Method for optimizing operation control parameters of powder concentrator based on multiple linear regression algorithm

Abstract

The invention relates to the technical field of optimization of operation control parameters of a powder concentrator, and discloses a method for optimizing the operation control parameters of the powder concentrator based on a multiple linear regression algorithm. The method comprises the following steps: step 1, setting sample parameters; 2, establishing a multiple linear regression equation model; step 3, iterative training of the multiple linear regression equation model; step 4, calculating an offset item w0; 5, calculating weight values w1, w2, w3, w4, w5, w6 and w7 of the multiple linear regression equation; and step 6, acquiring operation control parameters of the powder concentrator. The method has the beneficial effects that the operation control parameters of the powder concentrator are optimized, fitted and matched under the condition that the powder concentrator is not broken down or is not shut down, and scientific and reasonable historical operation control parameters are obtained under the working condition that the ideal efficiency of the powder concentrator is formulated.

Description

Technical Field

[0001] This invention relates to the field of air classifier operation control parameter optimization technology, and in particular to a method for optimizing air classifier operation control parameters based on a multiple linear regression algorithm. Background Technology

[0002] Air classifiers are crucial components in cement plants. Currently, the most widely used and advanced air classifier is the O-Sepa classifier, renowned for its high efficiency. Its eddy current classification principle is considered both advanced and scientifically sound. Its advantages include low energy consumption, simple operation, convenient fineness adjustment, minimal wear, easy maintenance, large powder processing capacity, and high classification efficiency. In cement plants, it is typically used in conjunction with raw material mills, coal mills, or cement mills, serving as a key piece of equipment for classifying the fineness of raw materials, coal powder, and cement. The O-Sepa classifier works as follows: the material to be classified is fed into the classifier through two inlets at the top, fully dispersed by a spreading disc and buffer plate, and falls into the classification zone. The classification airflow mostly originates from the mill, entering through the tangential primary and secondary air inlets, and then horizontally into the classification zone via guide vanes. Inside the classifier, a cage composed of vertical and horizontal blades maintains a constant pressure difference throughout the entire classification zone during rotation, ensuring stable and uniform airflow and creating favorable conditions for accurate powder classification. The material flows from top to bottom, providing each particle with multiple opportunities for repeated sorting, and each sorting is carried out under precise conditions of centrifugal force and horizontal airflow balance. Fine powder, moving from the outside in, overcomes the adverse effects of the sidewalls. Due to the extended sorting distance, the particles that finally fall into the cone section undergo three more air separations, greatly improving sorting efficiency.

[0003] In actual production, the main parameters affecting the air classifier efficiency of the O-Sepa classifier (without considering improvements to the classifier structure and local air pressure) are rotor speed, primary air volume, secondary air volume, tertiary air volume, primary air pressure, secondary air pressure, and tertiary air pressure. In existing technologies, optimizing the control parameters for improving the classifier's efficiency is primarily achieved through long-term production practice and numerous trial-and-error adjustments to these parameters. The main technical problem with the existing methods for optimizing the control parameters of O-Sepa air classifiers is that while the current air classifier efficiency is around 85% to 90%, improving efficiency requires balancing energy consumption and wear. Furthermore, each order-of-magnitude increase in efficiency necessitates adjusting numerous control parameters, and attempting to improve efficiency through adjustments introduces too many uncertainties, potentially leading to equipment malfunctions or shutdowns, thus disrupting normal production. Consequently, it has been impossible to achieve an efficiency exceeding 95%. To address these shortcomings, the inventors have developed a method based on a multiple linear regression algorithm to optimize the air classifier's operating control parameters, effectively resolving these technical problems. Summary of the Invention

[0004] To address the problems in the prior art, this invention provides a method for optimizing the operating control parameters of an air classifier based on a multiple linear regression algorithm. This invention uses historical operating control data of the air classifier's rotor speed, primary air volume, secondary air volume, tertiary air volume, primary air pressure, secondary air pressure, and tertiary air pressure as feature vector samples. Simultaneously, it uses the air classifier efficiency corresponding to different historical operating data of rotor speed, primary air volume, secondary air volume, tertiary air volume, primary air pressure, secondary air pressure, and tertiary air pressure as the target vector to fit and form a multiple linear regression equation. The multiple linear regression equation is iteratively trained using a mini-batch gradient descent algorithm to form a multiple linear regression equation model with a goodness-of-fit approaching one. The weights and biases of the factors influencing the feature vector and target vector are calculated using this multiple linear regression equation model with a goodness-of-fit approaching one. Under the condition that the weights and biases are known, and given an ideal efficiency for the air classifier, the operating control parameters of the air classifier that best match the ideal efficiency value are calculated using the multiple linear regression equation formula with a goodness-of-fit approaching one.

