Method and system for parameterized simulation of radian of main roller of roller press and storage medium

Abstract

The invention relates to the technical field of roller press main roller simulation, in particular to a parameterized simulation method and system for the radian of a roller press main roller and a storage medium. The method comprises the following steps that surface space scanning is conducted on a main roller of the roller press, and main roller space parameters are obtained; main roller radian distribution evaluation is conducted according to the main roller space parameters, and a main roller radian distribution diagram is obtained; carrying out regional division on the main roller radian distribution diagram, and importing mechanical analysis software to construct a roller press main roller initial state model; rolling material data are obtained; performing material-main roller stress parameterization simulation according to the rolled material data and the roller press main roller initial state model to obtain simulated main roller stress state data; and intelligent main roller bending compensation is carried out based on the simulated main roller stress state data, and main roller bending compensation parameters are obtained. Intelligent compensation of the radian state of the main roller of the roller press is achieved through main roller stress parameterization simulation, the product quality is effectively improved, energy consumption is reduced, and the service life of equipment is prolonged.

Description

Technical Field

[0001] The present invention relates to the technical field of roller press main roller simulation, and particularly to a parametric simulation method, system and storage medium for the arc of a roller press main roller. Background Art

[0002] In the actual production process, the main roller of the roller press bears complex and variable stress distributions, and the wear degrees in different regions are different, resulting in uneven changes in the arc of the main roller surface. This arc deviation will lead to uneven pressure distribution during the roller pressing process, causing a series of problems such as inconsistent material density, large fluctuations in product quality, and increased energy consumption. Especially under high-pressure and high-load working conditions, the problem of main roller arc deviation is more prominent, seriously affecting product quality and equipment service life. The traditional detection and maintenance methods for the main roller of the roller press mainly rely on periodic shutdown inspections and empirical repairs, and it is impossible to grasp the arc change of the main roller in real time. This passive maintenance strategy has obvious defects: on the one hand, the shutdown detection period is long, and it is difficult to detect arc anomalies in a timely manner; on the other hand, the detection means are simple, the accuracy is limited and it is easy to cause secondary damage. More importantly, the existing technology cannot accurately predict the arc change trend of the main roller under different working conditions, and maintenance personnel can only make compensation adjustments based on experience, making it difficult to achieve precise control. Summary of the Invention

[0003] Based on this, the present invention provides a parametric simulation method, system and storage medium for the arc of a roller press main roller to solve at least one of the above technical problems.

[0004] To achieve the above object, a parametric simulation method for the arc of a roller press main roller includes the following steps:

[0005] Step S1: Perform a surface space scan on the main roller of the roller press to obtain the main roller space parameters; perform multiple pre-measurements on the working conditions of the main roller of the roller press to generate the main roller operating state data;

[0006] Step S2: Divide the main roller sections according to the main roller operating state data, and then perform an analysis of the main roller arc deviation to obtain the main roller arc deviation data; calculate the main roller arc change rate according to the main roller arc deviation data; determine the actual working state of the main roller, and perform an evaluation of the main roller arc distribution according to the main roller arc deviation data to obtain the main roller arc distribution map;

[0007] Step S3: Divide the main roller arc distribution map into regions through the main roller arc change rate and mark the arc stress points; import the main roller space parameters and the arc stress points into a mechanical analysis software to construct an initial state model of the roller press main roller; obtain the roller pressed material data; perform a parametric simulation of the material-main roller stress according to the roller pressed material data and the initial state model of the roller press main roller to obtain the simulated main roller stress state data;

[0008] Step S4: Perform intelligent main roll bending compensation based on the simulated main roll stress state data to obtain the main roll bending compensation parameters.

[0009] Preferably, the present invention further provides a parametric simulation system for the main roll curvature of a roll press, which executes the parametric simulation method for the main roll curvature of the roll press as described above. The parametric simulation system for the main roll curvature of the roll press includes:

[0010] The main roll curvature scanning module is used to perform surface space scanning on the main roll of the roll press to obtain the main roll space parameters; perform multiple pre-measurements on the working conditions of the main roll of the roll press to generate the main roll operating state data;

[0011] The curvature analysis module is used to divide the main roll sections according to the main roll operating state data, and then perform main roll curvature deviation analysis to obtain the main roll curvature deviation data; calculate the main roll curvature change rate according to the main roll curvature deviation data; determine the actual working state of the main roll, and perform main roll curvature distribution evaluation according to the main roll curvature deviation data to obtain the main roll curvature distribution map;

[0012] The stress simulation module is used to divide the main roll curvature distribution map into regions through the main roll curvature change rate and mark the curvature stress points; import the main roll space parameters and the curvature stress points into the mechanical analysis software to construct the initial state model of the main roll of the roll press; obtain the roll-pressed material data; perform material-main roll stress parametric simulation according to the roll-pressed material data and the initial state model of the main roll of the roll press to obtain the simulated main roll stress state data;

[0013] The bending compensation module is used to perform intelligent main roll bending compensation based on the simulated main roll stress state data to obtain the main roll bending compensation parameters.

[0014] Preferably, the present invention further provides a computer-readable storage medium storing a computer program, and when the computer program is executed, it implements the parametric simulation method for the main roll curvature of the roll press as described in any one of the above.

[0015] By performing surface space scanning on the main roller of the roller press, complete and accurate spatial parameters of the main roller are obtained, overcoming the defects of sampling detection and limited accuracy in traditional detection methods, and ensuring the reliability of the analysis results. At the same time, by collecting multiple pre-measurements of the working conditions of the main roller to generate the main roller operation status data, not only the current geometric shape of the main roller is considered, but also its dynamic changes under different working conditions are fully considered, making the evaluation of the main roller status more comprehensive and accurate. Based on the main roller operation status data, the division of the main roller sections and the analysis of the arc deviation can accurately locate the specific position and degree of the arc deviation. By calculating the arc change rate of the main roller, the law of the arc change of the main roller over time can be revealed. By dividing the main roller arc distribution diagram into regions through the main roller arc change rate and marking the arc force points, the key stress regions on the surface of the main roller can be accurately identified. Importing this information together with the main roller spatial parameters into the mechanical analysis software to construct a highly realistic initial state model of the roller press main roller. This model not only contains the geometric information of the main roller but also reflects its stress state under actual working conditions. Combining the obtained roller-pressed material data for material-main roller stress parametric simulation can accurately simulate the complex interaction between the material and the main roller during the roller pressing process to obtain the simulated main roller stress state data. This parametric simulation method fully considers the characteristics of the material and the actual state of the main roller, overcoming the limitations of simplified models and ignoring non-linear factors in traditional simulation methods, making the simulation results closer to the actual situation. Through the analysis of the simulation results, the deformation trend of the main roller under different working conditions can be accurately predicted, and based on this, the optimal bending compensation scheme can be formulated. This intelligent compensation method can effectively offset the negative impact caused by the arc deviation of the main roller, ensure the uniformity of the pressure distribution during the roller pressing process, thereby improving product quality, reducing energy consumption, and extending the service life of the equipment. Compared with the traditional empirical compensation method, the intelligent compensation method of this solution is more scientific and accurate, can realize the active control of the roller press main roller status, and avoid secondary damage or performance degradation caused by improper compensation. Practical applications show that after adopting this method, the product qualification rate has increased by more than 15%, the energy consumption has been reduced by about 12%, the service life of the main roller has been extended by more than 30%, and the annual economic benefits have been significantly improved. At the same time, this method reduces the dependence on manual experience, reduces the maintenance cost, improves the equipment reliability, provides technical support for the intelligent and refined operation management of the roller press, and has broad application prospects for promotion. Therefore, a parametric simulation method for the arc of the roller press main roller of the present invention significantly improves the control level of the main roller arc deviation through the closed-loop logic of real-time monitoring, accurate analysis, simulation prediction, and intelligent compensation, greatly improves the stability and consistency of product quality, effectively reduces the energy consumption caused by uneven pressure, significantly extends the service life of the main roller and related components by reducing local overload and optimizing the stress, and at the same time reduces the unplanned downtime and maintenance cost, ultimately improving the efficiency and economic benefits of the entire roller pressing production process. Brief Description of the Drawings

[0016] Figure 1 It is a schematic flow chart of the steps of a parametric simulation method for the arc of the main roller of a roller press according to the present invention;

[0017] Figure 2 is Figure 1 a detailed implementation step flow chart of step S4 in

[0018] The realization, functional features and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. Specific Embodiments

[0019] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work shall fall within the scope of protection of the present invention.

[0020] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus their repeated description will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0021] It should be understood that although the terms "first", "second", etc. may be used here to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used here includes any and all combinations of one or more of the listed related items.

[0022] To achieve the above object, please refer to Figures 1 to 2 , the present invention provides a parametric simulation method for the arc of the main roller of a roller press, including the following steps:

[0023] Step S1: Perform a surface space scan on the main roller of the roller press to obtain the main roller space parameters; perform multiple pre-measurements on the working conditions of the main roller of the roller press to generate the main roller operating state data;

[0024] Step S2: Divide the main roller sections according to the main roller operation status data, then conduct an analysis of the main roller radian deviation to obtain the main roller radian deviation data; calculate the main roller radian change rate based on the main roller radian deviation data; determine the actual working status of the main roller, and conduct an evaluation of the main roller radian distribution according to the main roller radian deviation data to obtain the main roller radian distribution map;

[0025] Step S3: Divide the main roller radian distribution map into regions through the main roller radian change rate and mark the radian stress points; import the main roller space parameters and the radian stress points into a mechanical analysis software to construct an initial state model of the main roller of the roller press; obtain the roller pressing material data; conduct a material-main roller stress parametric simulation according to the roller pressing material data and the initial state model of the main roller of the roller press to obtain the simulated main roller stress state data;

[0026] Step S4: Conduct intelligent main roller bending compensation based on the simulated main roller stress state data to obtain the main roller bending compensation parameters.

[0027] In the embodiment of the present invention, the parametric simulation method for the main roller radian of the roller press includes the following steps:

[0028] Step S1: Conduct a surface space scan on the main roller of the roller press to obtain the main roller space parameters; conduct multiple pre-measurements on the working conditions of the main roller of the roller press to generate the main roller operation status data;

[0029] In an embodiment of the present invention, for example, on a 1200mm×2000mm high-pressure roll press in a certain cement production plant, an MLS-2100 laser ranging device is used to perform precise spatial scanning on the surface of the main roll. During the scanning process, the main roll is fixed on a rotating fixture, and axial scanning is performed every 10mm. The next scan is performed every 5 degrees of rotation. After a full 360-degree scan, a data set containing approximately 15,000 three-dimensional coordinate points is formed. The coordinate points obtained from the scan are processed using the least squares method for circle fitting to obtain the center coordinates and radius values at 21 axial positions. These data together constitute the spatial profile characteristics of the main roll. Seven measurement points are set on the main roll according to the axial position, and four measurement points are arranged circumferentially at each position (located at 0 degrees, 90 degrees, 180 degrees, and 270 degrees respectively), for a total of 28 measurement points. A PT100 temperature sensor, an eddy current displacement sensor, and a torque strain gauge are installed at the measurement point positions. The measurement accuracies of the sensors are ±0.1°C, 0.001mm, and 2.1 sensitivity coefficients respectively. Subsequently, three pre-measurements of the working conditions are performed: the first is to collect data for 5 seconds 10 seconds after the roll press starts without load; the second is to collect data for 5 seconds 10 seconds after the material enters the roll gap after loading; the third is to collect data for 10 seconds after stable operation for 5 minutes. The collected raw data includes the temperature distribution (range 25°C to 80°C), torque values (from no-load to full-load state), and radial displacement data of the main roll at different positions. The collected data is processed using digital filtering technology to remove outliers and noise interference, and a data set accurately reflecting the operating state of the main roll is generated.

[0030] Step S2: Divide the main roll into sections according to the main roll operating state data, and then perform an analysis of the main roll arc deviation to obtain the main roll arc deviation data; calculate the main roll arc change rate based on the main roll arc deviation data; determine the actual working state of the main roll, and perform an evaluation of the main roll arc distribution according to the main roll arc deviation data to obtain the main roll arc distribution diagram;

[0031] In the embodiments of the present invention, based on the main roll operating state data obtained in step S1, data classification matrix processing is first performed, and the measurement data is separated into two matrices of axial measurement data and circumferential measurement data according to spatial attributes. Subsequently, the 2000-mm long main roll is accurately divided into three sections: the first end section (0 - 700 mm), the central section (700 - 1300 mm), and the second end section (1300 - 2000 mm), accounting for 35%, 30%, and 35% of the total length respectively. Corresponding to the distribution of measurement points, the 1st, 2nd, and 3rd measurement points are located in the first end section, the 4th and 5th measurement points are located in the central section, and the 6th and 7th measurement points are located in the second end section. The least-squares circular arc fitting method is used to calculate the axial radian values of each section. The radian value of the first end section is approximately 1.36, the central section is approximately 0.95, and the second end section is approximately 1.42 (unit: 10^-2 mm^-1). Frequency domain analysis is performed on the circumferential torque measurement data to calculate the torque fluctuation characteristic value and the non-uniformity index, and the typical value is 7.5%. The temperature data is aligned in time series through cubic spline interpolation to create a complete time series data of circumferential temperature values. The radian difference and temperature gradient distribution between the central region and the two end regions are calculated to construct a weighted fusion comprehensive evaluation model of radian deviation. The change curve of the radian difference over time is analyzed through the sliding window technique, and the radian change rate is calculated and the positions of mutation points are marked. At the same time, the comprehensive load index 0.92 is calculated through the main motor load rate (92%), the main roll torque value (87% of the rated torque), and the hydraulic system pressure value (96% of the rated pressure), and it is determined that the main roll is in a full-load state. The piecewise polynomial fitting method is used to analyze the local radian data at 21 axial positions, and the principal component analysis method is used to process the radian field distribution characteristics, and finally a radian distribution map intuitively showing the radian distribution state of the main roll is generated.

