A numerical control vertical mill intelligent control system based on parameter self-optimization

Abstract

The application discloses a numerical control vertical mill intelligent control system based on parameter self-optimization and relates to the technical field of industrial grinding. The system collects the mill roller data of the vertical roller mill through the sensor group arranged in real time and transmits the mill roller data to the numerical control vertical mill intelligent control system for preprocessing to construct the stress coupling data group and the structure response data group. The loading heterogeneity index η load is calculated The loading heterogeneity demarcation threshold T η is calculated The disturbance return tendency index β rb is calculated when the loading heterogeneity exists The loading pressure reconstruction index P yq is calculated The preset first pressure control critical threshold T P1 and the second pressure control critical threshold T P2 are calculated The rebound failure risk is evaluated, adjustment instructions are generated according to the evaluation results, and if the benefit standards are not reached in three consecutive adjustment rounds, mechanical failure is automatically determined and a warning is generated. The system can realize the whole-process intelligent closed loop of loading heterogeneity state recognition, disturbance tendency judgment and risk response control.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial grinding, in particular to a numerical control vertical mill intelligent control system based on parameter self-optimization. BACKGROUND

[0002] In the industrial grinding application scenario, the numerical control vertical mill as the core equipment, its running state directly affects the crushing efficiency, energy consumption control and structural stability. In recent years, with the concept of parameter self-optimization of numerical control vertical mill, the system operation is no longer dependent on fixed control logic, but through real-time collection of mill roller running data, dynamic identification of system state and adjustment of control parameters, realizing the technical leap from static control to dynamic self-adaptation. However, on this basis, further entering the "intelligent control" stage still faces key problems, among which the non-linear dynamic problem represented by the local heterogeneous behavior in the loading process of the mill roller and the structural rebound interference phenomenon caused thereby is the most complex. The stress distribution of the mill roller in the loading process often produces heterogeneity due to uneven material distribution, mechanical wear or operation deviation, etc., and then induces the local disturbance return phenomenon under the coupling of thermal vibration. The composite state of such loading heterogeneity and rebound interference will directly interfere with the control system judgment, resulting in failure of the regulation and control strategy.

[0003] Most of the current numerical control vertical mill systems have certain degree of online monitoring capability, but their intelligent identification mechanism often stops at the mechanical overrun alarm level, lacking the identification ability of complex loading heterogeneous structure and the precise modeling means of subsequent rebound interference. In actual production process, the traditional control strategy usually only judges the system stability based on average pressure or vibration level, and is not easy to identify the regional loss behavior caused by asymmetric stress and multi-source response interference. In addition, the existing system relies on manual intervention and experience parameter adjustment for the judgment of the abnormal state of the mill roller, which not only lags in identification, but also may cause repeated regulation and control failure due to subjective misjudgment, increasing the frequency of equipment downtime. Lack of systematic heterogeneous identification mechanism and interference trend evaluation means makes the mill roller control system difficult to realize real self-optimization regulation and control when facing complex nonlinear state. SUMMARY

[0004] In view of the shortcomings of the prior art, the present application provides a numerical control vertical mill intelligent control system based on parameter self-optimization, which solves the problems in the background art.

[0005] To achieve the above purpose, the present application realizes the following technical scheme: a numerical control vertical mill intelligent control system based on parameter self-optimization, comprising a data acquisition module, a heterogeneous identification module, a reconstruction control module and a control execution module.

[0006] The data acquisition module is used for acquiring the data of the grinding roller of the vertical roller mill in real time according to the sensor group arranged at different positions of the numerical control vertical mill equipment, and transmitting the data to the numerical control vertical mill intelligent control system in real time to obtain a stress coupling data group and a structure response data group after preprocessing;

[0007] The heterogeneous recognition module is used for calculating a loading heterogeneity index η load according to the stress coupling data group, performing loading heterogeneity evaluation with a preset loading heterogeneity threshold T η , and executing a rebound interference analysis instruction when the evaluation result is that the grinding roller has loading heterogeneity;

[0008] The reconstruction control module is used for executing the rebound interference analysis instruction, calculating a disturbance return tendency index β rb according to the structure response data group, performing correlation calculation with the loading heterogeneity index η load to obtain a loading pressure reconstruction index P yq , and performing rebound failure risk evaluation with a preset first pressure control threshold T P1 and a second pressure control threshold T P2 ;

[0009] The control execution module is used for receiving the control instruction after the rebound failure risk evaluation in real time, and executing optimized adjustment according to the control instruction, and generating mechanical fault information when the optimized adjustment does not reach the efficiency standard for three times in succession.

[0010] Preferably, the data acquisition module comprises a data acquisition unit and a data processing unit;

[0011] The data acquisition unit is used for acquiring the data of the grinding roller of the vertical mill in real time according to the sensor group arranged at different positions of the numerical control vertical mill equipment;

[0012] The sensor group comprises a stress sensor, a symmetrical single-axis accelerometer, a current transformer, a three-axis acceleration sensor, a thermocouple and a differential pressure sensor;

[0013] The stress sensor is used for being installed below the contact surface of the grinding roller and in the bearing seat area, and acquiring the normal stress y l of the contact area of the grinding roller in real time;

[0014] The symmetrical single-axis accelerometer is used for being installed at symmetrical positions of the two grinding roller arm support structures respectively, and acquiring the left-right vibration difference a d of the grinding roller in real time;

[0015] The current transformer is used for being arranged at the front end of the connection between the main drive motor incoming line loop and the PLC, and acquiring the load current I r of the main drive motor in real time;

[0016] The triaxial acceleration sensor is arranged vertically above the side wall of the mill base and near the central axis of the inner wall of the mill cavity shell to collect the main frequency f of the regional vibration in real time v ;

[0017] The thermocouple is arranged at the air inlet port of the mill cavity top return air duct to collect the local temperature T of the mill cavity in real time m ;

[0018] The differential pressure sensor is installed on the upper air extraction pipeline of the mill cavity to collect the negative pressure P of the mill cavity region in real time v .