[0005] This invention provides a method for optimizing the operating control parameters of an air classifier based on a multiple linear regression algorithm:

[0006] Step 1: Setting Sample Parameters: Set the O-Sepa classifier efficiency as the target vector Y, the rotor speed as X1 (r / min), and the primary air volume as X2 (m³). 3 / min; the secondary air volume is set to X3, in m³ / min. 3 / min; the tertiary air volume is set to X4, in m³ / min. 3 / min; Primary air pressure is set to X5, in MPa; Secondary air pressure is set to X6, in MPa; Tertiary air pressure is set to X7, in MPa; The rotor speed, primary air volume, secondary air volume, tertiary air volume, primary air pressure, secondary air pressure, and tertiary air pressure mentioned above are the feature vectors; At the same time, the air classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7 corresponding to the historical operating control parameter sample data of X1, X2, X3, X4, X5, X6, and X7 are selected, and the air classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7 are the target vectors;

[0007] Step Two: Establishing the Multiple Linear Regression Equation Model: The seven historical control data points from Step One—rotor speed X1, primary air volume X2, secondary air volume X3, tertiary air volume X4, primary air pressure X5, secondary air pressure X6, and tertiary air pressure X7—are used as feature vector samples. For each of these seven historical control data points, the feature vector sample corresponds to a historical efficiency sample of the air classifier. Based on the linear regression equation y=wx+b, the feature vector samples of the seven historical operating data points are expressed using the fitting function formula as follows: The above formula can be expressed in vector form as follows: In the above formula, y represents the target vector classifier efficiency; x represents the feature vector sample; w1, w2, w3, w4, w5, w6, and w7 are the feature vector sample weights corresponding to rotor speed, primary air volume, secondary air volume, tertiary air volume, primary air pressure, secondary air pressure, and tertiary air pressure; T represents the transpose of the vector; b represents the bias term; the historical operating data samples of X1, X2, X3, X4, X5, X6, and X7 and the corresponding classifier efficiency samples Y1, Y2, Y3, Y4, Y5, Y6, and Y7 are fitted onto the coordinate system to form a fitting plane, which is expressed by the formula: To facilitate the calculation of the above formula, it is proposed to add a supplementary term X0 and set X0=1. Substituting X0 into the above formula, we can integrate it into a new fitting plane formula (i.e., the multiple linear regression equation):

[0008] ;

[0009] The integrated formula for fitting the plane is converted into a matrix calculation method. In the above formula, hw(x) represents the fitted plane, i.e., the multiple linear regression equation.

[0010] Step 3: Iterative Training of the Multiple Linear Regression Equation Model: The historical operation control data samples X1, X2, X3, X4, X5, X6, and X7 from Step 1, and the corresponding air classifier efficiency sample data Y1, Y2, Y3, Y4, Y5, Y6, and Y7, are input into the multiple linear regression equation model established in Step 2 in the maximum batch size. The loss function is then used to train the model. The mini-batch gradient descent algorithm is used to iteratively update the historical running data samples of X1, X2, X3, X4, X5, X6, and X7 and the efficiency sample data of Y1, Y2, Y3, Y4, Y5, Y6, and Y7 along the tangent direction of the descent of the iterative training curve with an infinitely small learning rate. This process is repeated until the lowest point of the iterative training curve is reached, resulting in a multiple linear regression equation model with a goodness of fit approaching one.