[0032] Step S3: Divide the main roll radian distribution map by the main roll radian change rate and mark the radian stress points; import the main roll spatial parameters and the radian stress points into the mechanical analysis software to construct the initial state model of the main roll of the roller press; obtain the roll-pressed material data; perform material-main roll stress parametric simulation based on the roll-pressed material data and the initial state model of the main roll of the roller press to obtain the simulated main roll stress state data;

[0033] In the embodiments of the present invention, the main roller radian distribution map is divided into regions according to a radian deviation range of 0.02 mm, and about 8 radian gradient regions are obtained. The ratio of the radian difference to the distance between adjacent points is measured in each region, and it is identified that the radian-distance difference ratios at the axial positions of 583 mm and 1417 mm are 0.012 and 0.014 respectively, exceeding the threshold judgment standard of 0.01, and these two positions are marked as the starting regions of the radian stress points. Since the deviation section is close to both ends of the roller press, the stress transmission direction is determined to be from both ends to the middle, and the stress magnitude is marked. The main roller space parameters and the radian stress point data are imported into the mechanical analysis system to create an accurate three-dimensional geometric model of the main roller, and the model is meshed, generating approximately 180,000 hexahedron elements in total. Boundary conditions are set in the model: all degrees of freedom are fixed at the driving end, axial displacement is allowed at the non-driving end, and elastic constraint conditions are set at the support positions. The radian stress points are set as the load application points, and the load magnitude and direction are calculated according to the radian gradient value of the region where the stress points are located. At the same time, a cement raw material sample is obtained from the production line, and its hardness value (Mohs hardness level 5), density value (2.8 g / cm³), moisture content (6.5%), and particle size distribution (26% for 10 - 30 mm, 32% for 5 - 10 mm, 42% for 0 - 5 mm) are measured. Based on the material characteristic parameters, the particle size distribution, cohesion coefficient, friction coefficient, and elastic modulus of the discrete element model are set to construct a material-roller surface interaction simulation model for the roller pressing process. A 60-second dynamic analysis is performed in the co-simulation system, and the main roller bending deformation data is recorded every 0.5 seconds. The maximum deformation amount, average deformation amount, and deformation amount change rate in the stable operation stage are extracted to generate complete simulation main roller stress state data.

[0034] Step S4: Perform intelligent main roller bending compensation based on the simulation main roller stress state data to obtain the main roller bending compensation parameters.

[0035] In the embodiment of the present invention, the hardware units of the main roller of the roll press are encoded and marked, and key parameters such as the working surface width (1400 mm) of the main roller body, the rated pressure range (500 - 2000 kN / m), and the allowable radial deformation value (1.8 mm) are analyzed. The position distribution data of the hydraulic support units are extracted. The 6 support units are located at the axial positions of 417 mm, 650 mm, 883 mm, 1117 mm, 1350 mm, and 1583 mm, and the piston stroke of each support unit is 20 mm. An accurate correspondence between the support position and the stroke is established through position - stroke capacity mapping, and the pressure fitness of all support units is calculated, with an average fitness reaching 98.4%. 11 high - precision pressure sensors are arranged to monitor the extrusion force values at the middle and both ends of the main roller in real - time. The average extrusion force in the middle region is calculated to be 1782 kN / m, the average extrusion force at both ends is 1645 kN / m, the difference is 137 kN / m, and the normalized difference is 8.02%. The extrusion force distribution data is processed by multi - threshold segmentation, and two extrusion force peak points are marked at the axial positions of 600 mm and 1400 mm, with peak pressures of 1894 kN / m and 1862 kN / m respectively. The force diffusion around these peak points is monitored at intervals of 5 cm to generate a diffusion gradient distribution map. The density clustering algorithm is used to perform clustering analysis on abnormal stress points, and 3 potential deformation regions are identified. The degree of deformation is evaluated by the multi - index weighted scoring method. Among them, the regions of 450 - 720 mm and 1320 - 1620 mm are classified as severely deformed regions, and the region of 830 - 980 mm is classified as a slightly deformed region. The support units 2 and 6 corresponding to the severely deformed regions are respectively increased by 42 kN and 37 kN of compensation force, and the support unit 4 corresponding to the slightly deformed region is increased by 11 kN of compensation force. The compensation force is applied through a piece - wise linear control model, and the arc recovery is continuously monitored, and finally a complete bending compensation parameter file of the main roller is generated.

[0036] Preferably, step S1 includes the following steps:

[0037] Step S11: The surface of the main roller of the roll press is scanned in three - dimensional space by a laser ranging device to obtain the coordinates of the main roller surface in the three - dimensional space. The scanning accuracy is ±0.05 mm, the scanning pitch is 10 mm axially and 5° radially, and it is recorded as the main roller surface coordinate data;

[0038] Step S12: Determine the axial position and radial dimension of the main roller according to the main roller surface coordinate data;

[0039] Step S13: Encode the geometric information of the main roller of the roll press through the axial position and radial dimension to obtain the main roller space parameters;

[0040] Step S14: Based on the positions of 7 measurement points evenly arranged along the axial direction of the main roller of the roll press, 4 measuring points are set along the circumferential direction of the main roller surface at each measurement point position to obtain the main roller measurement point data;

[0041] Step S15: Deploy temperature sensors, eddy current displacement sensors, and torque strain gauges according to the main roller measurement point data, thereby constructing a main roller measurement network;

[0042] Step S16: Use the main roller measurement network to perform multiple pre-measurements on the working conditions of the main roller of the roll press to obtain the original main roller measurement data; among them, the multiple pre-measurements of the working conditions include three acquisitions. The first acquisition is carried out 10 seconds after the roll press starts without load, and the acquisition duration is 5 seconds; the second acquisition is carried out 10 seconds after the material enters the roll gap after loading, and the acquisition duration is 5 seconds; the third acquisition is carried out after stable operation for 5 minutes, and the acquisition duration is 10 seconds; the original main roller measurement data includes the main roller speed value, the main roller torque value, and the main roller temperature value;

[0043] Step S17: Perform digital filtering processing on the original main roller measurement data to generate the main roller operating state data.

[0044] In the embodiment of the present invention, the main roller is fixed on a special rotating tooling so that the main roller can rotate at a constant angular velocity. The laser ranging device is installed 500 mm away from the main roller surface in a direction parallel to the main roller axis and is installed on a precision guide rail to achieve axial movement. When starting the ranging device for scanning, first rotate the main roller to the initial position (marked as 0°), and the laser ranging device collects a measurement point every 10 mm along the axial direction from one end of the main roller. After each axial scan is completed, the main roller rotates 5° to continue the next round of axial scan, and so on until the entire 360° of the main roller surface is scanned. The data obtained at each measurement point includes the x, y, and z coordinate values of the point in the three-dimensional coordinate system, and the measurement accuracy is controlled within ±0.05 mm. Import the main roller surface coordinate data into the numerical calculation program, group it according to the axial position, and perform circular fitting on the data points in one circle at each axial position. The circular fitting adopts the Gauss-Newton iteration method. Let the center coordinates of each axial position be (a, b) and the radius be r, then the objective function is F = Σ[(xi - a) 2 +(yi - b) 2 - r 2 2 ​, through iterative calculation until the objective function converges, and the convergence threshold is set to 10^-6. After fitting, a series of center coordinates and radius values along the axis are obtained. The axial position of the main roller is determined by the connection of the center points obtained by fitting, which is the central axis of the main roller; the radial dimension is determined by the radius values at each axial position, including the maximum radius, the minimum radius, and the average radius. For the main roller with uneven wear, the roundness error at each axial position also needs to be calculated, which is the difference between the maximum radius and the minimum radius. A cylindrical coordinate system with the central axis of the main roller as the z-axis is established, and the position of each point on the surface of the main roller is expressed as (r, θ, z), where r is the radial distance, θ is the circumferential angle, and z is the axial position. Then, the Fourier series expansion is used to fit the radial distance r: r(θ, z) = r0(z) + Σ[a n (z)cos(nθ) + b n (z)sin(nθ)], where n ranges from 1 to 10. For r0(z), a n (z), and b n (z) at different axial positions z, the cubic spline interpolation method is used for fitting to obtain a continuous function. Finally, the spatial parameters of the main roller consist of a set of Fourier coefficients and spline coefficients, including a total of 21 axial position points, and each position point has 21 Fourier coefficients (r0 and a n , b n, n = 1 - 10), a total of 441 parameters. These parameters completely describe the geometric characteristics of the main roll surface, including the basic cylindrical shape, axial taper, local concavity and convexity, and wear condition. Measure the total length L of the main roll, and obtain the effective working length L' after removing the non-working section lengths at both ends. Divide the effective working length L' into 6 equal parts on average to determine the axial positions of 7 measurement points, which are located at 0, L' / 6, 2L' / 6, 3L' / 6, 4L' / 6, 5L' / 6, and L' respectively. For the main roll with a standard effective length of 2000 mm, these 7 measurement points are located at 0, 333.33, 666.67, 1000, 1333.33, 1666.67, and 2000 mm. At each axial measurement point position, 4 measuring points are evenly arranged circumferentially along the main roll surface, and the circumferential angles are 0°, 90°, 180°, and 270° respectively, forming a uniformly distributed measuring point matrix. The positions of the measuring points are permanently marked with metal marking pieces. The diameter of the marking pieces is 10 mm and the thickness is 0.5 mm. They are fixed on the main roll surface through epoxy resin adhesive. Each measuring point marking piece is engraved with a unique number in the format of "A - B", where A represents the axial position serial number (1 - 7) and B represents the circumferential position serial number (1 - 4), forming a data acquisition network for a total of 28 measurement points. The temperature sensor uses a PT100 platinum resistance sensor, with a temperature measurement range of -50°C to 500°C and an accuracy of ±0.1°C. It is installed at the 0° measuring point at 7 axial positions, and the sensor probe penetrates 3 mm below the main roll surface. It is installed by opening an installation hole with a diameter of 4 mm and a depth of 3 mm and filling it with thermal conductive silicone grease. The eddy current displacement sensor uses the ECP - 3502 model, with a measurement range of 0 - 5 mm and a resolution of 0.001 mm. It is installed at the corresponding positions of the 90° and 270° measuring points at 7 axial positions. The sensor probe maintains an initial gap of 2 mm from the main roll surface and is fixed on the roll press frame through a special bracket. The torque strain gauge uses the BF350 - 3AA model, with a sensitivity coefficient of 2.1. It is pasted at the 180° measuring point at 7 axial positions. The strain gauge is installed at a 45° angle along the main roll surface to maximize the detection of shear strain caused by torque. All sensor signals are connected to the data acquisition system through shielded cables, and the sampling frequency is set to 1000 Hz. The data acquisition system uses a 24 - bit resolution A / D converter to convert the analog signal into a digital signal and transmit it to the data processing unit, forming a complete main roll measurement network. Use the main roll measurement network to perform multiple pre - measurements of the working conditions. The first acquisition starts after the roll press motor starts, when the main roll reaches the rated speed (usually 30 rpm) and runs stably for 10 seconds, and continuously acquires data for 5 seconds. During the acquisition process, record the temperature values of each measuring point. The initial temperature is usually the ambient temperature (about 25°C); at the same time, record the radial displacement values measured by the eddy current sensor, which characterize the jumping condition of the main roll during no - load rotation; and record the strain values measured by the torque strain gauge, and convert them into torque values. The no - load torque value mainly reflects the mechanical resistance of the system.The second acquisition is carried out 10 seconds after the material (such as cement raw meal) enters the roll gap through the feeding device. At this time, the roller press begins to bear the material pressure, and the acquisition lasts for 5 seconds. The acquisition content includes the temperature rise value of each measuring point (usually rising by 2 - 5 °C), the radial displacement change (reflecting the force deformation of the roll system), and the torque value (significantly higher than the no-load value). The third acquisition is carried out 5 minutes after the roller press operates stably. At this time, the system reaches the thermal equilibrium state, and the acquisition lasts for 10 seconds. The recorded data includes the temperature distribution under the stable working state (usually reaching 60 - 80 °C), the radial position of the main roll (reflecting the system stability after long-term operation), and the torque value (reflecting the energy consumption of material processing). All acquired data is marked with a timestamp and saved as the original main roll measurement data. Preprocessing is performed on all acquired signals, including removing outliers and baseline drift. The outlier determination criterion is the data points outside the range of 3 times the standard deviation, and the linear interpolation method is used to replace the outliers. Then, the low-pass Butterworth filter is applied to the temperature data for processing. The cut-off frequency is set to 2 Hz, the filter order is 4, and the filter transfer function is H(s) = 1 / [1+(s / 2π·2). 4 , to eliminate high-frequency noise while retaining the temperature change trend. The band-pass filter is applied to the torque data for processing, and the passband range is 1 - 50 Hz, to retain the periodic torque fluctuation information related to the main roll speed. After filtering, the average torque value and the torque fluctuation amplitude are calculated. The wavelet denoising technique is applied to the eddy current displacement sensor data, using the db5 wavelet basis function, with 5 decomposition levels, and the soft threshold method is used for the threshold. The threshold value is calculated using the Donoho formula th = σ√(2lnN), where σ is the noise standard deviation and N is the number of data points.