[0019] Preferably, the data processing unit is used to establish a communication connection between the sensor group and the intelligent control system of the numerical control vertical mill according to a wireless network, and the collected mill roller data is transmitted to the intelligent control system of the numerical control vertical mill in real time for preprocessing to obtain the stress coupling data group and the structure response data group;

[0020] The preprocessing includes denoising, missing value filling and dimensionless processing;

[0021] The denoising is a noise suppression processing of the mill roller data by a multi-dimensional filtering technology to eliminate the noise influence in the data, the missing value filling is a complete processing of the mill roller data with sampling interruption and communication packet loss by a missing value interpolation technology, and the dimensionless processing is a dimension influence removal of the mill roller data by a Max-Min maximum minimization method;

[0022] The stress coupling data group includes the normal stress y of the mill roller contact area l , the left and right vibration difference a of the mill roller d and the load current I of the main drive motor r ;

[0023] The structure response data group includes the main frequency f of the regional vibration v , the local temperature T of the mill cavity m and the negative pressure P of the mill cavity region v .

[0024] Preferably, the heterogeneous recognition module includes a heterogeneous recognition unit and a loaded heterogeneous evaluation unit;

[0025] The heterogeneous recognition unit is used to perform associated calculation according to the parameters in the stress coupling data group, analyze the local imbalance state of the loaded structure, and determine the loading heterogeneity index η of each section of the vertical roller mill load , specifically: , wherein, ln represents a logarithmic function, n represents the total number of mill roller segments, a d，i , I r，i and y l，i represent the normal stress y of the mill roller contact area of the i-th section lThe difference in left and right vibration of the grinding roller, a d and the load current I of the main drive motor r ε represents the zero perturbation prevention term, with a value of 0.001.

[0026] Preferably, the loading heterogeneity evaluation unit is used to calculate the loading heterogeneity index η over a six-month period using statistical methods. load mean and standard deviation and based on the mean and standard deviation Preset loading heterogeneous boundary threshold T η Specifically: Then, compared with the real-time acquired loading heterogeneity index η load A heterogeneous loading assessment was conducted, and the specific assessment scheme is as follows;

[0027] When the heterogeneity index η is loaded load <Loading heterogeneous boundary threshold T> η When the load distribution on the grinding roller is uniform, the normal operating procedure should be maintained, and normal monitoring should continue.

[0028] When the heterogeneity index η is loaded load ≥ Loading heterogeneous boundary threshold T η When this occurs, it indicates that there is a loading anomaly in the grinding roller, and at this time, the rebound interference analysis command is executed.

[0029] Preferably, the reconstruction control module includes an interference removal unit and a reconstruction analysis unit;

[0030] The interference removal unit is used to execute a rebound interference analysis command when the loading heterogeneity assessment indicates that the grinding roller has loading heterogeneity;

[0031] The bounce interference analysis command is used to perform correlation calculations based on the parameters within the structural response data set, analyze the disturbance return state of each loaded heterogeneous section of the vertical roller mill, and determine the disturbance return trend index β of each loaded heterogeneous section of the vertical roller mill. rb This reflects the transient interference surge characteristics caused by multi-source response coupling anomalies, specifically: In the formula, n represents the total number of segments of the grinding roller, and f v，i T m，i and P v，i The dominant frequency f of the regional vibration in the i-th segment is represented by f. v Local temperature T in the grinding cavity m Negative pressure P in the grinding cavity area v ε represents the zero perturbation prevention term, with a value of 0.001.

[0032] Preferably, the reconstruction analysis unit includes a bounce interference analysis unit and a risk assessment unit;

[0033] The rebound interference analysis unit is used to obtain a loading pressure reconstruction index P according to a loading heterogeneity index η load and a disturbance return tendency index β rb , perform correlation calculation, and obtain a loading pressure reconstruction index P yq , and comprehensively analyze the rebound interference risk of the region, specifically as follows: ; in the formula, P base represents a standard pressure value set by the system, tanh represents a hyperbolic tangent function, a represents a control sensitivity factor, which is set to 1.2 according to debugging experience, represents the maximum disturbance return tendency index in the entire region, represents the average value of the loading heterogeneity index in the entire region, and ε represents a zero perturbation prevention term, which is 0.001.

[0034] Preferably, the risk assessment unit is used to set a first pressure control critical threshold T base and a second pressure control critical threshold T P1 according to the standard pressure value P P2 set by the system, specifically as follows: T P1 = 0.9*P base , and T P2 = 0.97*P base , and then perform rebound failure risk assessment on the loading pressure reconstruction index P yq obtained in real time, and the specific assessment scheme is as follows:

[0035] When the loading pressure reconstruction index P yq is less than the first pressure control critical threshold T P1 , it indicates that there is a rebound and failure loading risk in the grinding roller contact area, and at this time, the first adjustment instruction is executed.

[0036] When the first pressure control critical threshold T P1 is less than the loading pressure reconstruction index P yq and the loading pressure reconstruction index P yq is less than the second pressure control critical threshold T P2 , it indicates that there is a rebound interference trend, and at this time, the second adjustment instruction is executed.

[0037] When the loading pressure reconstruction index P yq is greater than or equal to the second pressure control critical threshold T P2 , it indicates that there is no rebound interference effect, and the loading heterogeneity is caused by feed accumulation, and at this time, the third adjustment instruction is executed.

[0038] Preferably, the control execution module includes an instruction execution unit and an optimization feedback unit.

[0039] The instruction execution unit is used to receive the adjustment instruction after the rebound failure risk assessment in real time, and execute optimization adjustment according to the adjustment instruction, specifically as follows:

[0040] The first adjustment instruction: generates risk information transmission to the intelligent control system of the numerical control vertical mill, controls the vertical roller mill to stop material feeding, and reduces the roller pressure of the 15% mispressure section by 15% and reduces the hot air inlet wind speed by 10%, and then iteratively evaluates and adjusts through the data acquisition module until the roller load distribution is uniform;

[0041] The second adjustment instruction: generates interference information transmission to the intelligent control system of the numerical control vertical mill, controls the vertical roller mill to reduce the roller pressure of the 8% mispressure section by 8%, and iteratively evaluates and adjusts through the data acquisition module until the roller load distribution is uniform;

[0042] The third adjustment instruction: generates accumulation information transmission to the intelligent control system of the numerical control vertical mill, controls the vertical roller mill to close the abnormal area material feeding, and increases the feeding rate of the adjacent healthy section by 7%, and then iteratively evaluates and adjusts through the data acquisition module until the roller load distribution is uniform.

