Intelligent control method for particle grading of cement grinding

Abstract

The present invention discloses an intelligent control method for particle grading in cement grinding, which relates to the field of cement grinding control technology, including: S1: real-time acquisition of grinding system operating parameters, including mill current, powder selector speed, feed rate, grinding sound signal, and online particle size detection data; S2: calculation of core characteristic parameters based on the operating parameters, including effective crushing work index, dynamic fine powder retention rate, and material filling state index in the mill; S3: inputting the core characteristic parameters into a grading prediction model and outputting a predicted particle size distribution; S4: generating an adjustment instruction through a hierarchical control strategy based on the predicted particle size distribution; S5: outputting the adjustment instruction to an actuator. By real-time acquisition and analysis of grinding system operating parameters, combined with the core characteristic parameters, the particle size distribution is accurately predicted, and closed-loop intelligent control is achieved, thereby improving grinding efficiency, stability, and product quality, while optimizing control accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of cement grinding control, in particular to an intelligent control method for cement grinding particle gradation. Background Art

[0002] Particle grading during cement grinding significantly impacts the performance of the final product. Existing grinding systems often rely on fixed parameters or empirical control methods, lacking the ability to respond to dynamic operating conditions and making it difficult to achieve intelligent control of particle grading. This is especially true when raw material properties fluctuate or equipment loads change, leading to significant adjustment lags, resulting in reduced fine powder recovery, increased energy consumption, and poor product stability.

[0003] At the same time, traditional control systems typically rely on a single indicator, such as current or speed, which cannot fully reflect the complex crushing and grading behavior within the mill. They lack systematic modeling and feedback mechanisms for crushing efficiency, fine powder retention, and material filling status, making it difficult to accurately approximate the target particle size distribution. Therefore, an intelligent particle grading control method that integrates multiple operating parameters, establishes a predictive model, and implements closed-loop control is urgently needed to improve the control accuracy and operating efficiency of cement grinding systems. Summary of the Invention

[0004] Based on the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide an intelligent control method for cement grinding particle grading to solve the above-mentioned technical problems.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligently controlling the particle gradation of cement grinding particles, comprising:

[0006] S1: Real-time collection of grinding system operating parameters, including mill current, classifier speed, feed rate, grinding sound signal and online particle size detection data;

[0007] S2: Calculate core characteristic parameters based on operating parameters, including effective crushing work index, dynamic fine powder retention rate and mill material filling state index;

[0008] S3: Input the core characteristic parameters into the gradation prediction model and output the predicted particle size distribution;

[0009] S4: Generate adjustment instructions through hierarchical control strategy based on predicted particle size distribution;

[0010] S5: Output the adjustment command to the actuator.

[0011] The present invention is further configured such that step S2 specifically includes:

[0012] The effective crushing work index is calculated by integrating the mechanical energy obtained per unit material during the period of integration;

[0013] The dynamic fine powder retention rate is determined by the ratio of the online particle size to the theoretical fine powder amount combined with the speed ratio of the classifier;

[0014] The material filling state index in the mill is represented by the normalized difference between the grinding sound signal and the reference value.

[0015] The present invention is further configured such that the calculation logic of the effective crushing work index is: , is the effective crushing work index, For the current time point, is the time window length, is the current value, is the voltage value, is the motor mechanical efficiency, is the spindle speed of the mill, is the feeding rate, the motor mechanical efficiency The calculation logic is: , The maximum mechanical efficiency of the motor, The optimal working speed of the mill, is the expansion factor;

[0016] The calculation logic of dynamic fine powder retention rate is: , is the dynamic fine powder retention rate, The value measured by online particle size analyzer is less than The proportion of particles, In the theoretical equilibrium state, it is less than The proportion of fine powder The speed of the powder selector, is the reference critical speed value of the powder separator, is the preset threshold;

[0017] The calculation logic of the mill material filling state index is: , is the material filling state index in the mill, is the corrected linear unit function, is the current grinding power, is the base grinding power, is the mill structure sound response coefficient, is the current material density.

[0018] The present invention is further configured such that step S3 specifically includes:

[0019] The first distribution function term reflecting the distribution characteristics of coarse particles is established based on the effective crushing work index;

[0020] A second distribution function reflecting the fine particle enrichment characteristics is established based on the dynamic fine powder retention rate;

[0021] The third distribution function reflecting the coarse particle suppression characteristics is established based on the material filling state index in the mill;

[0022] Performing weighted fusion on the first, second and third distribution function terms and generating predicted particle size distribution through normalization;

[0023] The loss function based on distribution distance is used to optimize the parameters of the gradation prediction model.