[0045] Preferably, in step S2, the main roll section is divided according to the main roll operating state data, and then the main roll radian deviation analysis includes:

[0046] The main roll operating state data is divided into the main roll axial and the main roll surface circumferential measurement data, respectively obtaining the main roll axial measurement data and the main roll surface circumferential measurement data;

[0047] Based on the main roll spatial parameters, the main roll measurement points are divided. The main roll is divided into 3 sections according to the axial position, namely the two end sections and the central section, obtaining each section of the main roll; among them, the length of the two end sections is 35% of the full length of the main roll, and the length of the central section is 30% of the full length of the main roll;

[0048] Calculate the main roll axial radian value of each section of the main roll according to the main roll axial measurement data;

[0049] Analyze the circumferential torque value of each section of the main roll according to the main roll surface circumferential measurement data;

[0050] Perform time series alignment on the temperature values of the main roll surface circumferential measurement data to obtain the circumferential temperature value time series data;

[0051] Calculate the arc difference between the central area and the two end areas of the main roller based on the axial arc values of each section of the main roller to obtain the central-end arc difference value;

[0052] Compare the sectional numerical values according to the time series data of the circumferential temperature values and the circumferential torque values of each section of the main roller, and arrange them according to the numerical gradient to obtain the main roller temperature-torque gradient data;

[0053] Integrate the central-end arc difference value with the main roller temperature-torque gradient data for the main roller arc deviation to obtain the main roller arc deviation data.

[0054] In the embodiment of the present invention, the data classification matrix method is used to separate the data according to the spatial attributes. During the extraction process of the main roller axial measurement data, the sensor data at 7 axial positions are reorganized according to the time synchronization principle to form a data matrix M_axial of n×7, where n is the number of sampling time points. Each row in the matrix represents the measurement values at 7 different axial positions at the same moment, and each column represents the data sequence of the same axial position changing with time. When extracting the circumferential measurement data on the surface of the main roller, the data of 4 circumferential points at each axial position are arranged according to the circumferential angles (0°, 90°, 180°, 270°) to form a data matrix M_circ of n×28. The matrix is organized in a double-index structure of "axial position - circumferential position", that is, the first 4 columns are the data of 4 circumferential points at the first axial position, and so on. For a main roller with a standard length of 2000mm, each of the two end sections accounts for 35% of the total length, that is, 700mm at each end; the central section accounts for 30% of the total length, which is 600mm. When dividing, first determine the position of the origin of the main roller axial coordinate (usually one end face of the main roller), and then calculate the boundary positions of each section: the range of the first end section is 0 - 700mm, the range of the central section is 700 - 1300mm, and the range of the second end section is 1300 - 2000mm. Corresponding to the 7 set measurement points, the 1st, 2nd, and 3rd measurement points are located in the first end section, the 4th and 5th measurement points are located in the central section, and the 6th and 7th measurement points are located in the second end section. The section division follows the force and wear laws of the main roller. The two end sections usually bear greater stress concentration and non-uniform wear, while the central section maintains a relatively stable working state. Take the axial position points and their corresponding radial displacement values within each section to construct a coordinate set. For example, take the positions and radial displacement values of the 1st, 2nd, and 3rd measurement points in the first end section, the positions and radial displacement values of the 4th and 5th measurement points in the central section, and the positions and radial displacement values of the 6th and 7th measurement points in the second end section. Apply the least squares method to fit the circular arc equation (z - a) 2 +(r - b) 2 =R 2, where (a, b) are the coordinates of the center of the circle and R is the radius of the circle. The Levenberg-Marquardt algorithm is used in the fitting process to solve the non-linear least squares problem. The iteration termination condition is that the parameter change is less than 10^-8 or the number of iterations exceeds 500 times. After fitting, the curvature κ of the circular arc is calculated as κ = 1 / R, and the axial radian value is defined as 100·κ, with the unit of 10^-2mm^-1. For the main roller of the roller press under standard working conditions, the radian value of the first end section is usually 1.25 - 1.45, the radian value of the central section is 0.85 - 1.05, and the radian value of the second end section is 1.30 - 1.50. The torque data within each section is statistically analyzed in the time domain, and the mean value, standard deviation, maximum value, and minimum value are calculated. Taking the first end section as an example, the torque mean value of the 4 circumferential measurement points at the positions of the 1st, 2nd, and 3rd measurement points is calculated and denoted as T_avg_1, and the standard deviation is denoted as T_std_1; similarly, T_avg_2, T_std_2 and T_avg_3, T_std_3 of the central section and the second end section are calculated. Then, frequency domain analysis is carried out, and the fast Fourier transform (FFT) is performed on the torque time series of each section. The sampling window is the Hanning window, and the number of FFT points is 4096. The spectral distribution of the torque fluctuation is calculated. The amplitude of the main frequency component (usually related to the rotational speed of the main roller. For a main roller with a rotational speed of 30 rpm, the main frequency is 0.5 Hz) is extracted and denoted as the circumferential torque fluctuation characteristic values T_amp_1, T_amp_2, and T_amp_3 of each section. The torque non-uniformity index of each section is calculated, which is defined as the ratio of the standard deviation to the mean value: T_nonunif_i = T_std_i / T_avg_i × 100%, and this value should usually be less than 8% when the main roller is working normally. The starting moment of the first measurement (10 seconds after the roller press starts without load) is taken as the time zero point t0. Since the sampling of the temperature sensors at each measurement point is not completely synchronous and there is a small time difference, data interpolation alignment is required. The cubic spline interpolation method is used to resample the temperature time series of each measurement point on a unified time scale, and the sampling interval is 1 ms. The aligned temperature data is organized into a three-dimensional matrix T(i, j, k), where i represents the time index (from 1 to the total number of sampling points), j represents the axial position index (from 1 to 7), and k represents the circumferential position index (from 1 to 4). The time axis of the temperature data is normalized. The 5-second data of the first measurement is mapped to 0 - 5000 ms, the 5-second data of the second measurement is mapped to 10000 - 15000 ms, and the 10-second data of the third measurement is mapped to 60000 - 70000 ms. The temperature change in the intermediate period is estimated by linear interpolation to form the complete circumferential temperature value time series data. The average radian value of the two end sections is calculated as κ_ends = (κ_1 + κ_3) / 2, where κ_1 and κ_3 are the radian values of the first end section and the second end section respectively. Then, the central-two end radian difference Δκ = κ_2 - κ_ends is calculated, where κ_2 is the radian value of the central section.The positive or negative value of the radian difference represents the curvature distribution characteristic: when Δκ > 0, it means that the radian of the central section is greater than that of both ends, and the main roller is in a "convex" shape; when Δκ < 0, it means that the radian of the central section is less than that of both ends, and the main roller is in a "concave" shape. Under normal working conditions, the main roller should be slightly concave, and the typical radian difference Δκ should be between -0.2 and -0.5. Calculate the absolute value of the radian difference |Δκ| as the main roller radian uniformity index. The smaller this value is, the more uniform the axial radian distribution of the main roller is. At the same time, calculate the radian difference Δκ_ends = |κ_1 - κ_3| between the two end sections. This value reflects the symmetry of both ends of the main roller and should usually be controlled within 0.15. Calculate the average temperature of each section under the stable working state (the third measurement), denoted as T_avg_1, T_avg_2, and T_avg_3. Calculate the temperature gradients ΔT_1-2 = T_avg_1 - T_avg_2 and ΔT_3-2 = T_avg_3 - T_avg_2, representing the temperature differences between the ends and the center. Similarly, calculate the torque gradients ΔM_1-2 = T_avg_1 - T_avg_2 and ΔM_3-2 = T_avg_3 - T_avg_2. Then construct three-dimensional data points (κ_i, T_avg_i, M_avg_i), i = 1, 2, 3, representing the radian values, temperature values, and torque values of each section. Perform polynomial interpolation on these three data points to obtain the radian-temperature-torque relationship function κ = f(T, M). Calculate the partial derivatives of temperature and torque with respect to radian. and constitute a sensitivity matrix Arrange each section according to the numerical gradients of temperature and torque to form a gradient sequence table. Usually, the temperature gradient is sorted from high to low as the first end section > the second end section > the central section, and the torque gradient sorting depends on the material distribution. Define the main roller radian deviation index DI = w1·|Δκ| + w2·max(|ΔT_1-2|, |ΔT_3-2|) / T_ref + w3·max(|ΔM_1-2|, |ΔM_3-2|) / M_ref, where w1, w2, w3 are weight coefficients, taking values of 0.5, 0.3, and 0.2 respectively, T_ref is the reference temperature value (taking 50 °C), and M_ref is the reference torque value (taking 80% of the rated torque). Classify the main roller radian state according to the DI value: DI < 0.3 is in an excellent state, 0.3 ≤ DI < 0.6 is in a normal state, 0.6 ≤ DI < 0.9 is in a warning state, and DI ≥ 0.9 is in a dangerous state. At the same time, construct a spatial distribution diagram of the main roller radian deviation. The horizontal axis is the axial position z, and the vertical axis is the deviation value δ(z). The curve is obtained by polynomial fitting: δ(z) = a0 + a1z + a2z 2 + a3z 3 , where the coefficients a0, a1, a2, a3 are determined by least-squares fitting of the actual deviation values of each measurement point.

[0055] Preferably, in step S2, the main roller is divided into sections according to the operating state data of the main roller, and then the analysis of the arc deviation of the main roller includes:

[0056] Perform time series analysis on the center-to-end arc difference in the main roller arc deviation data, and calculate the change rate of the arc difference of the main roller at different time points to obtain the main roller arc change rate;

[0057] Judge whether there is an axis deviation of the main roller according to the main roller space parameters, and mark the sections with a deviation exceeding 0.05 mm to obtain the main roller deviation section data;

[0058] Determine the actual working state of the main roller, and divide the working state into three states: full load, half load and light load;

[0059] Based on the main roller deviation section data and the main roller temperature-torque gradient data in the main roller arc deviation data, measure the arc deformation amount of the main roller in three working states to obtain the main roller arc deformation data;

[0060] Determine the local arc data of the main roller according to the main roller arc deformation data;

[0061] Perform arc field statistics on the local arc data of the main roller and detect the arc field distribution data;

[0062] Import the arc field distribution data into a drawing tool, and map the data points to a two-dimensional coordinate system, where the coordinate axes represent the physical position and arc value of the main roller respectively, to obtain the main roller arc distribution map.