[0011] Step 4: Calculation of the bias term w0:

[0012] In the multiple linear regression equation model, for each sample, the true value is 3y. i And the predicted value 4w T x (i) There will be an error between them. Using formula In the above formula, i represents the sample matrix, and T represents the vector transpose; error Formula for following a normal distribution , the formula Substitute into the formula The formula is obtained from , the formula Expressed as a loss function Since the loss function formula involves cumulative multiplication, the cumulative multiplication in the above formula is converted into a cumulative summation expression as follows: By formula Simplify to obtain the loss function Then, by taking the partial derivative of the loss function and assuming the partial derivative is zero, the formula for calculating the weights is derived. In the formula for calculating the weight values ​​above, w0 represents the weight value, T represents the vector transpose, x represents the sample matrix of historical operating parameters including X1, X2, X3, X4, X5, X6, and X7, and y represents the sample matrix of classifier efficiency including Y1, Y2, Y3, Y4, Y5, Y6, and Y7; using the formula for calculating the weight values... Substitute the historical operating parameter sample matrices of X1, X2, X3, X4, X5, X6, and X7 and the classifier efficiency sample matrices of Y1, Y2, Y3, Y4, Y5, Y6, and Y7 into the formula to calculate the value of the w0 bias term.

[0013] Step 5: Calculation of weights w1, w2, w3, w4, w5, w6, and w7 in the multiple linear regression equation: Substitute the w0 bias term value calculated in Step 4 into the fitting plane formula. In the middle, the weight values ​​of w1, w2, w3, w4, w5, w6, and w7 are calculated respectively;

[0014] Step Six: Obtaining the Air Classifier's Operating Control Parameters: Set the ideal efficiency value for the air classifier (e.g., 96%), and simultaneously substitute the bias term w0 and weight values ​​w1, w2, w3, w4, w5, w6, and w7 calculated in Step Four into the fitting plane formula. The values ​​of the classifier's operating control parameters X1, X2, X3, X4, X5, X6, and X7, which are well-matched to the ideal efficiency value of the classifier, were calculated respectively, and the scientific and reasonable historical operating control parameters under the proposed ideal classifier efficiency were obtained.

[0015] In step one, the prerequisite for using the historical operating control parameter sample data of classifiers X1, X2, X3, X4, X5, X6, and X7, and the corresponding classifier efficiency data Y1, Y2, Y3, Y4, Y5, Y6, and Y7, is that the feed amount and model of the classifier are the same, and the fineness of the classifier is controlled at 300 mesh.

[0016] In step one, the historical operating control parameter sample data of classifiers X1, X2, X3, X4, X5, X6, and X7, and the corresponding classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7 are stored in the classifier's control system. Each set of historical operating control parameter data corresponds to the efficiency of a classifier; and abnormal historical operating control parameter data are removed.

[0017] In step three, for each target sample and each feature vector sample, the black dots projected in the coordinate system represent the true values, and a multiple linear regression equation is formed on the fitting plane. For each sample in the coordinate system, the change of the w0 bias term will cause the fitting plane to adjust upward or downward. During repeated iterative training, the fitting plane is fine-tuned by changing the W0 bias term to make it more accurate. The core influence is the weight values ​​of w1, w2, w3, w4, w5, w6, and w7. At this time, the bias term b is equal to the bias term w0.

[0018] In step three, the predicted value is the projection point of the true value sample onto the fitted multiple linear regression equation; error It is the bias term b (distance) between the actual value and the predicted value.

[0019] In step three, the error between the true value and the predicted value for each sample is... They are independent entities with the same distribution, and follow a mean of 0 and a variance of w. 2 It follows a normal distribution (Gaussian distribution).

[0020] The beneficial effects of this invention are as follows: By using the historical operating control parameters of the air classifier as feature vector samples and the corresponding air classifier efficiency as target vector samples, a multiple linear regression equation model with a goodness of fit approaching one is formed. Using the above model, the bias term and weight value of the multiple linear regression equation model are calculated. Under the condition of a proposed ideal air classifier efficiency, the calculated bias term and weight value are substituted into the multiple linear regression equation model to calculate the historical operating control parameters of the air classifier. This achieves optimized fitting of the air classifier operating control parameters without causing malfunctions or stopping the air classifier, resulting in scientifically reasonable historical operating control parameters under the proposed ideal air classifier efficiency. Attached Figure Description

[0021] Figure 1 This is a coordinate graph of the multiple linear regression equation model of the present invention;

[0022] Figure 2 This is a normal distribution diagram of the error in the multiple linear regression equation model of this invention;

[0023] Figure 3 This is a coordinate graph of the mini-batch gradient descent algorithm of this invention;

[0024] Figure 4 This is a comparison chart of the classifier operation control data and efficiency in the prior art of this invention with the classifier operation control data and efficiency after fitting and matching according to this invention;

[0025] The figures are labeled as follows: 1. Multiple linear regression equation, 2. Fitting plane, 3. True value, 4. Predicted value, 5. Intercept. Detailed Implementation

[0026] The present invention will be further described below with reference to specific embodiments. These embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention.