[0024] The present invention is further configured to predict the particle size distribution of material particles using a modified Rosin-Rammler distribution function based on the effective crushing work index, and the calculation logic of the first distribution function term is: ,in, is the coarse particle cumulative distribution function, is the particle diameter, is the effective crushing work index, The median particle size, is the distribution index, The calculation logic is: , is the scale factor, is the adjustment index, The calculation logic is: , is the benchmark distribution index, is the regulating factor;

[0025] The fine particle distribution is further corrected based on the dynamic fine powder retention rate. The calculation logic of the second distribution function term is: , is the fine powder correction function, is the dynamic fine powder retention rate, Characteristic particle size of fine powder, is the standard deviation of the distribution;

[0026] The secondary crushing probability of coarse particles is suppressed based on the filling state of the material in the mill. The calculation logic of the third distribution function term is: , is the coarse-grained suppression function, is the material filling state index in the mill, is the attenuation coefficient;

[0027] The first, second, and third distribution function items are weighted and fused, and the final predicted distribution is obtained by normalization through the Softmax function. The calculation logic of the predicted distribution is: , To predict particle size distribution, For the Item weighting function, is the standard Softmax normalization function, 、 and is the weight coefficient;

[0028] The calculation logic of the loss function is: , is the loss function, To predict particle size distribution, is the actual particle size distribution, is the gradient coefficient, To calculate the difference between the predicted particle size distribution and the actual particle size distribution, The calculation logic is: , To be allocated and The amount of transportation between and are the particle diameters of the predicted and actual distributions, respectively;

[0029] Grading prediction model parameter set Including distribution function parameters and weight coefficient ,in, , , parameter set The expression is: ;

[0030] Weight coefficient The update rule is: , For the The weight coefficient vector of the iteration, is the weight learning rate, is the loss function The gradient of the weight coefficient, is the updated weight coefficient vector.

[0031] The present invention is further configured such that step S4 specifically includes:

[0032] Obtain the current predicted particle distribution and the preset target particle distribution;

[0033] Based on the difference between the predicted particle size distribution and the preset target particle size distribution, a regulation function containing multiple sub-control objectives is constructed to obtain the adjustment values ​​of the mill loading force, mill speed and feed rate;

[0034] Output the adjustment instruction vector, the specific form of the adjustment instruction vector is: , To adjust the instruction vector, Adjust the amount of mill loading force, The speed adjustment value of the powder selector, Adjust the amount for feeding quantity.

[0035] The present invention is further configured such that the mill loading force adjustment amount The calculation logic is: , is the adjustment coefficient, is the effective crushing work index, To predict particle size distribution, is the preset target particle size distribution;

[0036] The speed adjustment amount of the powder selector The calculation logic is: , is the proportional coefficient, is the structural similarity index, is the symbolic function, is the median particle size of the target particle distribution, is the median particle size of the current predicted particle size distribution;

[0037] The feeding amount adjustment amount The calculation logic is: , is the sensitivity adjustment coefficient, It is the material filling status index in the mill.

[0038] The present invention is further configured such that step S5 specifically includes:

[0039] Converting the mill loading force adjustment instruction into a pressure control signal and outputting it to the mill pressure execution unit;

[0040] Convert the speed adjustment instruction of the powder classifier into a speed control signal and output it to the powder classifier drive unit;

[0041] The feeding amount adjustment instruction is converted into a flow control signal and output to the material conveying control unit.

[0042] The present invention is further configured such that the calculation logic of the mill loading force adjustment instruction is: , Set the value for the loading force, As the basic loading force, Adjust the amount of mill loading force, is the mill structure coefficient;

[0043] The calculation logic of the powder classifier speed adjustment instruction is: , For the moment The target setting speed of the powder selector is is the speed adjustment value of the powder classifier;

[0044] The calculation logic of the feeding amount adjustment instruction is: , For the moment Feeding speed, is the initial feeding speed, is the response time constant, Adjust the amount for feeding quantity.