[0063] In the embodiment of the present invention, a sampling time point sequence T = {t1, t2,..., t n} is established, where t1 corresponds to the start time of the first collection, and t n corresponds to the end time of the third collection. Corresponding to each time point t i , calculate the center-to-end arc difference Δκ(t i ) at this moment. The sliding window technique is used to process the time series data, the window length is 1 second (1000 data points), and the window overlap rate is 50%. Calculate the mean, variance and linear trend slope of the data in each window. The arc difference change rate is calculated using the central difference method: R(t i ) = [Δκ(t i+1 ) - Δκ(t i-1 )] / (t i+1 - t i-1 ), and the unit is 10 -2 mm -1 / s. For each working stage (no-load start, load operation, stable operation), calculate the average change rates R_avg_1, R_avg_2, and R_avg_3 respectively. Set the change rate threshold R_th = 1.5×10 -3 mm -1 / s. When |R(t i )| > R_th, mark this moment as the radian mutation point. According to the time series analysis results, plot the curve of the radian difference changing with time, and mark the mutation points on the curve. Typically, during the process of the roller press from no-load to full-load, the change rate of the central-end radian difference first increases and then decreases, and finally stabilizes. The change rate in the stable stage should be less than 5×10 -4 mm -1 / s. It is expressed as L(t) = P0 + t·v in a three-dimensional rectangular coordinate system, where P0 is the starting coordinate of the axis, and v is the axis direction vector. Ideally, v should be parallel to the Z-axis. Use the main roller axial position data to extract the center coordinates at each axial position to form a point set {(x1, y1, z1), (x2, y2, z2),..., (x n , y n , z n )}. Adopt the three-dimensional least squares method to fit the actual axis equation L'(t) = P0' + t·v'. Calculate the angle θ = arccos((v·v') / (|v|·|v'|)) between the theoretical axis and the actual axis, and the minimum distance d = |(P0 - P0')×v| / |v| between the two axes. When θ > 0.1° or d > 0.05 mm, it is determined that there is an axis deviation of the main roller. Calculate a cross-section every 50 mm along the axial direction of the main roller, and calculate the deviation distance δ i = |P' i - P i | for each cross-section, where P' i is the point on the actual axis, and P i is the corresponding point on the ideal axis. Mark all δ iThe section with a deviation greater than 0.05 mm is the deviation section. Record its starting position, ending position, and maximum deviation value to form the main roll deviation section data table. Establish a working state determination index system, including three core indicators: the main motor load rate, the main roll torque value, and the hydraulic system pressure value. The main motor load rate \(M_r\) is calculated by the ratio of the actual output power of the motor to the rated power, \(M_r = P_{actual} / P_{rated}×100\%\); the main roll torque value \(T\) is calculated from the measurement data of the torque strain gauge, taking the average value under stable working conditions; the hydraulic system pressure value \(P\) is directly measured by a hydraulic sensor. Standardize the three indicators: \(M_{r\_norm}=M_r / 100\%\), \(T_{norm}=T / T_{rated}\), \(P_{norm}=P / P_{rated}\), where \(T_{rated}\) is the rated torque of the main roll and \(P_{rated}\) is the rated pressure of the hydraulic system. Calculate the comprehensive load index \(LI = 0.4×M_{r\_norm}+0.4×T_{norm}+0.2×P_{norm}\). Divide the working state according to the \(LI\) value: full load state (\(0.8\leq LI\leq1.0\)), corresponding to the material filling the roll gap and reaching the designed production capacity; half load state (\(0.4\leq LI<0.8\)), corresponding to the material partially filling the roll gap and the production load being medium; light load state (\(0.1\leq LI<0.4\)), corresponding to a small amount of material entering the roll gap and the production load being low. For the case of \(LI > 1.0\), it is recorded as the overload state, and the process parameters need to be adjusted immediately or the material input needs to be reduced. For each working state, select the "main roll temperature - torque gradient data" collected in that state. For each deviation section marked in the "main roll deviation section data", according to its axial position, find the corresponding temperature - torque gradient value from the "main roll temperature - torque gradient data". Using the empirical formula, estimate the deformation amount of the main roll rigid material in this deviation section under this working state according to the temperature - torque gradient value, the length and position of the deviation section, and the working state (full load, half load, light load). Divide the main roll axis into 21 evaluation points, and the distance between adjacent evaluation points is 1 / 20 of the effective length of the main roll. For each evaluation point, extract its radial displacement data and calculate the local curvature in combination with the original geometric data. The calculation process of the local curvature \(\kappa_{local}\) is as follows: ① Take the positions and radial coordinates of 5 points, 2 points before and after the evaluation point; ② Fit these 5 points with a second-order polynomial \(y = ax\) 2 +bx + c; ③ Calculate the curvature of the fitted curve at the evaluation point \(\kappa_{local}=|2a| / [1+(b) 2 ^(3 / 2). Solve for the fitting parameters a, b, and c using the least squares method, and construct the objective function \(F=\sum[(ax_i 2 +bx_i + c - y_i) 2, where \(i = 1\) to \(5\), the normal equations are obtained by taking the derivative to be zero, and the fitting coefficients are solved. For the edge points (the first 2 points and the last 2 points), an asymmetric point-taking strategy is adopted to ensure the fitting accuracy. Organize the local radian values of the 21 evaluation points into a sequence \(\{\kappa_{local\_1},\kappa_{local\_2},...,\kappa_{local\_21}\}\) to form the local radian data of the main roll. At the same time, calculate the deviation rate of the local radian from the global average radian: \(\delta_i=(\kappa_{local\_i}-\kappa_{avg}) / \kappa_{avg}\times100\%\). Under normal circumstances, the absolute value of this deviation rate should be less than 12%. Regard the local radian data as a 21-dimensional vector \(\kappa_{local}=[\kappa_{local\_1},\kappa_{local\_2},...,\kappa_{local\_21}]\). First, standardize the data so that the mean of each dimension is 0 and the standard deviation is 1 to obtain the standardized radian vector \(\kappa_{norm}\). Calculate the covariance matrix \(C = \kappa_{norm}^T\cdot\kappa_{norm} / 20\) of the standardized data, and the matrix size is \(21\times21\). Perform eigenvalue decomposition on the covariance matrix \(C\) to obtain the eigenvalues \(\lambda_1\geq\lambda_2\geq...\geq\lambda_{21}\) and the corresponding eigenvectors \(v_1,v_2,...,v_{21}\). The eigenvalues represent the variance contributions of the principal components, and the eigenvectors represent the directions of the principal components. Calculate the principal component scores: \(s_i=\kappa_{norm}\cdot v_i\), where \(i = 1\) to \(21\). Take the first 3 principal components to construct the radian field feature space. Calculate the cumulative variance contribution rate: \(CVR_k = (\lambda_1+\lambda_2+...+\lambda_k) / (\lambda_1+\lambda_2+...+\lambda_{21})\), where \(k = 1\) to \(3\). Usually \(CVR_3>0.85\), indicating that the first 3 principal components can explain more than 85% of the radian field variation. The spatial distribution characteristics of the radian field are determined by the scores of the first 3 principal components to form the radian field distribution data.

[0064] Preferably, in step S3, the main roll radian distribution map is divided into regions by the main roll radian change rate, and the radian force points are marked, including:

[0065] The main roll radian distribution map is divided into regions according to the radian deviation range by the main roll radian change rate to obtain the radian gradient regions; among them, the radian deviation range of each region is set to 0.02 mm;

[0066] Measure the radian difference between adjacent measurement points in each radian gradient region, and calculate the ratio of the radian difference to the distance between adjacent measurement points to obtain the radian-distance difference ratio data;

[0067] If the radian-distance difference ratio data is greater than 0.01 mm / cm, it is preliminarily determined that the region corresponding to the radian-distance difference ratio data is the starting region of the radian force point;

[0068] Determine the radian stress transfer direction based on the starting area of the radian force application point; if the deviation section of the main roller is close to the middle of the roller press, the stress transfer direction is from the middle to both ends; if the deviation section of the main roller is close to one end of the roller press, the stress transfer direction is from this end to the other end;

[0069] Compare the radian differences between the radian gradient regions to determine the stress magnitude;

[0070] Mark the stress transfer direction and stress magnitude on the main roller radian distribution map, and superimpose the marked stress transfer path on the main roller radian distribution map to generate the radian force application point.

[0071] In the embodiment of the present invention, extract the dataset of the main roller radian change rate calculated in step S2. This dataset contains the radian change rate values at 21 axial positions and 16 circumferential positions on the main roller surface. Based on these data, use the isoline segmentation method to partition the main roller surface, and set the radian deviation range to 0.02 mm, that is, the radian difference between adjacent regions is fixed at 0.02 mm. In specific operations, first determine the maximum and minimum radian values on the main roller surface. For a standard main roller with a diameter of 1200 mm, the radian change range is usually 0.85 - 1.45×10 -2 mm -1 , and based on this, divide the entire range into about 8 regions, and the radian deviation within each region does not exceed 0.02 mm. During the partitioning process, use the bilinear interpolation algorithm to calculate the radian values between grid points to ensure the continuity of the region boundaries. When drawing the isoline map, mark different regions with different colors. For example, the region of 0.85 - 0.87 is blue, the region of 0.87 - 0.89 is cyan, and so on, until the region of 1.43 - 1.45 is red. Finally, generate a layered main roller radian distribution map that clearly shows the boundaries of each radian gradient region. The map contains 8 regions in total, which are respectively marked as D1 to D8. Determine the measurement point grid within each region. For a main roller with a diameter of 1200 mm and a length of 2000 mm, set a measurement point every 100 mm along the axial direction and every 45 degrees along the circumferential direction to form a 21×8 measurement point matrix. Calculate the radian difference Δκ ij =|κ i+1j -κ ij | for each pair of adjacent measurement points (i, j) and (i + 1, j), where κ ij represents the radian value at point (i, j). At the same time, measure the actual physical distance d ij between adjacent measurement points. For axially adjacent points, the distance is 100 mm; for circumferentially adjacent points, the distance is πD / 8≈471 mm, where D is the main roller diameter. Calculate the radian - distance difference ratio β ij =Δκ ij / d ij, in units of mm / cm. For example, at two measurement points at axial positions of 600 mm and 700 mm within region D3, the radian values are 0.92×10 -2 mm -1 and 1.05×10 -2 mm -1 , the radian difference is 0.13×10 -2 mm -1 , the distance is 100 mm, and the calculated radian - distance difference ratio is 0.013 mm / cm. The same calculation is performed for all measurement point pairs to generate a complete radian - distance difference ratio data matrix. For each radian - distance difference ratio β ij , a judgment is made: when β ij > 0.01 mm / cm, the corresponding measurement point pair is marked as the starting position of the potential stress - bearing point. In an actual main roller test case, it is found that the radian - distance difference ratio within the axial position range of 500 - 600 mm reaches 0.013 mm / cm, and within the axial position range of 1400 - 1500 mm, the radian - distance difference ratio reaches 0.014 mm / cm. These two regions exceed the determination threshold of 0.01 mm / cm. At the same time, regional connectivity analysis is carried out to connect adjacent measurement points with high radian - distance difference ratios into continuous regions, and finally two starting regions of the radian stress - bearing points are determined: the first region ranges from 450 - 650 mm axially and 315 - 45 degrees circumferentially, and the center position is approximately (550 mm, 0 degrees); the second region ranges from 1350 - 1550 mm axially and 135 - 225 degrees circumferentially, and the center position is approximately (1450 mm, 180 degrees). High - light marks are added to the determined starting regions of the stress - bearing points, and the region range, center position, and maximum radian - distance difference ratio are recorded in the data table. In this embodiment, the two determined starting regions of the stress - bearing points are located at axial positions of 550 mm and 1450 mm. Comparing with the total length of the main roller of 2000 mm, these two positions are respectively within the first - end section (0 - 700 mm) and the second - end section (1300 - 2000 mm) of the main roller, that is, the deviation sections are close to both ends of the roller press. According to the stress transfer rule, when the deviation section of the main roller is close to one end of the roller press, the stress transfer direction is from this end to the other end. Therefore, the stress transfer direction of the first region (at 550 mm) is determined to be from the first end to the middle, and the stress transfer direction of the second region (at 1450 mm) is determined to be from the second end to the middle. Using the vector representation method, the stress transfer vector of the first region is (0, 0, 1), indicating along the positive Z - axis direction (from the left end to the right); the stress transfer vector of the second region is (0, 0, - 1), indicating along the negative Z - axis direction (from the right end to the left). The above conclusion of the stress transfer direction is verified by the stress distribution gradient, and it is found that the stress gradient from both ends to the middle is significantly higher than other directions. Calculate the average radian value κ av g = Σκ ij / n, where n is the number of measurement points within the region. For adjacent radian gradient regions D m and D n , calculate the radian difference Δκ mn =|κ av g, m -κ av g, n |. Divide the radian difference between regions by the average distance d mn between regions to obtain the radian gradient value G mn =Δκ mn / d mn . The radian gradient value is directly related to the stress magnitude. Use the linear mapping function S = k·G to convert the gradient value to the standardized stress magnitude, where k is the proportionality coefficient, calibrated through experiments to be k = 5×10 4 N·cm. In this embodiment, the radian difference at the starting region of the first force application point (the junction of D3 and D4) is 0.10×10 -2 mm -1 , and the corresponding average distance between regions is 90 mm. The calculated radian gradient value is 0.011×10 -2 mm -1 / cm, which is converted to a stress magnitude of approximately 550 N; the radian difference at the starting region of the second force application point (the junction of D5 and D6) is 0.12×10 -2 mm -1 , and the corresponding average distance between regions is 85 mm. The calculated radian gradient value is 0.014×10 -2 mm -1 / cm, which is converted to a stress magnitude of approximately 700 N. These values indicate that the stress borne by the second force application point is approximately 27% greater than that of the first force application point, which is consistent with the actual measured pressure distribution. On the basic radian distribution map, mark the stress transfer direction with arrows of different colors. The red arrow indicates the transfer direction from left to right at the first force application point (at 550 mm), and the blue arrow indicates the transfer direction from right to left at the second force application point (at 1450 mm). The arrow lengths are set proportionally according to the stress magnitude. The arrow length at the first force application point is set to 55 mm, and the arrow length at the second force application point is set to 70 mm. Draw circular marks at the starting positions of the arrows. The diameter size represents the influence range of the force application point. The mark diameter at the first force application point is 120 mm, and the mark diameter at the second force application point is 140 mm. Use the ray tracing algorithm to generate the stress transfer path, project the stress from the force application point along the transfer direction, and draw the path width according to the stress magnitude attenuation law: W(d) = W0·exp(-αd), where W0 is the initial width, d is the distance from the force application point, and α is the attenuation coefficient (taking the value 0.005 mm -1 ). Overlay the generated stress transfer path with the original radian distribution map, and set the transparency to 40% to ensure that the underlying radian distribution is clearly visible.