[0027] like Figure 1-4 As shown, this invention provides a method for optimizing the operating control parameters of an air classifier based on a multiple linear regression algorithm:

[0028] Step 1: Setting Sample Parameters: Set the O-Sepa classifier efficiency as the target vector Y, the rotor speed as X1 (r / min), and the primary air volume as X2 (m³). 3 / min; the secondary air volume is set to X3, in m³ / min. 3 / min; the tertiary air volume is set to X4, in m³ / min. 3 / min; Primary air pressure is set to X5, unit is MPa; Secondary air pressure is set to X6, unit is MPa; Tertiary air pressure is set to X7, unit is MPa; The rotor speed, primary air volume, secondary air volume, tertiary air volume, primary air pressure, secondary air pressure, and tertiary air pressure mentioned above are the feature vectors; At the same time, the air classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7 corresponding to the historical operating control parameter sample data of X1, X2, X3, X4, X5, X6, and X7 are selected, and the air classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7 are the target vectors; Air classifier X1, X2, X 3. The historical operating control parameter sample data of X4, X5, X6, and X7, and the corresponding air classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7, are based on the premise that the air classifier feed rate and model are the same, and the air classifier fineness is controlled at 300 mesh. The historical operating control parameter sample data of air classifier X1, X2, X3, X4, X5, X6, and X7, and the corresponding air classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7 are stored in the air classifier's control system. Each set of historical operating control parameter data corresponds to a set of air classifier efficiencies; and abnormal historical operating control parameter data are excluded.

[0029] Step Two: Establishing the Multiple Linear Regression Equation Model: The seven historical control data points from Step One—rotor speed X1, primary air volume X2, secondary air volume X3, tertiary air volume X4, primary air pressure X5, secondary air pressure X6, and tertiary air pressure X7—are used as feature vector samples. For each of these seven historical control data points, the feature vector sample corresponds to a historical efficiency sample of the air classifier. Based on the linear regression equation y=wx+b, the feature vector samples of the seven historical operating data points are expressed using the fitting function formula as follows: The above formula can be expressed in vector form as follows: In the above formula, y represents the target vector powder classifier efficiency; x represents the feature vector sample; w1, w2, w3, w4, w5, w6, and w7 are the feature vector sample weight values ​​corresponding to rotor speed, primary air volume, secondary air volume, tertiary air volume, primary air pressure, secondary air pressure, and tertiary air pressure; T represents the transpose of the vector; and b represents the bias term.

[0030] The historical operating data samples of X1, X2, X3, X4, X5, X6, and X7, along with the corresponding classifier efficiency sample data of Y1, Y2, Y3, Y4, Y5, Y6, and Y7, are fitted onto a coordinate system to form a fitting plane 2. The fitting plane 2 is expressed by the formula: To facilitate the calculation of the above formula, it is proposed to add a supplementary term X0 and set X0=1. Substituting X0 into the above formula, we integrate it into a new fitting plane formula 2 (i.e., multiple linear regression equation 1):

[0031]

[0032] The integrated formula for the new fitting plane is converted into a matrix calculation method. In the above formula, hw(x) represents the fitted plane 2;

[0033] Step 3: Iterative Training of the Multiple Linear Regression Equation Model: The historical operation control data samples X1, X2, X3, X4, X5, X6, and X7 from Step 1, and the corresponding air classifier efficiency sample data Y1, Y2, Y3, Y4, Y5, Y6, and Y7, are input into the multiple linear regression equation model established in Step 2 in the maximum batch size. The loss function is then used to train the model. Mini-batch gradient descent algorithm (e.g.) is used. Figure 3 As shown), with a learning rate that approaches infinity, the historical running data samples of X1, X2, X3, X4, X5, X6, and X7, and the efficiency sample data of Y1, Y2, Y3, Y4, Y5, Y6, and Y7 are iteratively updated along the tangent direction of the descending iterative training curve until the lowest point of the iterative training curve is reached. This iterative training continues until a multiple linear regression equation model with a goodness of fit approaching one is obtained. For each target quantity sample and each feature vector sample, the black dot projected in the coordinate system represents the true value 3, and a multiple linear regression equation 1 (black straight line) is formed on the fitting plane 2. For each sample in the coordinate system, changing the w0 bias term will adjust the fitting plane upwards or downwards. During the iterative training process, fine-tuning is performed by changing the W0 bias term to make the fitting plane more accurate. The core impact is on the weight values ​​of w1, w2, w3, w4, w5, w6, and w7. At this time, the bias term b is equal to the bias term w0. Figure 1 As shown, for each sample, the true value is 3y. i And the predicted value 4w T x (i) There will be an error between them. Using formula The predicted value is the projection of the true value sample onto the fitted multiple linear regression equation 1; the error... The bias term b (distance) is the difference between the true value and the predicted value, where i represents the sample in the formula above; for example... Figure 2 As shown, the error between the true value and the predicted value for each sample. They are independent entities with the same distribution, and follow a mean of 0 and a variance of w. 2 It follows a normal distribution (Gaussian distribution).

