Method for intelligently regulating and controlling granulation effect of cylinder granulator in sintering plant

Abstract

The invention provides a method for intelligently regulating and controlling the granulation effect of a cylinder granulator in a sintering plant, which comprises the following steps of: acquiring the previous air temperature, humidity, raw material ratio, moisture added in the granulation process, utilization coefficient of a sintering machine, yield of sintered ore and drum strength as historical data information; performing positive sequence processing on the historical data information, removing abnormal data, and screening out a granulation process moisture sample added when the quality of the sintered ore is good as an optimal sample group; determining an initial structure of the BP neural network, performing optimization by using a white whale optimization algorithm BWO, and training the optimized BP neural network through the screened optimal sample group to obtain a BWO-BP model meeting boundary conditions; and predicting the optimal moisture to be added in the sintering mixture granulating process through the BWO-BP model, and granulating according to the optimal moisture. Compared with the prior art, the granulation effect of the mixture can be measured more accurately, intelligently and timely, and the granulation effect of the sintered mixture can be adjusted.

Description

Technical Field

[0001] The invention relates to the technical field of ironmaking pelletizing machines, and in particular to a method for intelligently regulating the pelletizing effect of a cylindrical pelletizing machine in a sintering plant. Background Art

[0002] As the front process of the steel production process, the ironmaking process plays a vital role in the steel production process. With the development of blast furnace ironmaking technology, the requirements for ironmaking raw materials are getting higher and higher, especially the product quality of sintered ore has a significant impact on the blast furnace. The permeability of the sintering mixture is an important process parameter in the sintering process, which significantly affects the output and quality of the sintered ore. The granulation process of the drum granulator in the sintering plant is the main equipment and method to ensure that the sintering mixture has good permeability, and it is also one of the indispensable process links in the iron ore sintering process of the sintering plant. Once the sintering plant is built, the granulation effect of the drum granulator is the most important means to ensure that the sintering mixture has good permeability if the proportion of the sintering mixture is not adjusted.

[0003] However, the current drum granulator does not have the ability to timely adjust the granulation effect of the sintering mixture. Sintering plants often rely on manual experience to adjust the granulation effect of the drum granulator when adjusting the sintering ore ratio, and will not adjust the granulation effect of the drum granulator for a long period of time. In this process, fluctuations in air temperature, humidity, and ore particle size will affect the granulation effect of the sintering mixture. Once the granulation effect fluctuates, it will directly affect the sintering process of the sintering mixture on the sintering machine, and ultimately affect the product quality of the sintered ore. As the most important equipment for controlling the granulation effect of the sintering mixture, the sintering drum granulator can timely detect the sintering mixture after granulation by the drum granulator, and adjust the granulator in time to stabilize the granulation effect of the sintering mixture. This is of great significance to stabilizing the iron ore sintering process and improving the quality of sintered ore.

[0004] At present, the main methods for measuring and controlling the pelletizing effect of the cylindrical pelletizer in the sintering plant are:

[0005] (1) Determine the particle size composition of the sintered mixture after granulation, and then adjust the granulation effect of the granulator according to the particle size distribution results.

[0006] (2) Based on the workers' existing experience, the granulation effect of the sintered mixture after granulation by the cylindrical granulator is judged, and the granulation effect of the granulator is adjusted based on this.

[0007] (3) The air permeability of the sintering material layer can be judged according to the exhaust gas temperature of the sintering trolley, and the granulation effect of the drum granulator can be inferred and adjusted accordingly.

[0008] The above prior art has the following shortcomings:

[0009] (1) The granulation effect is judged based on the particle size distribution of the sintered mixture after granulation by the granulator. The current method for measuring the particle size of the mixture in sintering plants has a long interval, often only a few times a month, and it is impossible to detect changes in the granulation effect of the mixture in time. In addition, the factors that affect the granulation effect of the mixture include not only the particle size distribution, but also other factors such as the moisture content of the mixture. Therefore, this method cannot reflect the granulation effect of the granulator in a timely and accurate manner.

[0010] (2) Judging the granulation effect of the sintering mixture based on the existing experience of the workers. This method has low accuracy in judging the granulation effect of the sintering mixture, and cannot timely detect and make adjustments when the granulation effect of the sintering mixture changes.

[0011] (3) The granulation effect of the sintered mixture after granulation by the pelletizer is judged based on the exhaust gas temperature of the sintering trolley. There is a serious lag, and it can only be judged qualitatively. The result is not accurate, and it is impossible to achieve precise control to improve the granulation effect of the mixture.