[0072] Preferably, in step S3, importing the main roll space parameters and the radian force application points into a mechanical analysis software to construct an initial state model of the main roll of the roll press includes:

[0073] Importing the main roll space parameters and the radian force application points into a mechanical analysis software, and constructing a three-dimensional geometric model of the main roll according to the main roll space parameters;

[0074] Setting the surface of the main roll as force-bearing grid nodes according to the three-dimensional geometric model of the main roll, determining the fixed end, free end and support structure of the main roll, and marking the positions and dimensions of each key section of the main roll;

[0075] Setting the radian force application points as load application points, and by setting the stress transfer parameters between the nodes;

[0076] Carrying out force correlation on the radian force application points on the surface of the main roll; if the main roll is in a full-load state, the force distribution is evenly connected; if the main roll is in a half-load or light-load state, the force distribution is uneven, so as to construct an initial state model of the main roll of the roll press.

[0077] In the embodiment of the present invention, the main roller space parameters and the marked arc force points are used as input data and imported into the ANSYS mechanical analysis system to construct the main roller three-dimensional geometric model. The import process first converts the main roller space parameters into a geometric data file in IGES format to ensure that the geometric accuracy is maintained during data transmission. 441 geometric control points are used for conversion, and the control point distribution density is 21 points in the axial direction × 21 points in the circumferential direction to form a complete NURBS surface description. After importing into the ANSYS working environment, the DesignModeler module is used to create a solid main roller model. The specific operations include: first, a cylindrical coordinate system is established based on the main roller axis as the reference, and the axis direction is defined as the Z axis; then the main roller surface is constructed according to the imported NURBS surface data; then the main roller solid model is generated by scanning operation (Sweep), and the journal structure is added at both ends at the same time, and the journal diameter is 40% of the main roller diameter and the length is 30% of the diameter; finally, internal structural features are added, including cooling channels (axial channels with a diameter of 25 mm, 30 mm from the surface) and keyways (width is 8% of the main roller diameter and depth is 4% of the diameter). The completed three-dimensional geometric model accurately reflects the outer dimensions and internal structure of the main roller, retains the slight changes in the surface curvature, and the model accuracy is controlled within the range of ±0.02mm. Based on the constructed three-dimensional geometric model of the main roller, the ANSYS Mechanical module is used to set the force grid nodes. The meshing adopts a hexahedron-dominated hybrid mesh method, and the following detailed parameters are used in the mesh generation process: the main roller surface mesh size is set to 10mm, the mesh in the curvature force point area is encrypted to 5mm, and the internal mesh size is gradually increased to 30mm; the surface mesh generation adopts surface fitting technology to ensure that the curvature capture accuracy is more than 95% of the original curvature; the mesh quality control parameters include skewness <0.85, orthogonal quality >0.65. After the mesh is generated, the number of typical units in the main roller model reaches 180,000, and the number of nodes is about 250,000. When setting the boundary conditions, first determine the fixed end, free end and support structure of the main roller: the driving end (usually the left end) is set as the fixed end, which is achieved by fixing all degrees of freedom of the end face of the journal (UX = UY = UZ = ROTX = ROTY = ROTZ = 0); the non-driving end is set as the free end, allowing axial displacement but limiting radial displacement (UX = UY = ROTX = ROTY = 0); the support structure adopts bearing constraints, and elastic support is applied to the outer surface of the journal, with an elastic coefficient of 5×109N / m. When marking the key sections of the main roller, it is determined according to the actual size: for the main roller with a total length of 2000mm, the two end sections are 700mm each (0-700mm and 1300-2000mm), and the central section is 600mm (700-1300mm); set an independent section number for each section. During implementation, create a load application node group at the corresponding position in the ANSYS working environment.The load application adopts the multi-point constraint (MPC) technique. First, virtual nodes are created at the radian force application points, and then the virtual nodes are connected to the surrounding solid mesh nodes through rigid beam elements (BEAM188). For each radian force application point on the main roll surface, three key parameters are set: the load magnitude \(F_i\) (unit: N), the load direction vector \(n_i\) (unit vector), and the load influence area radius \(R_i\) (unit: mm). The load magnitude is determined according to the radian gradient area where the force application point is located, and the calculation formula is \(F_i = k\cdot\Delta G_i\cdot A_i\), where \(k\) is the proportionality coefficient (typical value is \(1.2\times10^7\ N / mm\)), \(\Delta G_i\) is the radian gradient value (unit: \(mm^{-1}\)), and \(A_i\) is the force application point influence area (unit: \(mm^2\)). 2 ) The load direction vector is determined based on the stress transfer direction. When the stress transfers from the middle to both ends, is the angle with the axis, usually taking values from \(15^{\circ}\) to \(30^{\circ}\); when the stress transfers from one end to the other end, \(n_i = [0, 0, \pm1]\). The load influence area radius is determined according to the radian-distance difference ratio data, \(R_i = 50 + 200\cdot|\delta_i|\), where \(\delta_i\) is the radian-distance difference ratio value (unit: \(mm / cm\)). The setting of the stress transfer parameters between nodes includes: the contact stiffness coefficient \(K_c = 2\times10^8\ N / m\) 3, the friction coefficient μ = 0.15, the stress attenuation coefficient α = 0.85, and the stress transfer distance threshold d_th = 3·R_i. Based on the working state of the main roller, the force correlation of the surface radian stress points is carried out to construct a complete mechanical model of the initial state. Under the full-load state, the force distribution is evenly connected, and the operation method is as follows: First, divide a regular grid of 64 circumferential units × 21 axial units on the surface of the main roller; then assign a uniform pressure P_uniform = F_total / A_total to each grid cell, where F_total is the total load (usually 6×105 N), and A_total is the total surface area of the main roller; then correct the load in the area near the stress point, and the correction coefficient is c_i = 1 + 0.4·sin(πd_i / R_i), where d_i is the distance from the grid cell to the nearest stress point; finally, ensure that the pressure change rate between adjacent cells is less than 15% through the pressure continuity smoothing algorithm. Under the half-load state, the force is unevenly distributed, and a segmented pressure mode is adopted: The circumferential direction of the main roller is divided into a stressed area and a non-stressed area. The stressed area covers a range of 180°, and the pressure value is 1.6 times that of the full-load state; the pressure value of the non-stressed area is 0.4 times that of the full-load state; a 30° transition zone is set between the two areas, and a cosine transition function P_trans = P_loaded·(1 + cos(θ)) / 2 + P_unloaded·(1 - cos(θ)) / 2 is used, where θ ∈ [0, π]. Under the light-load state, the force is concentrated in a local area, and a Gaussian distribution model is used to describe the pressure distribution: P(x, y) = P_max·exp(-(x 2 / 2σ_x 2 +y 2 / 2σ_y 2 ))), where P_max is the maximum pressure value (usually 2.5 times the full-load pressure), and σ_x and σ_y are the distribution parameters in the axial and circumferential directions, respectively, with a value of 1 / 3 of the load application area. After the model is constructed, perform an initial balance calculation to ensure that the force system satisfies the static equilibrium condition, and adjust the local load until the difference between the resultant force and the reaction force is less than 0.5% of the total load.

[0078] Preferably, the material-main roller stress parametric simulation in step S3 based on the roller press material data and the initial state model of the main roller of the roller press includes:

[0079] Obtain the hardness value, density value, moisture content, and particle size distribution of the roller-pressed material as material characteristic parameter data according to the roller press material data;

[0080] Extract the target particle size distribution, maximum particle size limit, and minimum particle size limit as grinding task specification data according to the roller press material data;

[0081] Match the operating parameters of the roller press according to the grinding task specification data;

[0082] Set the discrete element parameters of the material for the initial state model of the main roller of the roller press through the material characteristic parameter data, including the material particle size distribution, cohesion coefficient, friction coefficient, and elastic modulus, and construct a material-roller surface interaction simulation model for the rolling process; among them, the material particle size distribution follows the Rosin-Rammler distribution, and the friction coefficient includes the particle-particle friction coefficient and the particle-roller surface friction coefficient;

[0083] Set the initial radian condition of the main roller, and set the initial radian to the standard straight line state, that is, the radian deviation is 0 mm; set the working mode of the main roller to continuous pressing, and set the pressing pressure to 500-2000 kN / m and the main roller linear speed to 0.5-1.5 m / s;

[0084] Start the calculation function of the dynamic analysis software, input the operating parameters of the roller press into the material-roller surface interaction simulation model for parametric simulation of the material-main roller stress, and obtain the simulated main roller stress state data.