[0034] Step 4: Calculation of the bias term w0:

[0035] In the multiple linear regression equation model, for each sample, the true value is 3y. i And the predicted value 4w T x (i) There will be an error between them. Using formula In the above formula, i represents the sample matrix, and T represents the vector transpose; error Formula for following a normal distribution , the formula Substitute into the formula The formula is obtained from , the formula Expressed as a loss function Since the loss function formula involves cumulative multiplication, the cumulative multiplication in the above formula is converted into a cumulative summation expression as follows: By formula Simplify to obtain the loss function Then, by taking the partial derivative of the loss function and assuming the partial derivative is zero, the formula for calculating the weights is derived. In the formula for calculating the weight values ​​above, w0 represents the weight value, T represents the vector transpose, x represents the sample matrix of historical operating parameters including X1, X2, X3, X4, X5, X6, and X7, and y represents the sample matrix of classifier efficiency including Y1, Y2, Y3, Y4, Y5, Y6, and Y7; using the formula for calculating the weight values... Substitute the historical operating parameter sample matrices of X1, X2, X3, X4, X5, X6, and X7 and the classifier efficiency sample matrices of Y1, Y2, Y3, Y4, Y5, Y6, and Y7 into the formula to calculate the value of the w0 bias term.

[0036] Step 5: Calculation of weights w1, w2, w3, w4, w5, w6, and w7 in the multiple linear regression equation: Substitute the w0 bias term value calculated in Step 4 into the fitting plane formula. In the middle, the weight values ​​of w1, w2, w3, w4, w5, w6, and w7 are calculated respectively;

[0037] Step Six: Obtaining the Air Classifier's Operating Control Parameters: Set the ideal efficiency value for the air classifier (e.g., 96%), and simultaneously substitute the bias term w0 and weight values ​​w1, w2, w3, w4, w5, w6, and w7 calculated in Step Four into the fitting plane formula. The values ​​of the classifier's operating control parameters X1, X2, X3, X4, X5, X6, and X7, which are well-matched to the ideal efficiency value of the classifier, were calculated respectively, and the scientific and reasonable historical operating control parameters under the proposed ideal classifier efficiency were obtained.

[0038] In the setting of the sample parameters in the aforementioned step, the classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7 corresponding to the historical operating control parameter sample data of X1, X2, X3, X4, X5, X6, and X7 are selected, and the classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7 are used as the target vector. The purpose of doing this is to use the historical operating control parameter sample data of the classifier as the feature vector, and simultaneously use the classifier efficiency sample data as the target vector, in... Figure 1 The coordinates are fitted on plane 2 to form a multiple linear regression equation 1.

[0039] In setting the sample parameters in the aforementioned steps, the historical operating control parameter sample data of classifiers X1, X2, X3, X4, X5, X6, and X7, and the corresponding classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7, are based on the premise that the classifier feed rate and model are the same, the classifier fineness is controlled at 300 mesh, and abnormal historical operating control parameter data are excluded. The main purpose of doing so is to ensure the scientific accuracy of the historical operating control parameter sample of the classifier under the conditions of the same feed rate and uniform classifier model; and to provide the necessary conditions for forming a multiple linear regression equation model with a goodness of fit approaching one.

[0040] like Figure 1 As shown, in step two, the establishment of the multiple linear regression equation model involves fitting the historical operating data samples of X1, X2, X3, X4, X5, X6, and X7, along with the corresponding air classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7, onto the coordinate system to form a fitting plane 2. This forms the model of multiple linear regression equation 1, aiming to fit as many of the true values ​​of the historical operating data samples of X1, X2, X3, X4, X5, X6, and X7, along with the corresponding air classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7, as close as possible to the vicinity of multiple linear regression equation 1.