[0012] Based on the above analysis, the existing methods for measuring and adjusting the granulation effect of the granulator have obvious shortcomings, which will lead to instability in the sintering production process. If a method and system can be developed based on actual production conditions, which can more accurately, intelligently and timely measure the granulation effect of the mixture, and accurately and timely adjust the granulation effect of the sintering mixture, it will be of great significance to stabilize the sintering production. Summary of the invention

[0013] The objectives of the present invention are achieved through the following technical solutions.

[0014] Specifically, the present invention provides a method for intelligently controlling the pelletizing effect of a cylindrical pelletizer in a sintering plant, comprising the following steps:

[0015] Obtain the past temperature, humidity, raw material ratio, water added during the granulation process, sintering machine utilization factor, sintering ore yield and drum strength as historical data information;

[0016] Process the historical data information in positive order and remove abnormal data, and select the moisture sample added in the granulation process when the sintered ore quality is good as the best sample group;

[0017] Determine the initial structure of the BP neural network, optimize it using the Beluga optimization algorithm BWO, train the optimized BP neural network through the selected best sample group, and obtain a BWO-BP model that meets the boundary conditions;

[0018] The BWO-BP model is used to predict the optimal moisture that should be added during the granulation process of the sintering mixture, and granulation is performed according to the optimal moisture.

[0019] Further, after granulating according to the optimal moisture content, the method further comprises:

[0020] The added moisture is adjusted multiple times within a certain interval near the optimal moisture content, and granulation is performed multiple times;

[0021] After each granulation is completed, the permeability index of the mixture is measured using the mixture permeability measuring device. According to the measured values ​​of the permeability index of the mixture corresponding to different water addition amounts, the permeability index reference value of the drum granulator for regulating the granulation effect of the mixture is determined.

[0022] Furthermore, after determining the air permeability index reference value for the granulation effect of the drum granulator, the method further includes:

[0023] The measured value is returned to the BWO-BP model, the BWO-BP model is trained, and the prediction result is further optimized.

[0024] The advantages of the present invention are:

[0025] (1) Compared with judging the granulation effect based on the particle size distribution of the sintered mixture after granulation by a cylindrical granulator. The present invention adopts the air permeability index to reflect the granulation effect of the cylindrical granulator, which can more accurately reflect the comprehensive granulation effect of the sintered mixture. The present invention can automatically and timely determine the granulation effect of the mixture, and can automatically adjust the granulation effect of the granulator. In addition, the present invention can also predict the granulation effect of the sintered mixture through the BWO-BP model. Therefore, compared with the method of measuring particle size, the present invention can timely and accurately reflect the granulation effect of the granulator and adjust the granulator in time.

[0026] (2) Compared with judging the granulation effect of the sintering mixture based on the existing experience of workers, the present invention can more accurately reflect the granulation effect of the granulator.

[0027] (3) Compared with judging the pelletizing effect of the pelletizer based on the exhaust gas temperature of the sintering trolley, the present invention has no hysteresis, and can timely judge the pelletizing effect of the pelletizer and make adjustments quickly. In addition, the present invention has a specific value for judging the pelletizing effect of the cylindrical pelletizer, and the judgment result is more accurate.

[0028] Based on the above analysis, the existing methods for measuring and adjusting the granulation effect of the granulator have obvious deficiencies, which will lead to instability in the sintering production process. Compared with the prior art, the present invention can more accurately, intelligently and timely measure the granulation effect of the mixture, and can accurately, intelligently and timely adjust the granulation effect of the sintering mixture, and give early warning for the mixture whose granulation effect is still poor after adjustment, which is of great significance to stable sintering production. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0030] Figure 1 A flow chart of a method for intelligently controlling the granulation effect of a cylindrical granulator in a sintering plant according to an embodiment of the present invention is shown.

[0031] Figure 2 A method flow chart of the White Whale algorithm according to an embodiment of the present invention is shown.

[0032] Figure 3 A schematic diagram of the optimized BP neural network structure according to an embodiment of the present invention is shown.

[0033] Figure 4 A schematic diagram of the best fitness curve of the BWO White Whale optimization algorithm according to an embodiment of the present invention is shown.