[0043] Preferably, the optimization feedback unit is used for sampling and archiving the rebound failure risk evaluation result of each time, and comparing the current iterative evaluation result with the previous iterative evaluation result, if the iterative evaluation result is not reached for three times, it means that there is a mechanical problem, at this time, the vertical roller mill is stopped through the intelligent control system of the numerical control vertical mill, and the mechanical fault information is generated and transmitted to the relevant personnel, reminding the equipment maintenance and repair;

[0044] The benefit standard includes that the particle size fluctuation decreases by more than 10%, the unit energy consumption decreases by more than 5%, and the roller surface temperature rise and vibration amplitude decreases by more than 8%.

[0045] The present application provides a numerical control vertical mill intelligent control system based on parameter self-optimization.

[0046] (1) The data acquisition module of the system acquires the roller data of the vertical roller mill in real time according to the stress sensors, symmetrical single-axis accelerometers, current transformers, three-axis acceleration sensors, thermocouples and differential pressure sensors arranged at different positions of the numerical control vertical mill equipment, and transmits the acquired roller data to the intelligent control system of the numerical control vertical mill for pre-noise removal, missing data filling and Max-Min normalization preprocessing, and constructs complementary stress coupling data sets and structure response data sets. This module not only realizes full-dimensional digital mapping of multiple physical signals, but also provides a cleaned physical state baseline for high-order correlation analysis for subsequent modules, significantly improving the accuracy and timeliness of system intelligent judgment.

[0047] (2) The heterogeneous recognition module of the system performs segmented analysis on the stress coupling data, and constructs a loading heterogeneity index η load, to reveal the mechanical imbalance trend between multi-zone loading states, and introduce a loading heterogeneity threshold T η With the loading heterogeneity index η load Perform loading heterogeneity evaluation to form a preliminary judgment mechanism for triggering interference analysis. If there is loading heterogeneity, enter the reconstruction control module to execute the rebound interference analysis instruction, and construct the disturbance return trend index β rb With the loading heterogeneity index η load Jointly generate the loading pressure reconstruction index P yq As a comprehensive characteristic variable driven by multiple source anomalies. This index is normalized and response suppressed by the hyperbolic tangent function tanh, and finally performs rebound failure risk assessment with the preset first pressure control critical threshold T P1 And the second pressure control critical threshold T P2 To form three types of physical state judgments: rebound failure risk area, interference trend area and non-interference accumulation area. This module realizes a multi-level reasoning logic chain from data features, index construction, level division and trend reconstruction, so that the originally implicit structural anomaly state is monitored and quantitatively identified in real time.

[0048] (3) The system control execution module generates three-level control instructions with partition accuracy and action executability based on the rebound failure risk assessment results, including pressure dynamic adjustment, hot air parameter adjustment and feed control optimization. After each round of control, the system introduces an optimization feedback mechanism to record and compare the change trend of benefit parameters. If there is no improvement for three consecutive iterations, it is determined that there is a mechanical problem, and the system actively triggers shutdown and generates a fault warning to improve safety and reliability. This module strengthens the closed-loop capability of the system, so that control is no longer a static threshold response, but forms a dynamic adjustment closed-loop path of identification, control, feedback and reevaluation, with the ability of self-learning, self-correction and self-evolution, and realizes fine management and intelligent fault tolerance of the vertical mill equipment under complex working conditions. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 A flowchart of the intelligent control system of the numerical control vertical mill based on parameter self-optimization of the present application;

[0050] Figure 2 A running principle diagram of the intelligent control system of the numerical control vertical mill based on parameter self-optimization of the present application. DETAILED DESCRIPTION

[0051] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present application.

[0052] Embodiment 1

[0053] Please refer to Figure 1 The present application provides a numerical control vertical mill intelligent control system based on parameter self-optimization. To achieve the above purpose, the present application is implemented by the following technical solutions: comprising a data acquisition module, a heterogeneous identification module, a reconstruction control module and a control execution module.

[0054] The data acquisition module is used for acquiring the mill roller data of the vertical roller mill in real time according to the sensor group arranged at different parts of the numerical control vertical mill equipment, and transmitting the mill roller data to the numerical control vertical mill intelligent control system in real time for preprocessing to obtain a stress coupling data group and a structure response data group.

[0055] The heterogeneous identification module is used for calculating and obtaining a loading heterogeneity index η load according to the stress coupling data group, performing loading heterogeneity evaluation with a preset loading heterogeneity threshold T η , and executing a rebound interference analysis instruction when the evaluation shows that the mill roller exists loading heterogeneity.

[0056] The reconstruction control module is used for executing the rebound interference analysis instruction, calculating a disturbance return trend index β rb according to the structure response data group, performing associated calculation with the loading heterogeneity index η load to obtain a loading pressure reconstruction index P yq , and performing rebound failure risk evaluation with a preset first pressure control critical threshold T P1 and a second pressure control critical threshold T P2 .

[0057] The control execution module is used for receiving the control instruction after the rebound failure risk evaluation in real time, and executing optimized adjustment according to the control instruction. When the optimized adjustment does not reach the efficiency standard for three times in succession, mechanical failure information is generated.