[0041] like Figure 1 , 2As shown in Figure 3, in step three, for each sample in the coordinate system, changing the w0 bias term will adjust the fitting plane upwards or downwards. During iterative training, fine-tuning is performed by changing the W0 bias term to make the fitting plane more accurate. The core influence is on the weight values ​​of w1, w2, w3, w4, w5, w6, and w7. At this time, the bias term b equals the bias term w0. Figure 1 As shown, for each sample, the true value is 3y. i And the predicted value 4w T x (i) There will be an error between them. Using formula The predicted value is the projection of the true value sample onto the fitted multiple linear regression equation 1; the error... The bias term b (distance) is the difference between the true value and the predicted value, where i represents the sample in the formula above; for example... Figure 2 As shown, the error between the true value and the predicted value for each sample They are independent entities with the same distribution, and follow a mean of 0 and a variance of w. 2 The model follows a normal (Gaussian) distribution. The main purpose of this approach is to utilize a loss function and a mini-batch gradient descent algorithm, with a learning rate that approaches an infinitesimally small value, to iteratively train the model along the tangent of the descent curve of the iterative training curve. This iteratively trains the model on historical running data samples (X1, X2, X3, X4, X5, X6, X7) and efficiency data samples (Y1, Y2, Y3, Y4, Y5, Y6, Y7), minimizing the error between the true and predicted values. Ultimately, this results in a multiple linear regression equation model with a goodness of fit approaching one.

[0042] like Figure 2 As shown, in step three, the error between the true value and the predicted value of each sample... They are independent entities with the same distribution, and follow a mean of 0 and a variance of w. 2 It follows a normal distribution (Gaussian distribution). According to Figure 2 From the normal distribution curve of the mean error, we can see that the sample data accounts for the majority when the normal distribution curve is close to 0, and the sample data accounts for the minority when the normal distribution curve is on the left and right sides.

[0043] In step four of this invention, the bias term w0 is calculated based on the error between the true and predicted values ​​of the feature vector sample data in the multiple linear regression equation model 1, and because the errors are independent and have the same distribution, and follow a mean of 0 and a variance of w. 2 The normal distribution characteristics of the formula Substituting into the normal distribution formula of the error The loss function formula is simplified to obtain the formula. Then, the loss function containing multiplication is converted into an summation formula. Subsequently, the partial derivative of the loss function is used to derive the calculation formula for w0. Finally, the known data of historical running data samples X1, X2, X3, X4, X5, X6, X7 and efficiency samples Y1, Y2, Y3, Y4, Y5, Y6, Y7 are substituted into the calculation formula for w0 to calculate the value of the partial term w0.

[0044] In step five of this invention, the weight values ​​w1, w2, w3, w4, w5, w6, and w7 of the multiple linear regression equation are calculated. The main purpose of this is to use the value of w0 calculated in step four as a known value and substitute it into the iteratively trained multiple linear regression equation 1 to calculate the values ​​of the weight values ​​w1, w2, w3, w4, w5, w6, and w7 respectively.

[0045] In step six of this invention, the acquisition of the air classifier's operating control parameters serves the following purpose: Under the condition of setting the ideal efficiency of the air classifier, the known bias term w0 value calculated in step four and the known weight values ​​of w1, w2, w3, w4, w5, w6, and w7 calculated in step five are substituted into the iteratively trained multiple linear regression equation 1 to calculate the air classifier operating control parameters X1, X2, X3, X4, X5, X6, and X7 values ​​that fit well with the set ideal air classifier efficiency. This achieves optimized fitting of the air classifier's operating control parameters without causing malfunctions or shutdowns, resulting in scientifically reasonable historical operating control parameters under the proposed ideal air classifier efficiency conditions.

[0046] In steps one through six above, each step interacts and influences the others. Those skilled in the art should view steps one through six as a whole. By using the historical operating control parameters of the air classifier as feature vector samples and the corresponding air classifier efficiency as target vector samples, a multiple linear regression equation model with a goodness of fit approaching one is formed. Using the above model, the bias term and weight value of the multiple linear regression equation model are calculated. Under the condition of a proposed ideal air classifier efficiency, the calculated bias term and weight value are substituted into the multiple linear regression equation model to calculate the historical operating control parameters of the air classifier. This achieves optimized fitting of the air classifier operating control parameters without causing malfunctions or stopping the air classifier, resulting in scientifically reasonable historical operating control parameters under the proposed ideal air classifier efficiency conditions.