[0034] Figure 5 A schematic diagram of the air permeability index of the sintered mixture when different moisture is added during the granulation process according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0035] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0036] like Figure 1 As shown, the present invention provides a method for intelligently controlling the granulation effect of a cylindrical granulator in a sintering plant, comprising the following steps:

[0037] S1: Obtain data information such as the previous temperature, humidity, raw material ratio, water added during the granulation process, utilization coefficient of the sintering machine, sintering ore yield and drum strength;

[0038] S2: Process the above data in positive order and remove abnormal data, and select the moisture samples added in the granulation process when the sintered ore quality is good as the best sample group;

[0039] S3: Determine the initial structure of the BP neural network, optimize it using the White Whale Optimization Algorithm (BWO), train the optimized network with the screened samples, and obtain a BWO-BP model that meets the boundary conditions;

[0040] S4: The optimal water content to be added during the granulation process of the sintering mixture is predicted by the model, and the added water content is adjusted within a certain range of the predicted value to perform granulation;

[0041] S5: After granulation is completed, the air permeability of this part of the mixture is measured using a mixture permeability measuring device, and the air permeability index benchmark value of the drum granulator for regulating the granulation effect of the mixture is determined based on the measured change value of the air permeability index of the mixture within different water addition ranges.

[0042] Furthermore, the measured values ​​can be returned to the BWO-BP model to train the BWO-BP model and further optimize the prediction results.

[0043] Air temperature, humidity, raw material ratio and water added in the granulation process significantly affect the granulation effect of the sintered mixture. The quality indicators of sintered ore such as the utilization coefficient of the sintering machine, the yield rate of sintered ore and the drum strength are directly related to the granulation effect of the sintered mixture. Therefore, the present invention inputs the data of air temperature, humidity, raw material ratio, water added in the granulation process, the utilization coefficient of the sintering machine, the yield rate of sintered ore and the drum strength into the computer, analyzes the data based on statistics, and under the 3σ principle, the abnormal value is a value in a group of measured values ​​whose deviation from the average value exceeds 3 times the standard deviation. The probability of the value outside the average value 3σ is P(|x-μ|>3σ)<=0.003, which is a very small probability event. Abnormal data was eliminated and the time series of data was processed in positive order. The evaluation indexes of sintering granulation effect and sintered ore quality were constructed using the utilization coefficient of the sintering machine, the yield of sintered ore and the drum strength. The water content added in the granulation process which had good correspondence with the above evaluation indexes was taken as the optimal sample according to the previous measured values. The initial structure of the BP neural network was determined and the BWO White Whale algorithm ( Figure 2 ) optimizes the BP neural network structure, and uses the screened data samples to optimize the BP neural network ( Figure 3 ) was trained offline to obtain the optimal moisture addition prediction model for the granulation process based on the BWO-BP neural network that met the boundary conditions.

[0044] In the present invention, Figure 2 As shown, the BWO White Whale algorithm process is as follows:

[0045] After setting the population size N and the maximum number of iterations T, the population is initialized, the fitness value and the balance factor Bf are evaluated, and the exploration phase (swimming behavior) or the development phase (predation behavior) is determined according to the value range of Bf. Then the whale fall phase is entered, and it is iterated in sequence until the maximum number of iterations is reached. The optimal solution of the fitness and the balance factor Bf (i.e., the optimal weight wij and the optimal node deviation value used in the following formula, respectively) is obtained, and the optimal solution is output to the BP neural network to obtain the neural network model with the optimal structure.

[0046] The BP neural network model optimized by the BWO algorithm is trained using the optimal data samples, and the forward propagation of input data and the back propagation of output result error are continuously performed until the output result error meets the calculation accuracy or reaches the maximum number of iterations.

[0047] The data of previous air temperature, humidity, raw material ratio, moisture added in the granulation process, utilization coefficient of sintering machine, air permeability index of sintering mixture, sintered ore yield and drum strength are selected as the input layer, and the output layer node is set as the optimal amount of moisture added in the granulation process.

[0048] Let m be the number of neurons in the hidden layer, n be the number of nodes in the input layer, L be the number of nodes in the output layer, and α be any number between 1 and 10; calculate the number of neurons in the hidden layer of the BP neural network according to the following formula:

[0049]

[0050] Select the function sigmod() as the transfer function between the input layer and the hidden layer, select the function tansig() as the transfer function between the hidden layer and the output layer, set the number of network learning times, learning rate and minimum training error; set w ij is the connection weight between the i-th node in the input layer and the j-th node in the hidden layer, b j is the bias value of the jth node in the hidden layer; then the output y of the jth node in the hidden layer j for

[0051]

[0052] In the above formula, x i is the data of previous temperature, humidity, raw material ratio, moisture added in the granulation process, etc., p is the number of nodes in the hidden layer, p=m, that is, the nodes and neurons in the hidden layer correspond one to one.