[0085] In the embodiment of the present invention, a standard sampling technique is used to obtain real-time material samples, and the sample size is one-thousandth of the hourly throughput of the production line, at least not less than 10 kg. Perform a standard test process on the obtained samples: the hardness value is measured by a Mohs hardness tester, and 10 random point measurements are made on the material, and the arithmetic mean is taken, and the measurement accuracy is ±0.2 level; the density value is measured by the volume displacement method. Put the material sample with known mass into a graduated cylinder, add water with known volume and record the total volume change. The calculation formula is ρ = m / (V2 - V1), where m is the sample mass, V1 is the initial water volume, and V2 is the total volume after adding the sample, accurate to ±0.01 g / cm 3 ; The moisture content is measured by the drying method. Take 100 g of the material sample and dry it to a constant weight at 105 ± 2 °C. The calculation formula is w = (m1 - m2) / m1 × 100%, where m1 is the mass before drying and m2 is the mass after drying, accurate to ±0.1%; the particle size distribution is measured by the standard sieving method. Use 10 layers of vibrating sieves (sieve hole sizes are 50 mm, 30 mm, 20 mm, 10 mm, 5 mm, 3 mm, 1 mm, 0.5 mm, 0.25 mm, 0.125 mm) for 15 minutes with an amplitude of 2 mm and a vibration frequency of 50 Hz. Record the sieve fractions of each layer and calculate the cumulative percentage of undersize, and draw a particle size distribution curve. For standard cement raw meal, the typical hardness value is 4-6 levels, the density value is 2.6-3.0 g / cm 3 , the moisture content is 3-8%, d 50 The particle size is 5-15 mm. For cement raw meal grinding, extract the target particle size distribution parameters from the enterprise production process regulations: the target particle size distribution is usually described by the Rosin-Rammler equation, and the equation form is R(d) = 100·exp[-(d / d') n, where R(d) is the cumulative sieve residue percentage of materials with particle size greater than d, d' is the characteristic particle size, and n is the distribution uniformity index. The parameters d' and n are determined by least squares fitting of the actual screening data. For standard cement raw material processing, the target d' value is 0.08 - 0.12 mm, and the n value is 0.8 - 1.2; the maximum particle size limit is taken as the particle size corresponding to R(d) = 5% in the screening curve, usually controlled at 30 - 40 mm; the minimum particle size limit is taken as the particle size corresponding to R(d) = 95%, usually controlled at 0.002 - 0.005 mm. For ore grinding, the target particle size class also needs to consider the feeding requirements of subsequent sorting equipment, and the typical target is that 70 - 80% passes through a 0.074 mm sieve hole. In actual operation, the piecewise linear interpolation method is used to obtain the particle size value corresponding to a specific sieve residue percentage in the discrete screening data. Establish a mapping relationship matrix between material properties and roll pressing parameters. For the extracted target particle size distribution, calculate the index ratio coefficient λ = d 50 / (d_max - d_min), where d 50 is the particle size corresponding to 50% cumulative undersize, and d_max and d_min are the maximum and minimum particle size limits respectively. Based on the λ value, determine the linear pressure P_line of the roll press, and the calculation formula is P_line = P_base·(1 + 2.5·λ), where P_base is the reference linear pressure, and for cement raw material, it takes the value of 950 kN / m. The main roll linear speed v is determined according to the material hardness index H_i, and the calculation formula is v = v_ref·(H_ref / H_i)^0.4, where v_ref is the reference speed of 1.2 m / s and H_ref is the reference hardness of 5 levels. The roll gap spacing s is calculated according to the target maximum particle size d_max, s = 0.6·d_max. For a maximum particle size limit of 30 mm, the roll gap is set to 18 mm. The feeding rate Q is calculated according to the geometric parameters and operating parameters of the roll press, Q = 60·π·D·L·s·ρ·v·η_fill, with the unit of t / h, where D is the main roll diameter (m), L is the main roll length (m), s is the roll gap spacing (m), ρ is the material density (t / m 3), where v is the linear velocity (m / s), η_fill is the roll gap filling coefficient, with a value range of 0.4 - 0.6, and the specific value is determined by the empirical formula η_fill = 0.38 + 0.05·w + 0.01·H_i, where w is the moisture content of the material (%). Create a roll press - material interaction model in the EDEM discrete element simulation environment. The particle size distribution of the material is set to follow the Rosin - Rammler distribution, and the mathematical expression is f(d) = 100·n·(d^(n - 1) / d'^n)·exp[-(d / d')^n], where the values of d' and n are obtained from the measured particle size distribution. When creating the material particle model, set the particle shape as a combination of polyhedrons. Each single particle is composed of 3 - 7 basic spheres, and the combination method is center packing type, with an overlap rate of 25 - 35% to simulate the irregular shape of the material. The cohesion coefficient of the material is calculated according to the moisture content w, and the formula is c = c0·(1 + 5·w), where c0 is the reference cohesion, with a value of 1.2×10 4 N / m 2 . The particle - particle friction coefficient μ_pp is set to is the internal friction angle of the material, determined by direct shear test. For standard cement raw meal, μ_pp = 0.45 - 0.60; the particle - roll surface friction coefficient μ_pw is set according to the roll surface material and roughness. For standard cemented carbide roll surfaces, μ_pw = 0.35 - 0.45. The elastic modulus E_m of the material is determined by compression test, and the calculation formula is E_m = σ / ε, where σ is the applied stress and ε is the relative deformation. For cement raw meal, the typical value is 2.0 - 4.5 GPa. The Poisson's ratio of the material is set to 0.25 - 0.30, and the restitution coefficient is set to 0.4 - 0.65, and the specific values are determined by bounce test. Through geometric transformation, reset the main roll surface to the standard straight state, that is, the radian deviation is 0 mm. During the reset process, perform radial adjustment on each point P(r, θ, z) on the main roll surface so that where is the average radius value at the corresponding axial position z. During specific operations, all the node coordinates on the surface grid of the main roller are extracted, a mapping table between the radial position and the axial position is established, the ideal cylindrical surface equation is fitted using the least squares method, the radial deviation of each node is calculated, and then it is adjusted to the ideal cylindrical surface by applying node displacements. The working mode of the main roller is set to the continuous pressing mode, and the specific working parameters are set according to the actual working conditions: the pressing pressure is determined by the ratio of the hydraulic cylinder force F to the effective length L of the main roller, P_line = F / L, and the setting range is 500 - 2000 kN / m. The specific value is set in stages, 600 kN / m in the start-up stage, 1200 kN / m in the transition stage, and 1600 kN / m in the stable operation stage; the linear speed v of the main roller is calculated according to the rotational speed n of the main roller and the diameter D of the main roller, v = π·D·n / 60, and the setting range is 0.5 - 1.5 m / s, corresponding to the rotational speed of 8 - 23 rpm of the main roller with a typical diameter of 1.2 m. After all the working parameters are set, a pre-load balance calculation is performed on the model to ensure that the system is in a static equilibrium state, and the pre-load balance convergence criterion is that the residual force is less than 0.1% of the total load. The operating parameters of the roller press are input into the material-roller surface interaction simulation model, and the material-main roller stress parametric simulation is executed. The simulation process sets multi-physical field coupling calculations: the discrete element module is responsible for simulating the interaction between material particles and their contact with the roller surface, and the time step is set to 2×10 -5 seconds to ensure calculation stability; the finite element module is responsible for calculating the stress distribution and deformation of the main roller under the action of the load, and the explicit integration method is used, with a time step of 1×10 -4 seconds. The two modules are coupled through interface data exchange, and the exchange frequency is to exchange data every 100 discrete element time steps. The following key data are recorded during the calculation process: the stress distribution σ(θ, z, t) on the surface of the main roller, including the radial stress σ_r, the circumferential stress σ_θ, and the axial stress σ_z; the deformation displacement u(θ, z, t) of the main roller, including the radial component u_r, the circumferential component u_θ, and the axial component u_z; the load distribution F(θ, z, t) of the main roller, reflecting the contact force distribution between the material and the roller surface; the material flow velocity field v(x, y, z, t), describing the motion state of the material in the roll gap area. The simulation calculation continues until the system reaches a steady state, and the determination criterion is that the change rate of the maximum deformation of the main roller is less than 1% within 10 consecutive calculation cycles. The finally obtained simulation main roller stress state data includes: the stress time history curve, the deformation time history curve, the stress distribution contour map, and the deformation distribution contour map of each monitoring point of the main roller, and the data accuracy is controlled within the range of ±0.5%.

[0086] Especially importantly, the material-main roller stress parametric simulation further includes:

[0087] The material is divided into a coarse particle zone, a medium particle zone, and a fine particle zone according to the material characteristic parameter data, and the proportion of different particle size zones is calculated; among them, the coarse particle zone is set to 10 - 30 mm, the medium particle zone is set to 5 - 10 mm, and the fine particle zone is set to 0 - 5 mm;

[0088] The material is divided into a high moisture zone, a medium moisture zone, and a low moisture zone according to the material characteristic parameter data, and the distribution of different moisture zones is analyzed; among them, the high moisture zone is set to 15 - 20%, the medium moisture zone is set to 8 - 15%, and the low moisture zone is set to 0 - 8%;

[0089] The material is divided into a high hardness zone, a medium hardness zone, and a low hardness zone according to the material characteristic parameter data, and the distribution of different moisture zones is analyzed; among them, the high hardness zone is set to 7 - 10 grades, the medium hardness zone is set to 4 - 6 grades, and the low hardness zone is set to 1 - 3 grades;

[0090] Create a simulation working condition matrix, including at least 27 different working condition combinations, where the simulation working condition matrix is specifically composed of 3 material particle sizes × 3 moisture contents × 3 hardnesses;

[0091] Conduct dynamic simulation analysis on each working condition combination, set the simulation duration to 60 seconds, and record the main roll bending deformation data every 0.5 seconds;

[0092] Extract the maximum deformation amount, average deformation amount, and deformation amount change rate of each monitoring point of the main roll during the simulation process to obtain the simulation main roll stress state data.

[0093] In the embodiments of the present invention, the materials are divided into three regions according to the particle size range: the materials with a particle size range of 10 - 30 mm are divided into the coarse particle region, the materials with a particle size range of 5 - 10 mm are divided into the medium particle region, and the materials with a particle size range of 0 - 5 mm are divided into the fine particle region. Calculate the proportion of each particle size region, expressed in mass fraction, and the calculation formula is \(W_i=(m_i / m_{total})\times100\%\), where \(W_i\) is the mass percentage of the \(i\)-th particle size region, \(m_i\) is the material mass of the \(i\)-th particle size region, and \(m_{total}\) is the total material mass. When calculating specifically, use the screening test data to read the cumulative undersize percentage corresponding to the key particle sizes from the particle size cumulative distribution curve: \(P1\) is the cumulative undersize percentage corresponding to 5 mm, \(P2\) is the cumulative undersize percentage corresponding to 10 mm, and \(P3\) is the cumulative undersize percentage corresponding to 30 mm. Then the proportion of the fine particle region is \(P1\%\), the proportion of the medium particle region is \((P2 - P1)\%\), the proportion of the coarse particle region is \((P3 - P2)\%\), and the proportion of the ultra - coarse particles larger than 30 mm is \((100 - P3)\%\). For standard cement raw meal, the typical distribution is that the fine particle region accounts for 35 - 45%, the medium particle region accounts for 25 - 35%, the coarse particle region accounts for 20 - 30%, and the ultra - coarse particles account for 0 - 5%. In the discrete element model, create the corresponding number of particles according to these proportions to ensure that the particle size distribution in the simulation is consistent with the actual materials. Use a Karl Fischer moisture analyzer to accurately measure the moisture content of the material samples, and the measurement accuracy is ±0.1%. Divide the materials into three moisture regions according to the moisture content range: the materials with a moisture content of 15 - 20% are divided into the high - moisture region, the materials with a moisture content of 8 - 15% are divided into the medium - moisture region, and the materials with a moisture content of 0 - 8% are divided into the low - moisture region. Conduct stratified sampling and analysis of the samples to determine the spatial distribution of different moisture regions. Use a columnar sampler to extract samples from different positions and depths of the material pile, record the three - dimensional coordinates \((x, y, z)\) of each sampling point and measure the moisture content \(w(x, y, z)\). Based on the measurement data, construct a three - dimensional moisture distribution model and use the trilinear interpolation algorithm to calculate the moisture content at any position: \(w(x, y, z)=\sum (i＝1) 8$w_i\cdot N_i(x, y, z)$, where $w_i$ is the moisture content value of 8 adjacent sampling points, and $N_i(x, y, z)$ is the shape function, and the calculation formula is $N_i(x, y, z) = (1\pm\xi)(1\pm\eta)(1\pm\zeta) / 8$, where $\xi$, $\eta$, and $\zeta$ are normalized coordinates. In the cement raw meal processed under standard conditions, the typical distribution is that the high moisture area accounts for 5 - 15%, the medium moisture area accounts for 40 - 60%, and the low moisture area accounts for 30 - 50%. The hardness is measured on 100 randomly selected material particles, the measurement load is 500 g, and the holding time is 15 s. Record the hardness value of each measurement point and establish a frequency distribution histogram of the hardness values. The materials are divided into three hardness zones according to the Mohs hardness scale: the materials with a hardness of 7 - 10 are divided into the high hardness zone, the materials with a hardness of 4 - 6 are divided into the medium hardness zone, and the materials with a hardness of 1 - 3 are divided into the low hardness zone. Calculate the proportion of each hardness zone, using the frequency statistics method, and the calculation formula is $P_i=n_i / N\times100\%$, where $P_i$ is the percentage of the $i$-th hardness zone, $n_i$ is the number of particles in the $i$-th hardness zone, and $N$ is the total number of test particles. For materials with multiple mineral combinations, X-ray diffraction (XRD) analysis is used to determine the mineral composition, and then the weighted average hardness is calculated according to the standard hardness values and content ratios of each mineral. Analyze the spatial distribution of different hardness regions, study its correlation with mineral components and particle sizes, and establish a hardness distribution probability density function $f(H)=dP / dH$, where $P$ is the cumulative percentage and $H$ is the hardness value. In the standard cement raw meal, the typical hardness distribution is that the high hardness zone accounts for 15 - 25%, the medium hardness zone accounts for 50 - 70%, and the low hardness zone accounts for 15 - 25%. The dimension of the simulation working condition matrix is 3×3×3, corresponding to different levels of three factors: material particle size (dominated by coarse particle zone, dominated by medium particle zone, dominated by fine particle zone), moisture content (high moisture, medium moisture, low moisture), and hardness (high hardness, medium hardness, low hardness). The material particle size combinations are designed as follows: dominated by coarse particle zone (coarse particle zone 60%, medium particle zone 25%, fine particle zone 15%), dominated by medium particle zone (coarse particle zone 25%, medium particle zone 50%, fine particle zone 25%), dominated by fine particle zone (coarse particle zone 15%, medium particle zone 25%, fine particle zone 60%). The moisture content combinations are designed as follows: high moisture type (average moisture content 17%, range 15 - 20%), medium moisture type (average moisture content 11%, range 8 - 15%), low moisture type (average moisture content 4%, range 0 - 8%). The hardness combinations are designed as follows: high hardness type (average hardness 8, range 7 - 10), medium hardness type (average hardness 5, range 4 - 6), low hardness type (average hardness 2, range 1 - 3). Each working condition combination is marked as "(P_i)(W_j)(H_k)", for example, "(P1)(W2)(H3)" represents the combination of dominated by coarse particle zone, medium moisture, and low hardness.Set unified operating parameters for the roller press for all 27 working conditions: the main roller speed is set to 15 rpm, corresponding to a linear speed of approximately 0.95 m / s; the pressing pressure is set to 1200 kN / m; the roll gap is set to 15 mm. Establish a working condition - parameter mapping table. For each monitoring point, extract three key deformation indicators: the maximum deformation u_max, defined as the maximum absolute value of the radial displacement of this point during the entire simulation process, and the calculation formula is u_max = max|u_r(t)|, t ∈ [0, 60 s]; the average deformation u_avg, defined as the time average of the radial displacement of this point during the stable operation stage (30 - 60 s), and the calculation formula is u_avg = (1 / T)∫. 30 60 u_r(t)dt, where T = 30 s; the deformation rate of change R_u, defined as the ratio of the standard deviation of the deformation to the average value during the stable stage, and the calculation formula is R_u = σ_u / |u_avg|×100%, where σ_u = √[(1 / T)∫ 30 60 (u_r(t) - u_avg) 2 dt]. Conduct a spatial distribution analysis on the deformation data of all monitoring points, draw a deformation nephogram, with the horizontal axis being the axial position and the vertical axis being the circumferential position, and the color representing the magnitude of the deformation. Calculate the overall bending degree index of the main roller: the axial bending index B_axial, defined as the ratio of the difference between the average deformation at both ends and the deformation at the center to the diameter of the main roller, and the calculation formula is B_axial = |(u_ends - u_center) / D|×100%, where u_ends is the average deformation of the monitoring points at both ends, u_center is the deformation of the central monitoring point, and D is the diameter of the main roller.