[0047] like Figure 4In existing technologies, the adjustment of the operating control parameters of air classifiers involves experimental adjustments to the rotor speed, primary air volume, secondary air volume, tertiary air volume, primary air pressure, secondary air pressure, and tertiary air pressure during production practice. However, excessive adjustments to these operating control parameters can cause abnormal operation of the air classifier, even leading to malfunctions or shutdowns, thus affecting its normal production operation.

[0048] Using the method of this invention, the values ​​of the air classifier operating control parameters X1, X2, X3, X4, X5, X6, and X7 calculated in step six are input into the virtual air classifier central control system software. The central control system's central operating parameter control program simulates operation, and the air classifier's operating efficiency is simulated through the air classifier efficiency calculation module within the control program. This allows for the scientific verification of the air classifier operating control parameters X1, X2, X3, X4, X5, X6, and X7 calculated in step six, under conditions where the air classifier will not malfunction or stop. Once the air classifier operating control parameters X1, X2, X3, X4, X5, X6, and X7 calculated in step six are verified to be correct, they are then input into the air classifier's central control program for normal operation and production. From the above... Figure 1 The comparison shows that the values ​​of the classifier operation control parameters X1, X2, X3, X4, X5, X6, and X7 calculated in step six have indeed improved the production efficiency of the classifier and produced unexpected technical effects, as verified by production practice.

[0049] Various modifications to the above embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be adapted to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for optimizing the operating control parameters of an air classifier based on a multiple linear regression algorithm, characterized in that: Step 1: Setting Sample Parameters: Set the O-Sepa classifier efficiency as the target vector Y, the rotor speed as X1 (r / min), and the primary air volume as X2 (m³). 3 / min; the secondary air volume is set to X3, in m³ / min. 3 / min; the tertiary air volume is set to X4, in m³ / min. 3 / min; Primary air pressure is set to X5, unit is MPa; Secondary air pressure is set to X6, unit is MPa; Tertiary air pressure is set to X7, unit is MPa; The rotor speed, primary air volume, secondary air volume, tertiary air volume, primary air pressure, secondary air pressure, and tertiary air pressure mentioned above are feature vectors; At the same time, the air classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7 corresponding to the historical operating control parameter sample data of X1, X2, X3, X4, X5, X6, and X7 are selected, and the air classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y7 are...