[0058] In the embodiment, the data acquisition module acquires the mill roller data of the vertical roller mill in real time through the sensor group arranged at the key structure parts, and transmits the mill roller data to the numerical control vertical mill intelligent control system for preprocessing to obtain the stress coupling data group and the structure response data group, thereby providing a high-quality data basis for subsequent identification and modeling. The heterogeneous identification module performs associated calculation of the loading heterogeneity index η load, for quantifying the synergy and consistency of the contact pressure, current load and vibration amplitude between each area of the grinding roller, and comparing with the preset loading heterogeneous boundary threshold T η Loading heterogeneous evaluation is performed, and when it is identified that the grinding roller has loading heterogeneity, the disturbance analysis instruction is started, and the trend analysis process is entered. The reconstruction control module performs associated calculation of disturbance return trend index β rb for analyzing whether there is a rebound enhancement trend in vibration frequency, heat accumulation and gas pressure environment after structural abnormal coupling, and comparing with the loading heterogeneous index η load Associated calculation of loading pressure reconstruction index P yq is performed, which comprehensively reflects the double coupling risks of loading abnormalities and rebound disturbances, and compares with the preset first pressure control critical threshold T P1 and the second pressure control critical threshold T P2 Rebound failure risk assessment is performed to form risk level division. The module uses the hyperbolic tangent function for normalized control to avoid physical extreme value out of control, and realizes the coupling judgment of trend prediction and control feasibility. Compared with the traditional empirical pressure control, the module realizes the fine modeling of pressure regulation logic and process dynamic feedback closed loop, greatly improving the timeliness of operation response and the resolution ability of pressure disturbance. The control execution module generates risk pressure regulation, disturbance area pressure fine adjustment and accumulation area feed control control instructions according to the rebound failure risk assessment results, and collects feedback data after each round of adjustment to judge whether the control effect meets the benefit standard. If the standard is not met for three consecutive rounds, it is judged that there is mechanical failure, and the system is automatically stopped and maintenance reminder is generated, realizing the whole process intelligent control from control strategy generation to execution feedback to fault closed loop identification. Compared with the traditional one-way instruction execution and manual experience intervention control mode, the module has the ability of self-evaluation, self-correction and self-evolution, which not only reduces the operation volatility and misoperation risk, but also improves the safety and stability of the system and the service life. It is the key support for the number control vertical mill to realize high intelligence and autonomous management.

[0059] Embodiment 2

[0060] Please refer to Figure 1 and Figure 2 , in particular: the data acquisition module comprises a data acquisition unit and a data processing unit;

[0061] The data acquisition unit is used to acquire the grinding roller data of the vertical mill in real time according to the sensor group arranged at different parts of the number control vertical mill equipment;

[0062] The sensor group comprises a stress sensor, a symmetrical single-axis accelerometer, a current transformer, a three-axis acceleration sensor, a thermocouple and a differential pressure sensor;

[0063] The stress sensor is used to be installed below the contact surface of the grinding roller and in the bearing seat area, and is used to collect the normal stress y of the contact area of the grinding roller in real time l , which represents the direct contact pressure between each section and the material.

[0064] The symmetric single-axis accelerometer is used to be installed at the symmetric positions of the two grinding roller arm support structures respectively, and is used to collect the left-right vibration difference a of the grinding roller in real time d , which represents the difference in vibration intensity of the left-right structure of the grinding roller and reflects the force deviation.

[0065] The current transformer is used to be arranged at the front end of the PLC connected to the main drive motor incoming line loop, and is used to collect the load current I of the main drive motor in real time r , which reflects the real-time performance of the material grinding resistance in the area and is a direct indicator of power consumption and load.

[0066] The three-axis acceleration sensor is used to be vertically arranged above the side wall of the grinding disc base and near the central axis of the inner wall of the grinding cavity shell, and is used to collect the main frequency f of the area vibration in real time v , which represents the main vibration frequency of the material and structure in the grinding cavity in the running state of the grinding roller and reflects the fluctuation trend of the air pressure frequency.

[0067] The thermocouple is used to be arranged at the inlet port of the return air duct at the top of the grinding cavity, and is used to collect the local temperature T of the grinding cavity in real time m , which reflects the degree of transient heat accumulation in the grinding cavity area and expresses the softening degree of the elastic modulus area.

[0068] The differential pressure sensor is used to be installed in the upper air extraction pipeline of the grinding cavity, and is used to collect the negative pressure P of the grinding cavity area in real time v , which represents the gas-solid negative pressure state in the grinding cavity and affects the airflow impact and powder suspension.

[0069] The data processing unit is used to establish a communication connection between the sensor group and the intelligent control system of the numerical control vertical mill according to the wireless network, and is used to transmit the collected grinding roller data to the intelligent control system of the numerical control vertical mill for preprocessing, to obtain the force coupling data group and the structure response data group.

[0070] The preprocessing includes denoising, missing value filling and dimensionless processing.

[0071] The denoising is performed by multi-dimensional filtering technology to suppress noise in the grinding roller data and eliminate noise influence in the data, the missing value filling is performed by missing value interpolation technology to complete the grinding roller data with sampling interruption and communication packet loss, and the dimensionless processing is performed by the Max-Min maximum and minimum method to remove the dimension influence of the grinding roller data.

[0072] The force coupling data group includes the normal stress y of the contact area of the grinding roller l , and the left-right vibration difference a of the grinding roller dand the main drive motor load current I r ;

[0073] The structure response data set includes the area vibration main frequency f v , the local temperature T m of the grinding cavity, and the local negative pressure P v of the grinding cavity.

[0074] In this embodiment, the data acquisition module uses stress sensors, symmetric single-axis accelerometers, current transformers, three-axis acceleration sensors, thermocouples, and differential pressure sensors distributed at different positions of the numerical control vertical mill to respectively collect the normal stress y l of the grinding roller contact area, the left-right vibration difference a d of the grinding roller, the main drive motor load current I r , the area vibration main frequency f v , the local temperature T m of the grinding cavity, and the local negative pressure P v of the grinding cavity, and transmits them to the numerical control vertical mill intelligent control system through a wireless network for preprocessing, thereby constructing the force coupling data set and the structure response data set. This method not only realizes high-precision and wide-coverage perception of the running state of the numerical control vertical mill, but also ensures the consistency and analysis availability of the data structure, significantly improves the response speed and discrimination accuracy of the subsequent state identification, trend modeling, and intelligent control links, lays a solid data foundation for realizing system-level intelligent closed-loop control, and has obvious improvement effect in adaptability in load evaluation, structure stability analysis, and pressure control compared with traditional single monitoring means.

[0075] Embodiment 3

[0076] Please refer to Figure 1 and Figure 2 , specifically: the heterogeneous identification module includes a heterogeneous identification unit and a loading heterogeneous evaluation unit.