[0053] Assume w jk is the connection weight between the jth node in the hidden layer and the kth node in the output layer, b k is the deviation value of the kth node in the output layer; then the output y of the output layer node k for:

[0054]

[0055] y k is the optimal water addition in the granulation process finally predicted, S is the number of data input from the hidden layer to the output layer, and S = p.

[0056] Let y k ' is the actual value of the data sample; the optimal sample obtained above is normalized and used as the training sample of the BP network, and the mean square error function between the network output and the actual sample is obtained as follows:

[0057]

[0058] Random numbers are selected as the initial weights and initial deviations of the network neuron connections. After iteration, the error value obtained by the error function is reduced to within the range.

[0059] In the present invention, according to the prediction result of the model on the amount of water added in the granulation process of the sintering mixture, the amount of water added in the cylindrical granulator is adjusted within the range of α% above and below the predicted value. The adjusted amount of water added in the mixture fluctuates above and below the corresponding predicted value, and the value of α is between 0-2.

[0060] After granulation is completed, a portion of the mixed material is taken out at a certain interval at the outlet of the drum granulator by an automatic material taking device, and the taken mixed material is placed on an air permeability measuring device to measure the air permeability of the mixed material. The material taking interval time can be any value between 0-24 hours.

[0061] In the past, the air permeability of the mixture was generally measured by the negative pressure of the bellows during the sintering process of the sintering machine. Due to the long interval between the drum pelletizer and the sintering machine, there is a serious lag and it cannot reflect the granulation effect of the drum pelletizer in time. In addition, since the iron ore sintering process on the sintering machine has already begun, the measured air permeability includes not only the original sintering material, but also the air permeability of the sintering ore layer, combustion layer and drying preheating layer on the sintering trolley. Obviously, this cannot accurately reflect the granulation effect of the drum pelletizer.

[0062] The present invention takes samples at the outlet end of the drum granulator. After the drum granulator and the air permeability measuring device are linked, the air permeability measuring device can accurately measure the air permeability of the mixture, thereby timely and accurately reflecting the granulation effect of the drum granulator, and timely adjusting the granulation process of the drum granulator.

[0063] In the present invention, according to the value measured by the sintering mixture permeability detection device, the computer uses the Voice formula to calculate the permeability index of this part of the sintering mixture, and returns the calculated permeability to the computer for recording. The present invention uses the permeability index to accurately characterize the permeability of the sintering mixture and comprehensively reflects the granulation effect of the drum granulator.

[0064] Voice formula:

[0065] Among them, Z represents the air permeability index, Q represents the air volume passing through the material layer, A represents the exhaust area, △P represents the exhaust negative pressure, H represents the feeding height, and M is a coefficient determined by the shape and size of the particles.

[0066] According to the measured change value of the air permeability index of the mixture within different water addition ranges, the reference value (of the air permeability index) for the drum granulator to regulate the granulation effect of the mixture is determined. The measured value is returned to the BWO-BP model, and the BWO-BP model is trained to further optimize the prediction results. The measured value is then compared with the reference value. If the deviation between the measured value and the reference value is greater than a certain range, the amount of water added to the sintered mixture in the drum granulator is adjusted by computer to optimize the granulation effect of the sintered mixture and control the fluctuation of the air permeability index of the sintered mixture within a certain range. (The deviation range between the measured value and the reference value is any value between 0-50%. The fluctuation range of the air permeability index is any value between 0-50%.)

[0067] Example 1

[0068] 10,000 sets of historical data of a steel plant were processed, and 7,680 sets of data were selected as relatively reliable data. Among them, they were divided into training set, validation set, and test set according to the ratio of 8:1:1. 6,912 sets of data were selected as training set and test set, and another 768 sets of data were selected as test set.