[0045] The present invention is further configured to include dynamically adjusting the adjustment instruction based on feedback information of the execution structure:

[0046] Real-time monitoring of the current change rate of the powder separator ,when Greater than the preset threshold When , the loading force compensation is triggered, and the loading force compensation mechanism is: , The compensation amount of the mill loading force adjustment, To adjust the step size coefficient;

[0047] Obtain actual particle size distribution through online particle size analyzer , every T minutes, the gradation prediction model parameters are modified. The gradation prediction model parameter modification logic is: , are the modified gradation prediction model parameters, The difference between the predicted particle size distribution and the actual particle size distribution, is the gradient operator, is the learning rate;

[0048] When the feeding amount is adjusted When the reference grinding power is triggered Adjustment, The threshold value of the number of adjustments is preset within a cycle. The adjustment logic of the reference grinding power is: , is the adjusted reference grinding power, The power adjustment amount of the grinding sound is The calculation logic is: , is the average grinding power of the current cycle, is the smoothing coefficient;

[0049] like , then reject this adjustment and adopt the rollback mechanism to keep the original unchanged, among which, is the long-term mean value of grinding power, is the standard deviation of grinding sound power, and Periodic statistics based on grinding noise signals.

[0050] The present invention provides an intelligent control method for cement grinding particle gradation, which includes the following steps: S1: real-time collection of grinding system operating parameters, including mill current, powder selector speed, feed rate, grinding sound signal, and online particle size detection data; S2: calculation of core characteristic parameters based on the operating parameters, including effective crushing work index, dynamic fine powder retention rate, and mill material filling state index; S3: inputting the core characteristic parameters into a gradation prediction model to output a predicted particle size distribution; S4: generating adjustment instructions based on the predicted particle size distribution through a hierarchical control strategy; and S5: outputting the adjustment instructions to an actuator. The beneficial effects produced include:

[0051] 1. Improve grinding efficiency and stability: By collecting and analyzing the operating parameters of the grinding system in real time and combining multiple core characteristic parameters such as effective crushing work index, dynamic fine powder retention rate and in-mill material filling state index, the particle size distribution can be predicted and adjusted more accurately, thereby significantly improving the efficiency and stability of the grinding process and avoiding the shortcomings of traditional methods that rely too much on experience in the grinding process.

[0052] 2. Optimize particle size distribution prediction: By establishing a multi-level distribution function based on the effective crushing work index, dynamic fine powder retention rate, and in-mill material filling state index, and weighted fusion to generate the particle size distribution, the distribution of various particle size components in the cement grinding process can be accurately predicted, thereby providing a more precise adjustment basis for the control system;

[0053] 3. Realize closed-loop intelligent control: Through real-time feedback and dynamic updating of the prediction model, a closed-loop adjustment is formed under the feedback control of the actuator, which effectively responds to various uncertainties in the grinding process, such as changes in mill loading force, classifier speed and feed rate. The model is optimized and trained through a loss function based on particle size distribution differences, which can continuously improve the accuracy of the prediction model.

[0054] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without inventive efforts. In the drawings:

[0056] Figure 1 The flowchart of the intelligent control method for cement grinding particle grading is shown as an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0057] The following describes the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art will readily appreciate the other advantages and benefits of the present invention from the disclosure herein. The present invention may also be implemented or applied through various other specific embodiments, and the various details in this specification may be modified or altered based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are intended only to illustrate the present invention and are not intended to limit the scope of protection of the present invention.

[0058] It should be noted that the illustrations provided in the following embodiments are merely schematic illustrations of the basic concept of the present invention. Therefore, the illustrations only show components related to the present invention and are not drawn according to the number, shape, and size of components in actual implementation. In actual implementation, the type, quantity, and proportion of each component may be changed arbitrarily, and the component layout may also be more complex.

[0059] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the embodiments of the present invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring the embodiments of the present invention.

[0060] Applied to the intelligent control method of cement grinding particle gradation, such as Figure 1 As shown, including:

[0061] S1: Real-time collection of grinding system operating parameters, including mill current, classifier speed, feed rate, grinding sound signal and online particle size detection data;

[0062] S2: Calculate core characteristic parameters based on operating parameters, including effective crushing work index, dynamic fine powder retention rate and mill material filling state index;

[0063] S3: Input the core characteristic parameters into the gradation prediction model and output the predicted particle size distribution;

[0064] S4: Generate adjustment instructions through hierarchical control strategy based on predicted particle size distribution;

[0065] S5: Output the adjustment command to the actuator.