[0094] As an example of the present invention, refer to Figure 2 shown in Figure 1 is a detailed implementation step flow diagram of step S4 in

[0095] Step S41: Continuously monitor the extrusion pressure values at the middle and both ends of the main roller, and calculate the difference between the extrusion pressure value at the middle of the main roller and the extrusion pressure values at both ends to obtain the main roller extrusion pressure distribution data;

[0096] In an embodiment of the present invention, when continuously monitoring the extrusion force values at the middle and both ends of the main roll, 11 high-precision pressure sensors are arranged axially along the working surface of the main roll. Among them, 5 sensors are arranged in the middle region (within 30% of the middle of the main roll length), numbered C1 to C5; 3 sensors are arranged in each of the two end regions (within 35% from both ends of the main roll). The left end is numbered L1 to L3, and the right end is numbered R1 to R3. The sampling frequency of each sensor is set to 200 Hz, and the data acquisition system records the pressure time series P_i(t), where i is the sensor number. The sensors are installed 3 mm below the surface of the main roll, and the measurement accuracy is ±0.5%. Calculate the average extrusion force value in the middle region P_center = (P_C1 + P_C2 + P_C3 + P_C4 + P_C5) / 5, and the average extrusion force value in the two end regions P_ends = (P_L1 + P_L2 + P_L3 + P_R1 + P_R2 + P_R3) / 6. When calculating the difference, use the extrusion force difference between the middle and the two ends ΔP = P_center - P_ends and the extrusion force difference between the left and right ends ΔP_LR = (P_L1 + P_L2 + P_L3) / 3 - (P_R1 + P_R2 + P_R3) / 3. Normalize the difference, ΔP_norm = ΔP / P_avg × 100%, where P_avg is the average extrusion force across the entire roll surface. Record the change curve of ΔP_norm over time in real-time, and calculate the mean value μ_ΔP and standard deviation σ_ΔP under steady-state conditions (running time greater than 30 minutes).

[0097] Step S42: Mark the extrusion force concentration areas for the extrusion force distribution data of the main roll, and divide the extrusion force concentration areas into extrusion force peak points;

[0098] In an embodiment of the present invention, set the high extrusion force threshold T_high = P_avg + 1.5σ_P, and the low extrusion force threshold T_low = P_avg - 1.5σ_P, where P_avg is the average extrusion force and σ_P is the standard deviation of the extrusion force. Judge the extrusion force value P(z_i) at each measurement position z_i: when P(z_i) is greater than T_high, mark this position as a high-pressure area; when P(z_i) is less than T_low, mark this position as a low-pressure area; mark the remaining positions as normal-pressure areas. Continuous high-pressure areas form extrusion force concentration areas. Calculate the range (starting position z_start to ending position z_end), length L_conc = z_end - z_start, average pressure P_conc_avg, and maximum pressure P_conc_max for each concentration area, and divide the extrusion force concentration areas into extrusion force peak points.

[0099] Step S43: Incrementally monitor the force parameters around the extrusion force peak points at a scanning interval of 5 cm according to the extrusion force peak points to obtain the main roll force diffusion data;

[0100] In the embodiments of the present invention, for each identified peak position z_peak of the extrusion force, measurement points are set at intervals of 5 cm to the left and right of this point: z_peak ± 5 cm, z_peak ± 10 cm, z_peak ± 15 cm, etc., until the extrusion force value drops to 20% of the peak value or reaches the end of the main roll. At each measurement point z_i, the following force parameters are recorded: the extrusion force value P(z_i), the radial deformation δ_r(z_i), the circumferential strain ε_θ(z_i), and the axial strain ε_z(z_i). The radial deformation is measured by a displacement sensor, and the strain value is measured by a surface strain gauge.

[0101] Step S44: Identify the diffusion gradient of the main roll force diffusion data, and evaluate the force distribution of the diffusion gradient to obtain the main roll force distribution data;

[0102] In the embodiments of the present invention, a force diffusion spatial distribution diagram is constructed. The horizontal axis is the axial position z of the main roll, the vertical axis is the diffusion parameter (extrusion force P, radial deformation δ_r, circumferential strain ε_θ, axial strain ε_z), and the height of the surface represents the magnitude of the parameter value. Cluster analysis is performed on the gradient field, and the K-means clustering algorithm is used to divide the surface of the main roll into k gradient regions (usually k = 3 - 5), and the gradient characteristics within each region are similar. Evaluate the force distribution of the diffusion gradient, and calculate the following indicators: the gradient uniformity index where and are the standard deviation and mean of the gradient magnitude respectively; the direction consistency index The value range is [0, 1], and the larger the value, the more consistent the gradient direction; the diffusion symmetry index where α_left and α_right are the diffusion coefficients on the left and right sides of the peak point respectively.

[0103] Step S45: Cluster the abnormal stress points of the main roll arc according to the simulated main roll stress state data to obtain the potential main roll arc deformation region;

[0104] In the embodiments of the present invention, key parameters are extracted from the stress state data: von Mises stress σ_vm, principal stresses σ_1, σ_2, σ_3, stress ratio σ_r = σ_1 / σ_3, and stress gradient Set the abnormal stress determination criterion: σ_vm is greater than 0.6σ_yield (where σ_yield is the material yield strength) or σ_r is greater than 5 or Greater than 10 MPa / mm. Points that meet any of the conditions are marked as abnormal stress points, forming an initial set of abnormal points P = {p_1, p_2,..., p_n}. The density-based spatial clustering algorithm is used for abnormal point clustering, and each cluster represents a potential arc deformation region. Calculate the characteristic indexes for each cluster: area A_i (the surface area covered by the cluster), average stress σ_avg_i (the average von Mises stress of the points within the cluster), maximum stress σ_max_i (the maximum von Mises stress within the cluster), and stress concentration factor.

[0105] Step S46: Evaluate the deformation degree of the potential main roll arc deformation region based on the main roll extrusion force distribution data, and divide it into a severely deformed region and a slightly deformed region;

[0106] In the embodiment of the present invention, the regional average stress σ_avg (weight w_σ = 0.3), maximum deformation δ_max (weight w_δ = 0.25), stress concentration factor K_t (weight w_K = 0.2), deformation duration t_def (weight w_t = 0.15), and regional area A (weight w_A = 0.1). Each index is scored on a 0-10 scale. According to the comprehensive score, the deformation region is divided into two categories: severely deformed region (S ≥ 6.5) and slightly deformed region (3.5 ≤ S < 6.5). Regions with scores lower than 3.5 do not require special treatment. Generate a detailed evaluation report for each region, including the regional boundary coordinates, center position, coverage range, scores of each index, and comprehensive score. Apply a red marking line at the boundary of the severely deformed region and a yellow marking line at the slightly deformed region, and display them as highlighted regions on the 3D model of the main roll.

[0107] Step S47: Perform main roll bending compensation according to the severely deformed region and the slightly deformed region to obtain the main roll bending compensation parameters; among them, increase the compensation force of the hydraulic support device in the severely deformed region by 20 - 50 kN, and increase the compensation force of the hydraulic support device in the slightly deformed region by 5 - 20 kN.

[0108] In the embodiments of the present invention, the numbers of the support devices within the influence range of each deformation region are determined. For severely deformed regions, the central position z_center of the region is identified, and the two nearest hydraulic support devices i and i + 1 are found. The distance ratio λ = (z_center - z_i) / (z_{i + 1} - z_i) is calculated, where z_i and z_{i + 1} are the axial positions of the two support devices. The compensation force distribution coefficients k_i = 1 - λ and k_{i + 1} = λ are determined. The total compensation force required for the severely deformed region is calculated as F_comp_severe = E·I·κ_max·L / (δ_max·c), where E is the elastic modulus, I is the cross-sectional moment of inertia, κ_max is the maximum curvature, L is the length of the deformation region, δ_max is the maximum deformation, and c is the compensation coefficient (with a value range of 1.2 - 1.5). The compensation force is determined according to the deformation amount by the formula ΔF (kN) = 20 + 30×(δ_max / δ_allow). For slightly deformed regions, a similar method is used, but the compensation force is reduced by adding a compensation force of 5 - 20 kN, and the calculation formula is ΔF (kN) = 5 + 15×(δ_max / δ_allow). The compensation force is applied through a hydraulic control system.

[0109] Particularly importantly, before the bending compensation of the main roll, it further includes:

[0110] Marking the hardware units of the main roll of the roll press, where the hardware units include the main roll body module, the hydraulic support unit module, and the pressure sensor module;

[0111] Analyzing the working surface width, rated pressure range, and radial deformation tolerance value of the main roll body according to the main roll body module;

[0112] Extracting the support position distribution data and the support piston stroke amount according to the hydraulic support unit module;

[0113] Performing a position - stroke capacity mapping on the support piston stroke amount through the support position distribution data to obtain a position - stroke mapping value;

[0114] Based on the position - stroke mapping value, performing a pressure level adaptability detection on the rated pressure range, and then parameterizing the adaptation relationship of the hydraulic support unit module through the pressure level adaptability to generate hydraulic support adaptation data;

[0115] Performing a pressure detection accuracy analysis on the pressure sensor module and performing a constraint relationship mapping to generate pressure sensor constraint data;

[0116] Determining the rated radian limit according to the working surface width of the main roll body and the radial deformation tolerance value to obtain the main roll radian limit data;

[0117] Construct a main roll bending compensation control mechanism based on hydraulic support adaptation data, pressure sensor constraint data, and main roll radian limit data.

[0118] In the embodiment of the present invention, the hardware unit of the main roll of the roll press is accurately marked, and the hardware unit is divided into a main roll body module, a hydraulic support unit module, and a pressure sensor module, and a unique identification code is assigned to each module. Subsequently, a morphological analysis is carried out on the main roll body module through precise measuring equipment to determine the effective width of the working surface of the main roll, calculate the rated pressure bearing range of the main roll according to the material strength characteristics, and at the same time calculate the allowable limit value of the radial deformation of the main roll under working conditions through the elastic deformation theory. Then, for the hydraulic support unit module, extract the accurate distribution position data of each support unit in the axial direction of the main roll, measure the maximum stroke of each support piston, and record the rated support force and working pressure values of each support unit. The piecewise linear function method is used to establish a mapping relationship between the support position and the piston stroke to form a continuous position-stroke mapping curve, ensuring that each position point in the axial direction of the main roll has a corresponding value of the support capacity. Based on the obtained position-stroke mapping values, a dynamic pressure response test is carried out, and the ratio of the actual support pressure to the theoretical required pressure is recorded under various working conditions to detect the pressure level adaptability, and the hydraulic support system is parameterized and adjusted accordingly to generate a complete hydraulic support adaptation data table. At the same time, a standard pressure calibration is performed on the pressure sensor module to detect the linearity, repeatability, and hysteresis error of each sensor, establish an accurate mapping relationship between the sensor readings and the actual pressure, and form a pressure sensor constraint data matrix. According to the geometric characteristics and material properties of the main roll, calculate and determine the rated radian limit of the main roll under various working conditions to generate a main roll radian limit data table. Finally, integrate the hydraulic support adaptation data, the pressure sensor constraint data, and the main roll radian limit data to construct a closed-loop control system with a triple protection mechanism, realize the real-time monitoring and accurate compensation of the main roll radian state, and ensure that the main roll always maintains the best working radian during the roll pressing process.

[0119] Preferably, the present invention also provides a parametric simulation system for the main roll radian of a roll press, which executes the parametric simulation method for the main roll radian of a roll press as described above. The parametric simulation system for the main roll radian of a roll press includes:

[0120] A main roll radian scanning module, which is used to perform a surface space scan on the main roll of the roll press to obtain the main roll space parameters; perform multiple pre-measurements on the working conditions of the main roll of the roll press to generate the main roll operation state data;

[0121] A radian analysis module, which is used to divide the main roll section according to the main roll operation state data, and then perform a main roll radian deviation analysis to obtain the main roll radian deviation data; calculate the main roll radian change rate according to the main roll radian deviation data; determine the actual working state of the main roll, and perform a main roll radian distribution evaluation according to the main roll radian deviation data to obtain the main roll radian distribution map;

[0122] A stress simulation module, which is used to divide the main roller radian distribution diagram into regions according to the main roller radian change rate, and mark the radian stress points; import the main roller spatial parameters and the radian stress points into mechanical analysis software to construct an initial state model of the main roller of the roller press; obtain roller pressing material data; perform material-main roller stress parametric simulation according to the roller pressing material data and the initial state model of the main roller of the roller press to obtain simulated main roller stress state data;

[0123] A bending compensation module, which is used to perform intelligent main roller bending compensation based on the simulated main roller stress state data to obtain main roller bending compensation parameters.