[0077] The heterogeneous identification unit is used to perform associated calculation according to the parameters in the force coupling data set, analyze the local imbalance state of the loading structure, and determine the loading heterogeneity index η load of each section of the vertical roller mill, specifically: , wherein, ln represents the logarithmic function, n represents the total number of grinding roller segments, a d，i , I r，i , and y l，i represent the normal stress y l of the grinding roller contact area, the left-right vibration difference a d of the grinding roller, and the main drive motor load current I r of the i-th section, respectively, and the value is 0.001, It reflects the degree of inconsistency between the loading torque offset and resistance in the region and the current actual pressure. The average of n regions is taken to reflect the balance of the overall loading structure of the system. The outermost layer adds a logarithmic function ln to compress the exponential growth trend into an exponential growth response, capturing the surge point of loading heterogeneity. Specific implementation examples are shown in Table 1.

[0078] Table 1: Loading Heterogeneity Index η load Calculation Example Table;

[0079]

[0080] The loading heterogeneity evaluation unit is used to calculate the loading heterogeneity index η over a six-month period based on statistical methods. load mean and standard deviation and based on the mean and standard deviation Preset loading heterogeneous boundary threshold T η Specifically: Then, compared with the real-time acquired loading heterogeneity index η load A heterogeneous loading assessment was conducted, and the specific assessment scheme is as follows;

[0081] When the heterogeneity index η is loaded load <Loading heterogeneous boundary threshold T> η When the load distribution on the grinding roller is uniform, the normal operating procedure should be maintained, and normal monitoring should continue.

[0082] When the heterogeneity index η is loaded load ≥ Loading heterogeneous boundary threshold T η When this occurs, it indicates that there is a loading anomaly in the grinding roller, and at this time, the rebound interference analysis command is executed.

[0083] In this embodiment, the heterogeneous identification unit is based on the normal stress y in the contact area of ​​the grinding roller in the force coupling data set. l The difference in left and right vibration of the grinding roller, a d and the load current I of the main drive motor r Construct the loading heterogeneity index η load It is used to quantify the phenomenon of uneven local loading.

[0084] The purpose of the formula is to reveal whether there is a loading heterogeneity phenomenon caused by mechanical abnormalities in each section of the vertical roller mill through the synergistic analysis of multi-source stress indicators. Used to measure the left-right vibration difference a of the grinding roller d With the load current I of the main drive motor r Multiplication creates an amplified term, capturing the coupling phenomenon of increased load caused by structural offset. The denominator is the normal stress y in the grinding roller contact area. l, which represents the actual pressure capacity of the current grinding roller, is squared to strengthen the extreme value and make the influence of the severely distorted structure more explicit, and the overall physical balance degree of whether the load and the structure deviation strength match the pressure intensity is reflected; The average of n segments reflects the overall loading balance of the system, forming a transition from the regional level to the overall structure level. The outermost logarithmic function ln is used to scale the compression of the explosive trend, avoiding the uncontrollable response caused by the expansion of extreme values, while retaining the continuity of the growth trend.

[0085] The loading heterogeneity evaluation unit generates the loading heterogeneity threshold T load by statistically calculating the mean and standard deviation of the loading heterogeneity index η η in the past six months, and evaluates the loading heterogeneity by using the real-time loading heterogeneity index η load to determine whether it is in a balanced loading state. If it is identified that the grinding roller has loading heterogeneity, the rebound interference analysis instruction will be triggered. This module realizes an adaptive loading abnormality identification mechanism based on structural mechanics response, avoiding the limitations of traditional fixed threshold or manual judgment, improving the system's sensitivity to abnormal states and the accuracy of pre-identification, significantly reducing the risk of structural overpressure, abnormal wear, and energy consumption fluctuations, and providing a high-reliability trigger basis for subsequent intelligent control. It is a key link to realize the intelligent and self-optimizing operation of numerical control vertical grinding.

[0086] Embodiment 4

[0087] Please refer to Figure 1 and Figure 2 , in detail: the reconstruction control module includes an interference elimination unit and a reconstruction analysis unit;

[0088] The interference elimination unit is used to execute the rebound interference analysis instruction when the loading heterogeneity evaluation indicates that the grinding roller has loading heterogeneity;

[0089] The rebound interference analysis instruction is used to perform correlation calculation based on the parameters in the structural response data set to analyze the disturbance return state of each loading heterogeneity section of the vertical roller mill, so as to determine the disturbance return trend index β rb of each loading heterogeneity section of the vertical roller mill, which reflects the transient disturbance return characteristics caused by multi-source response coupling abnormalities. Specifically: , where n represents the total number of grinding roller segments, f v，i , T m，i , and P v，i represent the regional vibration main frequency f v , the local temperature T m of the grinding cavity, and the negative pressure P v of the grinding cavity, respectively, and ε represents the zero disturbance term, which is 0.001, represents the rebound condition formed under the condition of low negative pressure, The rebound enhancement characteristics of the i-th section structure vibration frequency reaching the critical resonance state are analyzed, the product of the two is squared to strengthen the influence of the rebound severe area, the overall square root is taken and averaged to generate the overall rebound trend factor, and the specific implementation cases are shown in Table 2.

[0090] Table 2: disturbance return trend index β rb Operation example table;

[0091]

[0092] The reconstruction analysis unit includes a rebound interference analysis unit and a risk assessment unit;

[0093] The rebound interference analysis unit is used to perform correlation calculation according to the loading heterogeneity index η load and the disturbance return trend index β rb to obtain the loading pressure reconstruction index P yq , which comprehensively analyzes the rebound interference risk in the region, specifically: ; in the formula, P base represents the standard pressure value set by the system, tanh represents the hyperbolic tangent function, α represents the control sensitivity factor, which is set to 1.2 according to debugging experience, represents the maximum disturbance return trend index in the whole region, represents the average value of the loading heterogeneity index in the whole region, and ε represents the zero perturbation term to be prevented, which is 0.001, the numerator term β rb * η load is used to jointly analyze the double superimposed risk of structural abnormalities and powder rebound, and the denominator term is used to normalize the abnormal product to ensure a reasonable control interval, and the hyperbolic tangent function tanh is used to smooth the pressure suppression curve to avoid sudden response, and specific implementation cases are shown in Table 3.