[0066] The present invention is further configured such that step S2 specifically includes: obtaining an effective crushing work index by calculating the mechanical energy obtained per unit material during an integration period; determining a dynamic fine powder retention rate by combining a ratio of an online particle size to a theoretical fine powder amount with a powder selector speed ratio; and characterizing an in-mill material filling state index by a normalized difference between a grinding sound signal and a reference value. The present invention is further configured such that the calculation logic of the effective crushing work index is: , is the effective crushing work index, For the current time point, is the time window length, is the current value, is the voltage value, is the motor mechanical efficiency, is the spindle speed of the mill, is the feeding rate, the motor mechanical efficiency The calculation logic is: , The maximum mechanical efficiency of the motor, The optimal working speed of the mill, is the expansion factor; the calculation logic of dynamic fine powder retention rate is: , is the dynamic fine powder retention rate, The value measured by online particle size analyzer is less than The proportion of particles, In the theoretical equilibrium state, it is less than The proportion of fine powder The speed of the powder selector, is the reference critical speed value of the powder separator, is the preset threshold; the calculation logic of the mill material filling state index is: , is the material filling state index in the mill, is the corrected linear unit function, is the current grinding power, is the base grinding power, is the mill structure sound response coefficient, is the current material density.

[0067] Specifically, the effective crushing work index It is used to measure the mechanical energy obtained by unit material during the grinding process. The mechanical energy obtained represents the grinding efficiency of the mill. The higher the energy obtained by unit material at each moment, the higher the grinding efficiency of the mill. It is used to reflect the dependence of the motor's mechanical efficiency on the mill speed, and to achieve the highest efficiency near the optimal operating speed; Used to control the sensitivity width of the efficiency function to the speed, the value range is [50,200], the unit is rpm; dynamic fine powder retention rate Reflects the dynamic accumulation of fine particles in the grinding system, and is used to measure whether abnormal accumulation or reduction in separation efficiency occurs during the selection and classification process of fine powder; The threshold value of particle size is preset to characterize the limit of fine powder, with a value range of [20,30] and a unit of nanometers; the filling state index of the material in the mill It is used to dynamically evaluate the filling degree of the material inside the mill cylinder. By analyzing the changes in the grinding sound power generated when the mill is running, combined with the structural characteristics of the mill and the material density, it can be inferred whether there is a non-optimal state such as too much or too little material. The grinding sound power is a physical indicator calculated by filtering and energy extraction after the grinding sound signal obtained by the sound signal acquisition device installed outside the mill. The acquisition method belongs to the existing technology and will not be repeated here. By converting the key physical parameters of effective crushing work, dynamic fine powder retention rate and material filling state in the mill into mathematical expressions, it no longer relies on empirical judgment, and the understanding of the operating state of the grinding system is transformed from qualitative to quantitative, providing a basis for intelligent prediction and control.

[0068] The present invention is further configured such that step S3 specifically includes: establishing a first distribution function term reflecting the coarse particle distribution characteristics based on the effective crushing work index; establishing a second distribution function term reflecting the fine particle enrichment characteristics based on the dynamic fine powder retention rate; establishing a third distribution function term reflecting the coarse particle suppression characteristics based on the in-mill material filling state index; weighted fusion of the first, second, and third distribution function terms, and generating a predicted particle size distribution through normalization; and optimizing and training the gradation prediction model parameters using a loss function based on distribution distance. The present invention is further configured such that a modified Rosin-Rammler distribution function is used to predict the particle size distribution of the material particles based on the effective crushing work index, and the calculation logic of the first distribution function term is: ,in, is the coarse particle cumulative distribution function, is the particle diameter, is the effective crushing work index, The median particle size, is the distribution index, The calculation logic is: , is the scale factor, is the adjustment index, The calculation logic is: , is the benchmark distribution index, is the adjustment factor; the fine particle distribution is further corrected based on the dynamic fine powder retention rate, and the calculation logic of the second distribution function term is: , is the fine powder correction function, is the dynamic fine powder retention rate, Characteristic particle size of fine powder, is the distribution standard deviation; based on the filling state of the material in the mill, the secondary crushing probability of coarse particles is suppressed, and the calculation logic of the third distribution function term is: , is the coarse-grained suppression function, is the material filling state index in the mill, is the attenuation coefficient;

[0069] The first, second, and third distribution function items are weighted and fused, and the final predicted distribution is obtained by normalization through the Softmax function. The calculation logic of the predicted distribution is: , To predict particle size distribution, For the Item weighting function, is the standard Softmax normalization function, 、 and is the weight coefficient; the calculation logic of the loss function is: , is the loss function, To predict particle size distribution, is the actual particle size distribution, is the gradient coefficient, To calculate the difference between the predicted particle size distribution and the actual particle size distribution, The calculation logic is: , To be allocated and The amount of transportation between and The particle diameters of the predicted distribution and the actual distribution are respectively; the gradation prediction model parameter set Including distribution function parameters and weight coefficient ,in, , , parameter set The expression is: ; Weight coefficient The update rule is: , For the The weight coefficient vector of the iteration, is the weight learning rate, is the loss function The gradient of the weight coefficient, is the updated weight coefficient vector;