[0124] Preferably, the present invention also provides a computer-readable storage medium storing a computer program, and when the computer program is executed, it implements the parametric simulation method for the main roller radian of the roller press as described in any one of the above.

[0125] The present invention can achieve active control of the main roller state of the roller press, avoiding secondary damage or performance degradation caused by improper compensation. Practical applications show that after adopting this method, the product qualification rate is increased by more than 15%, the energy consumption is reduced by about 12%, the service life of the main roller is extended by more than 30%, and the annual economic benefit is significantly improved. At the same time, this method reduces the dependence on manual experience, reduces the maintenance cost, improves the equipment reliability, provides technical support for the intelligent and refined operation management of the roller press, and has broad application prospects for promotion. Therefore, the parametric simulation method for the main roller radian of the present invention significantly improves the control level of the main roller radian deviation through the closed-loop logic of real-time monitoring, precise analysis, simulation prediction and intelligent compensation, greatly improves the stability and consistency of the product quality, effectively reduces the energy consumption caused by uneven pressure, significantly extends the service life of the main roller and related components by reducing local overload and optimizing the force, and at the same time reduces the unplanned downtime and maintenance cost, ultimately improving the efficiency and economic benefit of the entire roller pressing production process.

[0126] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, it is intended to cover all changes falling within the meaning and scope of the equivalent elements of the application documents within the present invention.

[0127] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features invented herein.

Claims

1. A parametric simulation method for the arc of the main roller of a roller press, characterized in that, It includes the following steps: Step S1: Perform surface space scanning on the main roller of the roll press to obtain the main roller space parameters; perform multiple pre-measurements on the working conditions of the main roller of the roll press to generate the main roller operation state data; Step S2: Divide the main roller into sections according to the main roller operation state data, and then perform the main roller radian deviation analysis to obtain the main roller radian deviation data; calculate the main roller radian change rate according to the main roller radian deviation data; determine the actual working state of the main roller, and evaluate the main roller radian distribution according to the main roller radian deviation data to obtain the main roller radian distribution map; Step S3: Divide the main roller radian distribution map into regions through the main roller radian change rate, and mark the radian stress points; import the main roller space parameters and the radian stress points into the mechanical analysis software to construct the initial state model of the main roller of the roll press; obtain the roll-pressed material data; Perform material-main roller stress parametric simulation according to the roll-pressed material data and the initial state model of the main roller of the roll press to obtain the simulated main roller stress state data; Step S4: Perform intelligent main roller bending compensation based on the simulated main roller stress state data to obtain the main roller bending compensation parameters.

2. The parametric simulation method for the main roller arc of a roller press according to claim 1, characterized in that Step S1 includes the following steps: Step S11: Perform surface space scanning on the main roller of the roll press through a laser ranging device to obtain the coordinates of the main roller surface in three-dimensional space. The scanning accuracy is ±0.05 mm, the scanning spacing is 10 mm axially and 5° radially, and record it as the main roller surface coordinate data; Step S12: Determine the axial position and radial dimension of the main roller according to the main roller surface coordinate data; Step S13: Encode the geometric information of the main roller of the roll press through the axial position and radial dimension to obtain the main roller space parameters; Step S14: Based on the main roller of the roll press, 7 measurement point positions are evenly arranged along the main roller axis, and 4 measuring points are set along the circumferential direction of the main roller surface at each measurement point position to obtain the main roller measurement point data; Step S15: Deploy temperature sensors, eddy current displacement sensors and torque strain gauges according to the main roller measurement point data to construct the main roller measurement network; Step S16: Use the main roller measurement network to perform multiple pre-measurements on the working conditions of the main roller of the roll press to obtain the original main roller measurement data; among them, the multiple pre-measurements of the working conditions include three acquisitions. The first acquisition is carried out 10 seconds after the roll press starts without load, and the acquisition duration is 5 seconds; the second acquisition is carried out 10 seconds after the material enters the roll gap after loading, and the acquisition duration is 5 seconds; the third acquisition is carried out after stable operation for 5 minutes, and the acquisition duration is 10 seconds; the original main roller measurement data includes the main roller speed value, the main roller torque value and the main roller temperature value; Step S17: Perform digital filtering processing on the original main roller measurement data to generate the main roller operation state data.

3. The parametric simulation method for the main roll curvature of a roll press according to claim 1, characterized in that, In step S2, the division of the main roller into sections according to the main roller operation state data and then the main roller radian deviation analysis includes: Divide the main roller operation state data into the main roller axial and the main roller surface circumferential measurement data to obtain the main roller axial measurement data and the main roller surface circumferential measurement data respectively; Divide the main roll measurement points based on the spatial parameters of the main roll. The main roll is divided into 3 sections according to the axial position, namely the two end sections and the central section, to obtain each section of the main roll. Among them, the length of the two end sections is 35% of the total length of the main roll, and the length of the central section is 30% of the total length of the main roll. Calculate the axial radian value of each section of the main roll according to the axial measurement data of the main roll. Analyze the circumferential torque value of each section of the main roll according to the circumferential measurement data on the surface of the main roll. Perform time series alignment of the temperature values for the circumferential measurement data on the surface of the main roll to obtain the time series data of the circumferential temperature values. Calculate the radian difference between the central area and the two end areas of the main roll based on the axial radian value of each section of the main roll to obtain the central-two end radian difference value. Conduct sectional numerical comparison according to the time series data of the circumferential temperature values and the circumferential torque value of each section of the main roll, and arrange them according to the numerical gradient to obtain the main roll temperature-torque gradient data. Integrate the central-two end radian difference value with the main roll temperature-torque gradient data in the main roll radian deviation data to obtain the main roll radian deviation data.

4. The parametric simulation method for the main roller arc of a roller press according to claim 1, wherein, In step S2, divide the main roll sections according to the main roll operation state data, and then conduct the main roll radian deviation analysis, including: Conduct time series analysis on the central-two end radian difference value in the main roll radian deviation data, and calculate the change rate of the radian difference value of the main roll at different time points to obtain the main roll radian change rate. Judge whether there is an axis deviation of the main roll according to the main roll spatial parameters, and mark the sections with a deviation exceeding 0.05 mm to obtain the main roll deviation section data. Determine the actual working state of the main roll, and divide the working state into three states: full load, half load, and light load. Based on the main roll deviation section data and the main roll temperature-torque gradient data in the main roll radian deviation data, measure the radian deformation amount of the main roll under the three working states to obtain the main roll radian deformation data. Determine the local radian data of the main roll according to the main roll radian deformation data. Conduct radian field statistics on the main roll local radian data and detect the radian field distribution data. Import the radian field distribution data into a drawing tool, and map the data points into a two-dimensional coordinate system, where the coordinate axes represent the physical position and the radian value of the main roll respectively, to obtain the main roll radian distribution map.

5. The parametric simulation method for the main roller curvature of a roller press according to claim 1, characterized in that In step S3, divide the main roll radian distribution map through the main roll radian change rate and mark the radian stress points, including: Divide the main roll radian distribution map according to the radian deviation range through the main roll radian change rate to obtain the radian gradient area. Among them, set the radian deviation range of each area to 0.02 mm. Measure the radian difference between adjacent measurement points within each radian gradient area, and calculate the ratio of the radian difference to the distance between adjacent measurement points to obtain the radian-distance difference ratio data. If the radian-distance difference ratio data is greater than 0.01 mm / cm, it is initially determined that the area corresponding to the radian-distance difference ratio data is the starting area of the radian stress point. Determine the radian stress transmission direction based on the starting area of the radian stress point. If the main roll deviation section is close to the middle of the roll press, the stress transmission direction is from the middle to the two ends; if the main roll deviation section is close to one end of the roll press, the stress transmission direction is from this end to the other end. Determine the stress magnitude by comparing the radian differences between the radian gradient regions; Mark the stress transfer direction and stress magnitude on the main roller radian distribution diagram, and superimpose the marked stress transfer path on the main roller radian distribution diagram to generate radian force application points.

6. The parametric simulation method for the main roll curvature of a roll press according to claim 1, characterized in that, In step S3, importing the main roller spatial parameters and radian force application points into the mechanical analysis software to construct the initial state model of the main roller of the roller press includes: Import the main roller spatial parameters and radian force application points into the mechanical analysis software, and construct a three-dimensional geometric model of the main roller according to the main roller spatial parameters; Set the surface of the main roller as force-bearing grid nodes according to the three-dimensional geometric model of the main roller, determine the fixed end, free end and support structure of the main roller, and mark the positions and dimensions of each key section of the main roller; Set the radian force application points as load application points, and set the stress transfer parameters between the nodes; Perform force correlation on the radian force application points on the surface of the main roller; if the main roller is in a full-load state, the force distribution is evenly connected; if the main roller is in a half-load or light-load state, the force distribution is uneven, so as to construct the initial state model of the main roller of the roller press.

7. The parametric simulation method for the main roller curvature of a roller press according to claim 1, characterized in that, In step S3, performing material-main roller stress parametric simulation according to the roller-pressed material data and the initial state model of the main roller of the roller press includes: Obtain the hardness value, density value, moisture content, and particle size distribution of the roller-pressed material as material characteristic parameter data according to the roller-pressed material data; Extract the target particle size distribution, maximum particle size limit, and minimum particle size limit as the grinding task specification data according to the roller-pressed material data; Match the operating parameters of the roller press according to the grinding task specification data; Perform material discrete element parameter setting on the initial state model of the main roller of the roller press through the material characteristic parameter data, including the material particle size distribution, cohesion coefficient, friction coefficient, and elastic modulus, to construct a material-roller surface interaction simulation model of the roller pressing process; among them, the material particle size distribution follows the Rosin-Rammler distribution, and the friction coefficient includes the particle-particle friction coefficient and the particle-roller surface friction coefficient; Set the initial radian condition of the main roller, and set the initial radian to the standard straight state, that is, the radian deviation is 0 mm; set the working mode of the main roller to continuous pressing, and set the pressing pressure to 500 - 2000 kN / m and the main roller linear speed to 0.5 - 1.5 m / s; Start the calculation function of the mechanical analysis software, input the operating parameters of the roller press into the material-roller surface interaction simulation model for material-main roller stress parametric simulation, and obtain the simulation main roller stress state data.

8. The parametric simulation method for the main roller arc of a roller press according to claim 1, wherein Step S4 includes the following steps: Step S41: Continuously monitor the extrusion force values at the middle and both ends of the main roller, and calculate the difference between the extrusion force value at the middle of the main roller and the extrusion force values at both ends to obtain the main roller extrusion force distribution data; Step S42: Mark the extrusion force concentration areas for the main roller extrusion force distribution data, and divide the extrusion force concentration areas into extrusion force peak points; Step S43: Monitor the force parameters around the extrusion force peak points at an increment of 5 cm as the scanning interval according to the extrusion force peak points to obtain the main roller force diffusion data; Step S44: Identify the diffusion gradient of the main roller force diffusion data, and evaluate the force distribution of the diffusion gradient to obtain the main roller force distribution data; Step S45: Cluster the abnormal stress points of the main roll arc based on the simulation main roll stress state data to obtain the potential main roll arc deformation area; Step S46: Evaluate the deformation degree of the potential main roll arc deformation area based on the main roll extrusion pressure distribution data, and divide it into a severe deformation area and a slight deformation area; Step S47: Perform main roll bending compensation according to the severe deformation area and the slight deformation area to obtain the main roll bending compensation parameters; among them, increase the compensation force of the hydraulic support device in the severe deformation area by 20 - 50 kN, and increase the compensation force of the hydraulic support device in the slight deformation area by 5 - 20 kN.

9. A parametric simulation system for the main roll arc of a roller press, characterized in that, A parameterized simulation system for the main roll arc of a roller press for implementing the parameterized simulation method of the main roll arc of a roller press as described in claim 1, the parameterized simulation system for the main roll arc of the roller press comprising: A main roll arc scanning module, configured to perform surface space scanning on the main roll of the roller press to obtain main roll space parameters; perform multiple pre-measurements on the working conditions of the main roll of the roller press to generate main roll operating state data; An arc analysis module, configured to divide the main roll sections according to the main roll operating state data, and then perform main roll arc deviation analysis to obtain main roll arc deviation data; calculate the main roll arc change rate according to the main roll arc deviation data; determine the actual working state of the main roll, and perform main roll arc distribution evaluation according to the main roll arc deviation data to obtain the main roll arc distribution map; A stress simulation module, configured to divide the main roll arc distribution map into regions through the main roll arc change rate, and mark the arc stress points; import the main roll space parameters and the arc stress points into a mechanical analysis software to construct an initial state model of the main roll of the roller press; obtain roller pressing material data; perform material-main roll stress parameterized simulation according to the roller pressing material data and the initial state model of the main roll of the roller press to obtain simulation main roll stress state data; A bending compensation module, configured to perform intelligent main roll bending compensation based on the simulation main roll stress state data to obtain main roll bending compensation parameters.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed, it implements the parameterized simulation method of the main roll arc of a roller press as described in any one of claims 1 to 8.

Images