[0094] Table 3: loading pressure reconstruction index P yq Operation example table;

[0095]

[0096] The risk assessment unit is used to set a first pressure control critical threshold T base and a second pressure control critical threshold T P1 according to the standard pressure value P P2 set by the system, specifically: T P1 = 0.9*P base , T P2 = 0.97*P base , and then combined with the real-time obtained loading pressure reconstruction index P yqThe rebound failure risk assessment is performed, and the specific assessment scheme is as follows:

[0097] When the loading pressure reconstruction index P yq ≤ the first pressure control critical threshold T P1 , it indicates that there is a rebound and failure loading risk in the contact area of the grinding roller, and at this time the first adjustment instruction is executed;

[0098] When the first pressure control critical threshold T P1 < loading pressure reconstruction index P yq < second pressure control critical threshold T P2 , it indicates that there is a rebound interference trend, and at this time the second adjustment instruction is executed;

[0099] When the loading pressure reconstruction index P yq ≥ the second pressure control critical threshold T P2 , it indicates that there is no rebound interference effect, and the loading heterogeneity is caused by feed accumulation, and at this time the third adjustment instruction is executed.

[0100] In this embodiment, after identifying that the grinding roller has loading heterogeneity, the disturbance rejection unit calls the area vibration main frequency f v , the local temperature T m of the grinding cavity and the negative pressure P v of the grinding cavity region in the structure response data set to construct the disturbance return trend index β rb , so as to capture the resonance return trend induced under low negative pressure and high temperature accumulation, and depict the local transient disturbance risk caused by multi-source response coupling. The formula is used to measure the influence of the three physical phenomena of vibration main frequency, heat accumulation and negative pressure fluctuation on triggering local resonance, thermal excitation and dust suspension coupling feedback non-steady state in each structure segment.f v，i Reflects the main vibration frequency of the structure in this segment, and frequency anomaly usually indicates structure loosening, load unevenness and driving frequency coupling phenomenon, Reflects the coupling characteristics of heat accumulation and negative pressure fluctuation, and the product square and average are used to amplify the influence of local severe disturbance. The formula essentially constructs a non-linear energy concentration trend function, and the product of vibration frequency and thermal pressure anomaly expresses the rebound release trend, the square amplifies the local disturbance of large amplitude, and the square root average value processing effectively avoids accidental extreme value misjudgment, and takes into account the coordination of structure response in multiple segments.

[0101] The reconstruction analysis unit performs correlation calculation on the loading heterogeneity index η load and the disturbance return trend index β rb to generate the loading pressure reconstruction index Pyq, and adopts normalization and tanh function processing to form a risk index response curve with gradient stability. The purpose of the formula is to pass the loading heterogeneity index η load and the disturbance return trend index βrb The product of the two abnormal phenomena enhances the effect, builds a dynamic adjustment amount of pressure control benchmark, and is used to determine whether there is a local rebound and loading failure risk. Disturbance return trend index β rb Reflects the disturbance return trend, which is a risk signal of the structure response layer, and the loading heterogeneity index η load is a stability signal of the stress distribution layer, and the product of the two represents the rebound trend in the background of loading deviation, that is, the cumulative risk of negative coupling, and respectively represent the maximum disturbance return trend index in the whole region and the average of the loading heterogeneity index in the whole region, which are used as a comparison benchmark to standardize the risk level. The hyperbolic sine function tanh is used to buffer and limit the control response curve, achieving a balance between high sensitivity identification and low disturbance output, and the standard pressure value P base are multiplied to realize dynamic correction of the given operating condition.

[0102] The risk assessment unit constructs a first pressure control critical threshold T base and a second pressure control critical threshold T P1 according to the system standard pressure P P2 , and performs rebound failure risk assessment with the real-time obtained loading pressure reconstruction index P yq , and triggers the corresponding pressure adjustment, hot air control or feed optimization instructions, to realize dynamic classification and differentiation of rebound failure, disturbance trend and accumulation overload three types of risk states. This embodiment realizes the whole process quantitative linkage control of abnormal grinding roller loading state, structure response return trend, system pressure abnormal modeling and dynamic control decision, has the beneficial effects of strong early identification ability, high control response rate and low risk misjudgment rate, compared with the traditional experience pressure regulation or alarm post-processing mode, significantly improves the self-control ability, operation stability and energy efficiency of the system, effectively prolongs the service life of the equipment and reduces the risk of unplanned downtime.

[0103] Embodiment 5

[0104] Please refer to Figure 1 and Figure 2 , specifically: the control execution module includes an instruction execution unit and an optimization feedback unit;

[0105] The instruction execution unit is used to receive the control instruction after the rebound failure risk assessment in real time, and execute the optimization adjustment according to the control instruction, specifically as follows:

[0106] The first adjustment instruction: generates risk information transmission to the intelligent control system of the numerical control vertical mill, controls the vertical roller mill to stop material feeding, and reduces the roller pressure of the 15% error pressure section by 15% and reduces the hot air inlet wind speed by 10%, and then iteratively evaluates and adjusts through the data acquisition module until the roller load distribution is uniform;

[0107] The second adjustment instruction: generates interference information transmission to the intelligent control system of the numerical control vertical mill, controls the vertical roller mill to reduce the roller pressure of the 8% error pressure section by 8%, and iteratively evaluates and adjusts through the data acquisition module until the roller load distribution is uniform;

[0108] The third adjustment instruction: generates accumulation information transmission to the intelligent control system of the numerical control vertical mill, controls the vertical roller mill to close the abnormal area material feeding, and increases the feeding rate of the adjacent healthy section by 7%, and then iteratively evaluates and adjusts through the data acquisition module until the roller load distribution is uniform.