[0070] Specifically, the first distribution function term is used to predict the cumulative distribution of the coarse particle portion of the material particles; Used to control the location of particle distribution; Used to control the steepness of the distribution; Used to calculate the effective crushing work index The power function is used to map the median particle size to a physically meaningful value, with a value range of [20, 1000]; It is used to control the sensitivity of median particle size to the change of crushing work, and the value range is [0.1, 0.6]; The baseline used to control the steepness of the particle distribution, with a value range of [0.5, 2.5]; Used to control the response of distribution steepness to energy consumption, with a value range of [0.1,1]; 、 、 and The initial fitting parameters in the first distribution function term of the gradation prediction model can be obtained based on the existing technology, specifically including: collecting a certain number of grinding product particle size distribution samples, and combining the effective crushing work index under the corresponding working conditions , the nonlinear least squares method is used to fit the model parameters, which is an existing technology and will not be described in detail here; the second distribution function term is used to correct the distribution of fine powder particle size segments during cement grinding to better reflect the retention of fine particles under actual operating conditions; Used to control the dispersion of fine powder distribution, it can be dynamically updated according to the current particle size distribution data obtained by the online particle size analyzer, for example, the particle size interval span is calculated and estimated based on the distribution morphology characteristics. , with units of nanometers; the third distribution function term reflects the dynamic regulation of the probability of secondary crushing of large particles based on the filling state of the material in the mill. During the grinding process, if coarse particles are in a high filling state, they are easily protected by the cushion effect and are not easily effectively crushed. Therefore, it is necessary to introduce a mechanism to inhibit the secondary crushing of coarse particles. It is used to control the sensitivity of the filling state to coarse particle suppression, with a value range of [0.001, 0.01]. The weighted distribution is normalized by the Softmax function to ensure that the final output is a legal probability distribution and that the sum of all predicted values ​​is 1. 、 and It is used to adjust the contribution of each distribution function term to the predicted particle size distribution, and 、 and The sum is 1, 、 and The specific value is obtained by training historical data. The weight training uses a neural network to construct a loss function based on the collected particle size distribution data and the corresponding mill operating data. By minimizing the difference between the predicted particle size distribution and the actual particle size distribution, the most suitable weight coefficient is optimized. This is the existing technology and will not be described in detail here. By integrating the three distribution function terms, the complex factors of multiple grinding processes can be comprehensively considered to improve the accuracy of particle size prediction. The loss function Used to measure the difference between the predicted and actual particle size distributions, as well as gradient constraints; Used to control the gradient constraint For the loss function The influence of, the value range is [0.0001,0.1]; the weight coefficient Optimization is performed using the gradient descent method. The above weight coefficient update process calculates the loss function after each iteration. And determine whether to terminate the iteration by any of the following convergence conditions: ,in is the preset convergence threshold; It is used to control the adjustment step size of the weight vector in each iteration, with a value range of [0.0001, 0.1]. Accurate particle size distribution prediction can be used for feedback control to optimize parameter settings in the grinding process and achieve more efficient grinding operations.

[0071] The present invention is further configured such that step S4 specifically includes: obtaining a current predicted particle size distribution and a preset target particle size distribution; constructing a regulation function including a plurality of sub-control targets based on the difference between the predicted particle size distribution and the preset target particle size distribution, and obtaining adjustment values ​​for the mill loading force, the mill speed, and the feed rate; and outputting a regulation instruction vector, wherein the specific expression of the regulation instruction vector is: , To adjust the instruction vector, Adjust the amount of mill loading force, The speed adjustment value of the powder selector, The present invention is further configured such that the mill loading force adjustment amount The calculation logic is: , is the adjustment coefficient, is the effective crushing work index, To predict particle size distribution, is the preset target particle size distribution; the speed adjustment amount of the powder selector is The calculation logic is: , is the proportional coefficient, is the structural similarity index, is the symbolic function, is the median particle size of the target particle distribution, is the median particle size of the currently predicted particle size distribution; the feed rate adjustment amount The calculation logic is: , is the sensitivity adjustment coefficient, is the filling state index of the material in the mill; specifically, the purpose of step S4 is to realize intelligent control of the operation condition of the mill, and by comparing the difference between the current particle distribution and the target distribution, dynamically calculate the adjustment instruction, so as to accurately control the particle gradation in the mill; the current predicted particle distribution is predicted by the gradation prediction model, which reflects the particle size gradation under the current mill operation state; the preset target particle distribution is preset according to product requirements or experience curves; the current predicted particle distribution and the preset target particle distribution are obtained to clarify the degree of deviation between the current state and the target, providing a basis for control; the sub-control objectives include gradation shape optimization, overall fineness adjustment and operation stability maintenance; the mill loading force adjustment amount determines the pressure in the mill and affects the crushing strength; the powder selector speed adjustment amount controls the fine powder separation efficiency and adjusts the particle retention time; the feed amount adjustment amount affects the load of the mill and the grinding effect of the material. Too high a feed amount may cause the mill to overload, while too low a feed amount may affect the grinding efficiency; is a divergence function used to measure the information loss between the current predicted particle distribution and the preset target particle distribution. The larger the value, the more significant the difference. Used to control the adjustment amount of mill loading force The adjustment range is [0.001, 0.1]; It is used to determine the response speed of the powder concentrator speed to the difference in particle size distribution, and the value range is [0.1, 2]; The adjustment sensitivity used to control the feeding amount has a value range of [0.01, 0.05]. The above steps can improve the accuracy of particle control and reduce energy consumption and over-grinding during the working process.