6. Y7 data is the target vector; Step 2, Establishment of the multiple linear regression equation model: The seven historical control data points from Step 1 above—rotor speed X1, primary air volume X2, secondary air volume X3, tertiary air volume X4, primary air pressure X5, secondary air pressure X6, and tertiary air pressure X7—are used as feature vector samples; for each of the above seven historical control data points, the feature vector sample of each historical operating data point corresponds to a historical efficiency sample of the air classifier; according to the linear regression equation y=wx+b, the feature vector samples of the seven historical operating data points are expressed by the fitting function formula as follows: The above formula can be expressed in vector form as follows: In the above formula, y represents the target vector classifier efficiency; x represents the feature vector sample; w1, w2, w3, w4, w5, w6, and w7 are the feature vector sample weights corresponding to rotor speed, primary air volume, secondary air volume, tertiary air volume, primary air pressure, secondary air pressure, and tertiary air pressure; T represents the transpose of the vector; b represents the bias term; the historical operating data samples of X1, X2, X3, X4, X5, X6, and X7 and the corresponding classifier efficiency samples Y1, Y2, Y3, Y4, Y5, Y6, and Y7 are fitted onto the coordinate system to form a fitting plane, which is expressed by the formula: To facilitate the calculation of the above formula, it is proposed to add a supplementary term X0 and set X0=1. Substituting X0 into the above formula, we can integrate it into a new fitting plane formula (i.e., the multiple linear regression equation): ； The integrated formula for fitting the plane is converted into a matrix calculation method. In the above formula, hw(x) represents the fitting plane, i.e., the multiple linear regression equation; Step 3: Iterative training of the multiple linear regression equation model: The historical operation control data samples of X1, X2, X3, X4, X5, X6, and X7 from Step 1, and the corresponding air classifier efficiency sample data of Y1, Y2, Y3, Y4, Y5, Y6, and Y7, are input into the multiple linear regression equation 1 model established in Step 2 in the maximum batch size, and the loss function is used. The mini-batch gradient descent algorithm is used, with a learning rate that approaches an infinitesimally small value. It iteratively updates the historical running data samples of X1, X2, X3, X4, X5, X6, and X7, and the efficiency sample data of Y1, Y2, Y3, Y4, Y5, Y6, and Y7 along the tangent direction of the descent of the iterative training curve, until the lowest point of the iterative training curve is reached. This iterative training continues until a multiple linear regression equation model with a goodness of fit approaching one is obtained. Step four: Calculation of the bias term w0: In the multiple linear regression equation model, for each sample, the true value 3y... i And the predicted value 4w T x (i) There will be an error between them. Using formula In the above formula, i represents the sample matrix, and T represents the vector transpose; error Formula for following a normal distribution , the formula Substitute into the formula The formula is obtained from , the formula Expressed as a loss function Since the loss function formula involves cumulative multiplication, the cumulative multiplication in the above formula is converted into a cumulative summation expression as follows: By formula Simplify to obtain the loss function Then, by taking the partial derivative of the loss function and assuming the partial derivative is zero, the formula for calculating the weights is derived. In the formula for calculating the weight values ​​above, w0 represents the weight value, T represents the vector transpose, x represents the sample matrix of historical operating parameters including X1, X2, X3, X4, X5, X6, and X7, and y represents the sample matrix of classifier efficiency including Y1, Y2, Y3, Y4, Y5, Y6, and Y7; using the formula for calculating the weight values... Substitute the historical operating parameter sample matrices of X1, X2, X3, X4, X5, X6, and X7, and the air classifier efficiency sample matrices of Y1, Y2, Y3, Y4, Y5, Y6, and Y7 into the formula to calculate the value of the w0 bias term; Step 5: Calculation of the weight values ​​w1, w2, w3, w4, w5, w6, and w7 of the multiple linear regression equation: Substitute the w0 bias term value calculated in Step 4 into the fitting plane formula. In the next step, the weight values ​​of w1, w2, w3, w4, w5, w6, and w7 are calculated respectively. Step six: Obtaining the classifier's operating control parameters: Set the ideal efficiency value for the classifier, and substitute the bias term w0 and weight values ​​w1, w2, w3, w4, w5, w6, and w7 calculated in step four into the fitting plane formula. The values ​​of the classifier's operating control parameters X1, X2, X3, X4, X5, X6, and X7, which are well-matched to the ideal efficiency value of the classifier, were calculated respectively, and the scientific and reasonable historical operating control parameters under the proposed ideal classifier efficiency were obtained.

2. The method for optimizing the operating control parameters of an air classifier based on a multiple linear regression algorithm according to claim 1, characterized in that: In step one, the prerequisite for using the historical operating control parameter sample data of classifiers X1, X2, X3, X4, X5, X6, and X7, and the corresponding classifier efficiency data Y1, Y2, Y3, Y4, Y5, Y6, and Y7, is that the feed amount and model of the classifier are the same, and the fineness of the classifier is controlled at 300 mesh.

3. The method for optimizing the operating control parameters of an air classifier based on a multiple linear regression algorithm according to claim 2, characterized in that: In step one, the historical operating control parameter sample data of classifiers X1, X2, X3, X4, X5, X6, and X7, and the corresponding classifier efficiencies Y1, Y2, Y3, Y4, Y5, Y6, and Y7 are stored in the classifier's control system. Each set of historical operating control parameter data corresponds to the efficiency of a classifier; and abnormal historical operating control parameter data are removed.

4. The method for optimizing the operating control parameters of an air classifier based on a multiple linear regression algorithm according to claim 1, characterized in that: In step three, the predicted value is the projection point of the true value sample onto the fitted multiple linear regression equation; error It is the bias term b between the actual value and the predicted value.

5. The method for optimizing the operating control parameters of an air classifier based on a multiple linear regression algorithm according to claim 4, characterized in that: In step three, the error between the true value and the predicted value for each sample... They are independent entities with the same distribution, and follow a mean of 0 and a variance of w. 2 It follows a normal distribution.

Images