[0069] The data of previous temperature, humidity, raw material ratio, moisture added during granulation, utilization coefficient of sintering machine, yield rate of sintered ore and drum strength are input into the computer, and the data are analyzed based on statistics. Under the 3σ principle, the abnormal value is the value in a group of measured values ​​that deviates from the average value by more than 3 times the standard deviation. The probability of a value outside the average value by 3σ is P(|x-μ|>3σ)<=0.003, which is a very rare low-probability event. The abnormal data is eliminated and the time series of the data is processed in positive order. The evaluation index of sintering granulation effect and sintered ore quality is constructed using the data of utilization coefficient of sintering machine, yield rate of sintered ore and drum strength. According to the previous measured values, the moisture data added during the granulation process that corresponds well to the above evaluation indicators is used as the optimal sample to determine the initial structure of the BP neural network. The BP neural network is optimized using the BWO White Whale optimization algorithm. After setting the population size N and the maximum number of iterations to 500, the population is initialized and the fitness value ( Figure 4 ), and then get the optimal weight w used in the following formula ij and the optimal node deviation value, output the optimal solution to the BP neural network, and obtain the neural network model with the optimal structure.

[0070] Assume that the number of network learning times is 2000, the learning rate is 0.01, and the minimum training error is 0.01;

[0071] The function sigmod() is selected as the transfer function between the input layer and the hidden layer, and the previous data such as temperature, humidity, raw material ratio, and moisture added during the granulation process are used as x in the following formula: i Input into the model and calculate y j

[0072]

[0073] The function tansig() is selected as the transfer function between the hidden layer and the output layer to obtain the optimal water addition y in the granulation process obtained by prediction. k

[0074]

[0075] Let y' k is the actual value of the data sample; the optimal sample obtained above is normalized and used as the training sample of the BP network, and the mean square error function between the network output and the actual sample is obtained as follows:

[0076]

[0077] Random numbers are selected as the initial weights and initial deviations of the network neuron connections. After iteration, the error value obtained by the error function is reduced to within the range.

[0078] After training the model, select the validation set to verify the prediction effect. After the prediction effect meets the requirements, select the test set data for testing.

[0079] At present, the original moisture content of the sintering mixture is 3%, and the model predicts that the amount of moisture added during the current sintering mixture granulation process is 5%. The amount of moisture added to the sintering mixture in the drum granulator is adjusted within the range of 1% above and below the predicted value of 5%, and the specific amount of moisture added is 4%, 4.5%, 5.5% and 6%.

[0080] After granulation is completed, a portion of the mixture is taken out at the outlet of the drum granulator through the automatic material taking device at intervals of 15 minutes, and the taken out mixture is placed on the permeability measuring device to measure the permeability of the mixture. The computer uses the Voice formula to calculate the permeability index of this part of the sintered mixture, and the calculated permeability index is returned to the computer for recording. Figure 5 The air permeability index of the sintered mixture is measured when different moisture is added during the granulation process. It can be seen from the figure that the optimal moisture addition amount is 5.5%, at which time the air permeability index is 80.

[0081] According to the measured change value of the air permeability index of the mixture within different water addition ranges, the reference value of the (air permeability index) of the drum granulator to control the granulation effect of the mixture is determined to be 80. The measured value is returned to the BWO-BP model, and the error is calculated based on the predicted optimal water addition value and the measured optimal water addition value combined with the error function, and the prediction accuracy is controlled at 99.9%. The mean square error function is:

[0082]

[0083] After that, the measured value is compared with the reference value each time. If the deviation between the measured value and the reference value is greater than 30%, the amount of water added to the sintering mixture in the cylindrical granulator is adjusted by the computer. For example, when the permeability index of the sintering mixture is 160, the amount of water added in the cylindrical granulator process is increased by 0.5%. At this time, the permeability index of the sintering mixture is reduced to 90. At this time, the deviation from the reference value of 80 is within 30%, and the adjustment of the water content in the granulation process is stopped. If the permeability index is 120 after the water content is increased by 0.5%, the amount of water added is continued to be increased by 0.5% until the permeability index of the sintering mixture is within 104. Through the linkage of the cylindrical granulator and the permeability index measuring device, the fluctuation of the permeability index of the sintering mixture can be stably controlled within the range of 80-104.

[0084] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for intelligently controlling the pelletizing effect of a cylindrical pelletizing machine in a sintering plant, characterized in that: The steps include: Obtain the past temperature, humidity, raw material ratio, water added during the granulation process, sintering machine utilization factor, sintering ore yield and drum strength as historical data information; Process the historical data information in positive order and remove abnormal data, and select the moisture sample added in the granulation process when the sintered ore quality is good as the best sample group; Determine the initial structure of the BP neural network, optimize it using the Beluga optimization algorithm BWO, train the optimized BP neural network through the selected best sample group, and obtain a BWO-BP model that meets the boundary conditions; The BWO-BP model is used to predict the optimal moisture that should be added during the granulation process of the sintering mixture, and granulation is performed according to the optimal moisture.