[0072] The present invention is further configured such that step S5 specifically includes: converting the mill loading force adjustment instruction into a pressure control signal and outputting it to the mill pressure execution unit; converting the powder classifier speed adjustment instruction into a speed control signal and outputting it to the powder classifier drive unit; converting the feed amount adjustment instruction into a flow control signal and outputting it to the material conveying control unit; the present invention is further configured such that the calculation logic of the mill loading force adjustment instruction is: , Set the value for the loading force, As the basic loading force, Adjust the amount of mill loading force, is the mill structure coefficient; the calculation logic of the powder selector speed adjustment instruction is: , For the moment The target setting speed of the powder selector is is the speed adjustment value of the powder selector; the calculation logic of the feeding amount adjustment instruction is: , For the moment Feeding speed, is the initial feeding speed, is the response time constant, The amount of feed adjustment; specifically, The final target loading force obtained by the control system according to the adjustment amount; The target speed is calculated based on the speed adjustment of the powder classifier; is the target feeding speed obtained by adjusting the feeding amount; Used to adjust the response sensitivity of the mill, the value range is [0.1, 2]; Used to control the response speed of feed rate adjustment, with a value range of [0.1,1]. By calculating the adjustment instructions for mill loading force, classifier speed, and feed rate, the entire cement grinding process can be effectively controlled to optimize grinding efficiency and cement particle size distribution.

[0073] The present invention is further configured to include dynamically adjusting the adjustment instruction based on the feedback information of the execution structure: real-time monitoring of the current change rate of the powder selector ,when Greater than the preset threshold When , the loading force compensation is triggered, and the loading force compensation mechanism is: , The compensation amount of the mill loading force adjustment, To adjust the step size coefficient; obtain the actual particle size distribution through the online particle size analyzer , every T minutes, the gradation prediction model parameters are modified. The gradation prediction model parameter modification logic is: , are the modified gradation prediction model parameters, The difference between the predicted particle size distribution and the actual particle size distribution, is the gradient operator, is the learning rate; when the feeding amount is adjusted When the reference grinding power is triggered Adjustment, The threshold value of the number of adjustments is preset within a cycle. The adjustment logic of the reference grinding power is: , is the adjusted reference grinding power, The power adjustment amount of the grinding sound is The calculation logic is: , is the average grinding power of the current cycle, is the smoothing coefficient;