[0109] The optimization feedback unit is used for sampling and archiving the rebound failure risk evaluation results each time, and comparing the current iterative evaluation results with the previous iterative evaluation results, if the continuous three times of iterative evaluation results do not reach the benefit standard, it means that there is mechanical problem failure, at this time, the vertical roller mill is stopped through the intelligent control system of the numerical control vertical mill, and mechanical failure information is generated and transmitted to the relevant personnel, reminding to carry out equipment maintenance and repair;

[0110] The benefit standard includes that the particle size fluctuation decreases by more than 10%, the unit energy consumption decreases by more than 5%, and the roller surface temperature rise and vibration amplitude decreases by more than 8%.

[0111] In the embodiment, the control execution module cooperates with the optimization feedback unit through the instruction execution unit, which is used to receive the regulation and control instruction after the rebound failure risk assessment in real time and execute the optimization adjustment according to the regulation and control instruction. When the rebound failure risk assessment is that there is a rebound and failure loading risk in the roller contact area, the intelligent control system of the numerical control vertical mill automatically executes the first adjustment instruction to stop the abnormal area feeding, reduce the pressure by 15% and the hot air speed by 10%, and quickly suppress the rebound source. When the rebound failure risk assessment is that there is a rebound interference trend, the second adjustment instruction is executed to reduce the pressure by 8% to balance the micro load deviation. When the rebound failure risk assessment is that there is no rebound interference influence, the third adjustment instruction is executed to adjust the local feeding structure to realize load redistribution. After each round of regulation and control, the optimization feedback unit samples and archives the rebound failure assessment results, compares the adjustment effect with the previous one, and takes the particle size fluctuation, unit energy consumption and temperature rise amplitude as the evaluation indexes. If the benefit standard is not reached for three consecutive rounds, it is actively identified as a mechanical fault and a stop command is issued. The embodiment builds a complete path from risk identification, hierarchical regulation and control to closed-loop verification and fault warning, not only realizes the dynamic balance adjustment of the roller loading distribution, but also effectively reduces the rebound energy consumption fluctuation, structural heat accumulation and vibration abnormal risk in the running process, significantly improves the running stability, energy saving and self-diagnosis ability of the system, and breaks through the limitations of the traditional vertical mill system relying on manual adjustment, delayed response and late fault discovery.

[0112] Although the embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made thereto without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A CNC vertical mill intelligent control system based on parameter self-optimization, characterized in that: It includes a data acquisition module, a heterogeneous identification module, a reconfiguration and control module, and a control execution module; The data acquisition module is used to collect the grinding roller data of the vertical roller mill in real time based on the sensor group set in different parts of the CNC vertical mill equipment, and transmit it to the intelligent control system of the CNC vertical mill in real time for preprocessing to obtain the force coupling data group and the structural response data group. The data acquisition module includes a data acquisition unit and a data processing unit; The data acquisition unit is used to collect the grinding roller data of the vertical mill in real time based on the sensor groups installed at different parts of the CNC vertical mill equipment. The sensor group includes a stress sensor, a symmetrical single-axis accelerometer, a current transformer, a triaxial accelerometer, a thermocouple, and a differential pressure sensor. The stress sensor is installed below the contact surface of the grinding roller and in the bearing housing area to collect the normal stress y in the contact area of ​​the grinding roller in real time. l ; The symmetrical single-axis accelerometers are installed at symmetrical positions on the two grinding roller arm support structures to collect the left-right vibration difference 'a' of the grinding rollers in real time. d ; The current transformer is installed at the connection point between the main drive motor's input circuit and the PLC to collect the main drive motor's load current I in real time. r ; The triaxial accelerometer is vertically arranged above the side wall of the grinding disc base and near the central axis of the inner wall of the grinding cavity housing to collect the dominant frequency f of regional vibration in real time. v ; The thermocouple is installed at the inlet port of the return air duct at the top of the grinding cavity to collect the local temperature T of the grinding cavity in real time. m ; The differential pressure sensor is installed in the upper exhaust duct of the grinding chamber to collect the negative pressure P in the grinding chamber area in real time. v ; The heterogeneity identification module is used to calculate and obtain the loading heterogeneity index η based on the force coupling data set. load and the preset loading heterogeneous boundary threshold T η Perform loading heterogeneity assessment, and execute rebound interference analysis command when the assessment indicates that loading heterogeneity exists in the grinding roller; The reconfiguration and control module is used to execute rebound disturbance analysis commands and calculate the disturbance return trend index β based on the structural response data set. rb Then, with the loading heterogeneity index η load Perform correlation calculations to obtain the loading pressure reconstruction index P. yq Then, compared with the preset first pressure control critical threshold T P1 Second pressure control critical threshold T P2 Conduct a risk assessment of rebound failure; The reconstruction control module includes an interference removal unit and a reconstruction analysis unit; The interference removal unit is used to execute a rebound interference analysis command when the loading heterogeneity assessment indicates that the grinding roller has loading heterogeneity; The bounce interference analysis command is used to perform correlation calculations based on the parameters within the structural response data set, analyze the disturbance return state of each loaded heterogeneous section of the vertical roller mill, and determine the disturbance return trend index β of each loaded heterogeneous section of the vertical roller mill. rb This reflects the transient interference surge characteristics caused by multi-source response coupling anomalies, specifically: In the formula, n represents the total number of segments of the grinding roller, and f v，i T m，i and P v，i The dominant frequency f of the regional vibration in the i-th segment is represented by f. v Local temperature T in the grinding cavity m Negative pressure P in the grinding cavity area v ε represents the zero perturbation prevention term, with a value of 0.001; The reconstruction analysis unit includes a bounce interference analysis unit and a risk assessment unit; The bounce interference analysis unit is used to analyze the heterogeneity index η based on the loading. load And the perturbation retracement trend index β rb Perform correlation calculations to obtain the loading pressure reconstruction index P. yq A comprehensive analysis of the regional rebound interference risk is as follows: In the formula, P base This represents the standard pressure value set by the system, tanh represents the hyperbolic tangent function, and α represents the control sensitivity factor, which is set to 1.2 based on debugging experience. This represents the index indicating the maximum disturbance return trend across the entire region. ε represents the mean of the heterogeneity index loaded across the entire region, and ε represents the zero perturbation term, with a value of 0.