2. The method for intelligently controlling the pelletizing effect of a cylindrical pelletizer in a sintering plant according to claim 1, characterized in that: After granulating according to the optimal moisture content, the method further comprises: The added moisture is adjusted multiple times within a certain interval near the optimal moisture content, and granulation is performed multiple times; After each granulation is completed, the permeability index of the mixture is measured using the mixture permeability measuring device. According to the measured values ​​of the permeability index of the mixture corresponding to different water addition amounts, the permeability index reference value of the drum granulator for regulating the granulation effect of the mixture is determined.

3. The method for intelligently controlling the pelletizing effect of a cylindrical pelletizer in a sintering plant according to claim 2, characterized in that: After determining the permeability index reference value of the drum granulator for regulating the granulation effect of the mixed material, it further includes: The measured value is returned to the BWO-BP model, the BWO-BP model is trained, and the prediction result is further optimized.

4. The method for intelligently controlling the pelletizing effect of a cylindrical pelletizer in a sintering plant according to claim 1, characterized in that: Under the 3σ principle, the abnormal data refers to the value in a group of measured values ​​whose deviation from the mean value exceeds 3 times the standard deviation.

5. The method for intelligently controlling the pelletizing effect of a cylindrical pelletizer in a sintering plant according to claim 1, characterized in that: The process of the White Whale Optimization Algorithm BWO includes: After setting the population size N and the maximum number of iterations T, the population is initialized, the fitness value and the balance factor Bf are evaluated, and the exploration phase or the development phase is determined based on the value range of Bf. Then the whale fall phase is entered, and it is iterated in sequence until the maximum number of iterations is reached, thereby obtaining the optimal solution for the fitness and the balance factor Bf. The optimal solution is output to the BP neural network to obtain the BWO-BP model with the optimal structure.

6. The method for intelligently controlling the pelletizing effect of a cylindrical pelletizer in a sintering plant according to claim 1, characterized in that: The step of training the optimized BP neural network by using the selected best sample group includes: The historical data information is used as the input layer, and the output layer node is set as the optimal amount of water added in the granulation process.

7. The method for intelligently controlling the pelletizing effect of a cylindrical pelletizer in a sintering plant according to claim 6, characterized in that: Further including: Let m be the number of neurons in the hidden layer, n be the number of nodes in the input layer, L be the number of nodes in the output layer, and α be any number between 1 and 10; calculate the number of neurons in the hidden layer of the BP neural network according to the following formula:

8. The method for intelligently controlling the pelletizing effect of a cylindrical pelletizer in a sintering plant according to claim 7, characterized in that: Further including: Select the function sigmod() as the transfer function between the input layer and the hidden layer, select the function tansig() as the transfer function between the hidden layer and the output layer, set the number of network learning times, learning rate and minimum training error; set w ij is the connection weight between the i-th node in the input layer and the j-th node in the hidden layer, b j is the bias value of the jth node in the hidden layer; then the output y of the jth node in the hidden layer j for In the above formula, x i is the historical data information, p is the number of nodes in the hidden layer; p=m, that is, the nodes and neurons in the hidden layer correspond one to one; Assume w jk is the connection weight between the jth node in the hidden layer and the kth node in the output layer, b k is the deviation value of the kth node in the output layer; then the output y of the output layer node k for: y k is the optimal water addition in the granulation process finally predicted, S is the number of data input from the hidden layer to the output layer, and S = p.

9. The method for intelligently controlling the pelletizing effect of a cylindrical pelletizer in a sintering plant according to claim 8, characterized in that: Further including: Let y k ' is the actual value of the data sample; the optimal sample group is normalized and used as the training sample of the BP neural network, and the mean square error function between the network output and the actual sample is obtained as Random numbers are selected as the initial weights and initial deviation values ​​of the network neuron connections, and after iteration, the error value obtained by the mean square error function is reduced to within the preset range.

10. The method for intelligently controlling the pelletizing effect of a cylindrical pelletizer in a sintering plant according to claim 1, characterized in that: The calculation formula of the air permeability index is as follows: Among them, Z is the air permeability index, Q represents the air volume passing through the material layer, A represents the exhaust area, △P represents the exhaust negative pressure, H represents the feeding height, and M is a coefficient determined by the shape and size of the particles.

Images