[0074] like , then reject this adjustment and adopt the rollback mechanism to keep the original unchanged, among which, is the long-term mean value of grinding power, is the standard deviation of grinding sound power, and Based on the periodic statistics of the grinding sound signal; specifically, real-time monitoring of the current change rate of the powder separator , The value reflects the dynamic change of the current of the powder separator. Exceeding the preset threshold When , it indicates that the classifier has load imbalance or system abnormality, and the loading force needs to be adjusted; is a sign function that indicates the direction of the current change rate. When the current change rate is positive, the sign function is 1, indicating an increase in the loading force. When the current change rate is negative, the sign function is -1, indicating a decrease in the loading force. Used to control the size of the compensation amount, the value range is [0.1, 1]; every T minutes, the actual particle size distribution is obtained through the online particle size analyzer , based on its predicted particle size distribution The EMD error between them is calculated by using the gradient descent method according to the learning rate Modified gradation prediction model parameters ; T is the preset time interval parameter, which is used to control the update frequency of the gradation prediction model parameters. The value range is [3,30], and the unit is minutes; in the gradation prediction model parameter modification logic is the gradient operator, which represents the parameter set Distribution function parameters The partial derivative vector of The specific manifestations are: ,but The calculation logic is: ; Used to control the step size of each correction, the value range is [0.00001,0.1]; whenever the number of feed amount adjustments within a unit cycle reaches or exceeds the preset threshold , indicating that the material supply is frequently adjusted, and there is a deviation in material characteristics or operating status. At this time, it is necessary to re-evaluate and appropriately adjust the reference grinding noise power; the adjustment logic of the reference grinding noise power is essentially an exponential weighted average update to avoid drastic jumps; if the updated Deviations from the long-term mean Exceeding two standard deviations means: , the update is considered abnormal, the update is rejected, and the original ; Used to control the update amplitude of the reference grinding noise power, with a value range of [0.05, 0.3]. Based on the dynamic adjustment mechanism of feedback information, when an abnormality occurs, such as the current change rate of the classifier exceeds the threshold, the loading force compensation or the reference grinding noise power adjustment can be carried out in time to ensure that the grinding system can still maintain a good operating state under complex operating conditions and improve the robustness of the grinding system.

[0075] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0076] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0077] In this application, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0078] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0079] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0080] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0081] In the several embodiments provided in this application, it should be understood that the disclosed system can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0082] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0083] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0084] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0085] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. An intelligent control method for particle grading of cement grinding, characterized in that: include: S1: Real-time collection of grinding system operating parameters, including mill current, classifier speed, feed rate, grinding sound signal and online particle size detection data; S2: Calculate core characteristic parameters based on operating parameters. The core characteristic parameters include effective crushing work index, dynamic fine powder retention rate, and mill material filling state index. Step S2 specifically includes: the effective crushing work index is calculated by integrating the mechanical energy obtained per unit material during the integration period; the dynamic fine powder retention rate is determined by the ratio of the online particle size to the theoretical fine powder amount combined with the classifier speed ratio; the mill material filling state index is characterized by the normalized difference between the grinding sound signal and the reference value; the calculation logic of the effective crushing work index is as follows: , is the effective crushing work index, For the current time point, is the time window length, is the current value, is the voltage value, is the motor mechanical efficiency, is the spindle speed of the mill, is the feeding rate, the motor mechanical efficiency The calculation logic is: , The maximum mechanical efficiency of the motor, The optimal working speed of the mill, is the expansion factor; the calculation logic of dynamic fine powder retention rate is: , is the dynamic fine powder retention rate, The value measured by online particle size analyzer is less than The proportion of particles, In the theoretical equilibrium state, it is less than The proportion of fine powder The speed of the powder selector, is the reference critical speed value of the powder separator, is the preset threshold; The calculation logic of the mill material filling state index is: , is the material filling state index in the mill, is the corrected linear unit function, is the current grinding power, is the base grinding power, is the mill structure sound response coefficient, is the current material density; S3: Input the core characteristic parameters into the gradation prediction model and output the predicted particle size distribution; S4: Generate adjustment instructions through hierarchical control strategy based on predicted particle size distribution; S5: Output the adjustment command to the actuator.

2. The intelligent control method for cement grinding particle gradation according to claim 1 is characterized in that: Step S3 specifically includes: The first distribution function term reflecting the distribution characteristics of coarse particles is established based on the effective crushing work index; A second distribution function reflecting the fine particle enrichment characteristics is established based on the dynamic fine powder retention rate; The third distribution function reflecting the coarse particle suppression characteristics is established based on the material filling state index in the mill; Performing weighted fusion on the first, second and third distribution function terms and generating predicted particle size distribution through normalization; The loss function based on distribution distance is used to optimize the parameters of the gradation prediction model.