001. The control execution module is used to receive control instructions after the rebound failure risk assessment in real time, and to perform optimization adjustments according to the control instructions. If the benefit standard is not met after three consecutive optimization adjustments, mechanical fault information is generated.

2. The intelligent control system for a CNC vertical mill based on parameter self-optimization according to claim 1, characterized in that: The data processing unit is used to establish a communication connection between the sensor group and the intelligent control system of the CNC vertical mill based on the wireless network, and transmit the collected grinding roller data to the intelligent control system of the CNC vertical mill in real time for preprocessing to obtain the force coupling data group and the structural response data group. The preprocessing includes denoising, filling in missing values, and dimensionless processing; The denoising process uses multidimensional filtering technology to suppress noise in the grinding roller data, eliminating the influence of noise in the data. The missing value filling process uses missing value interpolation technology to complete the grinding roller data affected by sampling interruption and communication packet loss. The dimensionless processing uses the Max-Min method to remove the dimension influence of the grinding roller data. The force coupling data set includes the normal stress y in the contact area of ​​the grinding roller. l The difference in left and right vibration of the grinding roller, a d and the load current I of the main drive motor r ; The structural response data set includes the dominant regional vibration frequency f. v Local temperature T in the grinding cavity m Negative pressure P in the grinding cavity area v .

3. The intelligent control system for a CNC vertical mill based on parameter self-optimization according to claim 2, characterized in that: The heterogeneity identification module includes a heterogeneity identification unit and a heterogeneity evaluation loading unit; The heterogeneity identification unit is used to perform correlation calculations based on the parameters within the force coupling data set, analyze the local imbalance state of the loading structure, and determine the loading heterogeneity index η of each section of the vertical roller mill. load Specifically: In the formula, ln represents the logarithmic function, n represents the total number of segments of the grinding roller, and a d，i I r，i and y l，i Let y represent the normal stress y in the contact area of ​​the grinding roller in the i-th segment. l The difference in left and right vibration of the grinding roller, a d and the load current I of the main drive motor r ε represents the zero perturbation prevention term, with a value of 0.

001.

4. The intelligent control system for a CNC vertical mill based on parameter self-optimization according to claim 3, characterized in that: The loading heterogeneity assessment unit is used to calculate the loading heterogeneity index η over a six-month period based on statistical methods. load mean and standard deviation and based on the mean and standard deviation Preset loading heterogeneous boundary threshold T η Specifically: Then, compared with the real-time acquired loading heterogeneity index η load A heterogeneous loading assessment was conducted, and the specific assessment scheme is as follows; When the heterogeneity index η is loaded load <Loading heterogeneous boundary threshold T> η When the load distribution on the grinding roller is uniform, the normal operating procedure should be maintained, and normal monitoring should continue. When the heterogeneity index η is loaded load ≥ Loading heterogeneous boundary threshold T η When this occurs, it indicates that there is a loading anomaly in the grinding roller, and at this time, the rebound interference analysis command is executed.

5. The intelligent control system for a CNC vertical mill based on parameter self-optimization according to claim 1, characterized in that: The risk assessment unit is used to assess the standard pressure value P set by the system. base Set the first pressure control threshold T. P1 Second pressure control critical threshold T P2 Specifically: T P1 =0.9*P base T P2 =0.97*P base Then, it is compared with the real-time acquired loading pressure reconstruction index P. yq A risk assessment of rebound failure will be conducted, and the specific assessment plan is as follows; When the pressure reconstruction index P is applied yq ≤ First pressure control critical threshold T P1 When this occurs, it indicates that there is a risk of rebound and failure loading in the contact area of ​​the grinding roller, and the first adjustment command is executed at this time; When the first pressure control critical threshold T P1 <Loading pressure reconstruction index P yq <Second pressure control critical threshold T P2 When this occurs, it indicates a rebound interference trend, and the second adjustment instruction is executed. When the pressure reconstruction index P is applied yq ≥Second pressure control critical threshold T P2 When the condition is met, it indicates that there is no rebound interference, and the loading heterogeneity is caused by feed accumulation. At this time, the third adjustment command is executed.

6. The intelligent control system for a CNC vertical mill based on parameter self-optimization according to claim 5, characterized in that: The control execution module includes an instruction execution unit and an optimization feedback unit; The instruction execution unit is used to receive control instructions after the rebound failure risk assessment in real time, and to perform optimization adjustments according to the control instructions, as follows; The first adjustment instruction is to generate risk information and transmit it to the intelligent control system of the CNC vertical mill, control the vertical roller mill to stop feeding material, reduce the pressure of the grinding roller in the mis-pressure zone by 15% and reduce the hot air inlet velocity by 10%, and then perform iterative evaluation and adjustment through the data acquisition module until the grinding roller load distribution is uniform. The second adjustment command: generates interference information and transmits it to the intelligent control system of the CNC vertical mill, controls the vertical roller mill to reduce the pressure of the grinding roller in the 8% error pressure zone, and performs iterative evaluation and adjustment through the data acquisition module until the grinding roller load distribution is uniform; The third adjustment command generates accumulation information and transmits it to the intelligent control system of the CNC vertical mill. It controls the vertical roller mill to close the material feeding in the abnormal zone and increases the feeding rate of the adjacent healthy zone by 7%. Then, it performs iterative evaluation and adjustment through the data acquisition module until the roller loading distribution is uniform.

7. The intelligent control system for a CNC vertical mill based on parameter self-optimization according to claim 6, characterized in that: The optimization feedback unit is used to sample and archive the risk assessment results of each rebound failure, and compare the current iteration assessment results with the previous iteration assessment results. If the benefit standard is not met for three consecutive iteration assessments, it indicates that there is a mechanical problem. At this time, the vertical roller mill is stopped by the CNC vertical mill intelligent control system, and mechanical fault information is generated and transmitted to relevant personnel to remind them to carry out equipment maintenance and repair. The benefit criteria include a reduction of more than 10% in particle size fluctuation, a reduction of more than 5% in unit energy consumption, and a reduction of more than 8% in the surface temperature rise and vibration amplitude of the grinding roller.

Images