uneven heating area inside the current equipment, and the heating power is adjusted by controlling the heating component.

6. A multi-stage demand control method for fused magnesia smelting process according to claim 1, characterized in that: In the multi-level power demand forecasting control in S4, a grading mechanism will be established according to the power consumption situation, including normal control, early warning control and abnormal control, and different interval values ​​will be allocated according to the divided power consumption levels, with 15 minutes as a group of demand cycles. At the same time, when the current power consumption status tends to be saturated, the start-up of the standby electric furnace will be stopped.

7. A multi-stage demand control method for fused magnesia smelting process according to claim 1, characterized in that: In the multi-level electricity demand forecasting control in S4, a mathematical model will be established based on the current electricity consumption status, and historical data will be used as the basic data of the model forecasting control algorithm to predict the future electricity consumption status. At the same time, the forecasting algorithm will be supplemented with weighting based on the current regional temperature, air pressure and time data.

8. The multi-stage demand control method for fused magnesia smelting process according to claim 1, characterized in that: In the multi-level power demand forecasting control in S4, the occasional power peak data that may occur in the actual power consumption process will be recorded, and the corresponding calculation allowable interval value will be established, that is, the power fluctuation allowable interval within the interval range. When the power demand reaches the warning level but does not exceed the fluctuation allowable interval, the abnormal status alarm will not be triggered.

9. A multi-stage demand control method for fused magnesia smelting process according to claim 1, characterized in that: The prediction algorithm update in S5 is mainly used to self-update and optimize the power multi-level demand state prediction algorithm established in S4. At the same time, during the self-update optimization parameter generation process, manual modification of specific parameters is supported.

10. A multi-stage demand control method for fused magnesia smelting process according to claim 9, characterized in that: The prediction algorithm data and historical usage data updated in S5 will be synchronously stored in the data storage system. When the updated prediction algorithm has a large error and cannot be self-updated and optimized, the system will call the historical data version for data backtracking.

Images