3. The intelligent control method for cement grinding particle gradation according to claim 2 is characterized in that: The modified Rosin-Rammler distribution function is used to predict the particle size distribution of material particles based on the effective crushing work index. The calculation logic of the first distribution function term is: ,in, is the coarse particle cumulative distribution function, is the particle diameter, is the effective crushing work index, The median particle size, is the distribution index, The calculation logic is: , is the scale factor, is the adjustment index, The calculation logic is: , is the benchmark distribution index, is the regulating factor; The fine particle distribution is further corrected based on the dynamic fine powder retention rate. The calculation logic of the second distribution function term is: , is the fine powder correction function, is the dynamic fine powder retention rate, Characteristic particle size of fine powder, is the standard deviation of the distribution; The secondary crushing probability of coarse particles is suppressed based on the filling state of the material in the mill. The calculation logic of the third distribution function term is: , is the coarse-grained suppression function, is the material filling state index in the mill, is the attenuation coefficient; The first, second, and third distribution function items are weighted and fused, and the final predicted distribution is obtained by normalization through the Softmax function. The calculation logic of the predicted distribution is: , To predict particle size distribution, For the Item weighting function, is the standard Softmax normalization function, 、 and is the weight coefficient; The calculation logic of the loss function is: , is the loss function, To predict particle size distribution, is the actual particle size distribution, is the gradient coefficient, To calculate the difference between the predicted particle size distribution and the actual particle size distribution, The calculation logic is: , To be allocated and The amount of transportation between and are the particle diameters of the predicted and actual distributions, respectively; Grading prediction model parameter set Including distribution function parameters and weight coefficient ,in, , , parameter set The expression is: ; Weight coefficient The update rule is: , For the The weight coefficient vector of the iteration, is the weight learning rate, is the loss function The gradient of the weight coefficient, is the updated weight coefficient vector.

4. The intelligent control method for cement grinding particle gradation according to claim 1 is characterized in that: Step S4 specifically includes: Obtain the current predicted particle distribution and the preset target particle distribution; Based on the difference between the predicted particle size distribution and the preset target particle size distribution, a regulation function containing multiple sub-control objectives is constructed to obtain the adjustment values ​​of the mill loading force, mill speed and feed rate; Output the adjustment instruction vector, the specific form of the adjustment instruction vector is: , To adjust the instruction vector, Adjust the amount of mill loading force, The speed adjustment value of the powder selector, Adjust the amount for feeding quantity.

5. The intelligent control method for cement grinding particle gradation according to claim 4 is characterized in that: The mill loading force adjustment amount The calculation logic is: , is the adjustment coefficient, is the effective crushing work index, To predict particle size distribution, is the preset target particle size distribution; The speed adjustment amount of the powder selector The calculation logic is: , is the proportional coefficient, is the structural similarity index, is the symbolic function, is the median particle size of the target particle distribution, is the median particle size of the current predicted particle size distribution; The feeding amount adjustment amount The calculation logic is: , is the sensitivity adjustment coefficient, It is the material filling status index in the mill.

6. The intelligent control method for cement grinding particle gradation according to claim 4 is characterized in that: Step S5 specifically includes: Converting the mill loading force adjustment instruction into a pressure control signal and outputting it to the mill pressure execution unit; Convert the speed adjustment instruction of the powder classifier into a speed control signal and output it to the powder classifier drive unit; The feeding amount adjustment instruction is converted into a flow control signal and output to the material conveying control unit.

7. The intelligent control method for cement grinding particle gradation according to claim 6 is characterized in that: The calculation logic of the mill loading force adjustment instruction is: , Set the value for the loading force, As the basic loading force, Adjust the amount of mill loading force, is the mill structure coefficient; The calculation logic of the powder classifier speed adjustment instruction is: , For the moment The target setting speed of the powder selector is is the speed adjustment value of the powder classifier; The calculation logic of the feeding amount adjustment instruction is: , For the moment Feeding speed, is the initial feeding speed, is the response time constant, Adjust the amount for feeding quantity.

8. The intelligent control method for cement grinding particle gradation according to claim 7 is characterized in that: It also includes dynamic adjustment instructions based on feedback information from the execution structure: Real-time monitoring of the current change rate of the powder separator ,when Greater than the preset threshold When , the loading force compensation is triggered, and the loading force compensation mechanism is: , The compensation amount of the mill loading force adjustment, To adjust the step size coefficient; Obtain actual particle size distribution through online particle size analyzer , every T minutes, the gradation prediction model parameters are modified. The gradation prediction model parameter modification logic is: , are the modified gradation prediction model parameters, The difference between the predicted particle size distribution and the actual particle size distribution, is the gradient operator, is the learning rate; When the feeding amount is adjusted When the reference grinding power is triggered Adjustment, The threshold value of the number of adjustments is preset within a cycle. The adjustment logic of the reference grinding power is: , is the adjusted reference grinding power, The power adjustment amount of the grinding sound is The calculation logic is: , is the average grinding power of the current cycle, is the smoothing coefficient; like , then reject this adjustment and adopt the rollback mechanism to keep the original unchanged, among which, is the long-term mean value of grinding power, is the standard deviation of grinding sound power, and Periodic statistics based on grinding noise signals.

Images