Adaptive control system and method for intelligent roller centrifuge

Abstract

The present invention discloses an adaptive control system and method for an intelligent roller centrifuge, belonging to the field of roller centrifuges. The control system comprises: real-time acquisition of multi-source heterogeneous data during the operation of the roller centrifuge and transmission to a central control platform; flow uniformity detection of the molten metal in the roller centrifuge; when uneven flow of the molten metal is detected, a correction step length of the roller centrifuge speed is obtained based on the viscosity-temperature curve of the molten metal material, a roller centrifuge speed instruction is generated, the roller centrifuge speed is controlled, the molten metal pouring flow rate is synchronously controlled, and abnormal vibration monitoring is performed during the roller centrifuge speed regulation process. If potential abnormal vibration is detected, the abnormal vibration is corrected. The control system solves the problem of realizing coordinated control of the pouring speed and rotation speed of the roller centrifuge under complex working conditions, while suppressing vibration interference caused by rotation, ensuring uniform distribution of the molten metal and consistency of performance of the inner and outer layers of the roller.

Description

Technical Field

[0001] The present invention relates to the field of roller centrifuges, and more particularly to an adaptive control system and method for an intelligent roller centrifuge. Background Art

[0002] Roller centrifuges are core equipment in the metallurgical, mining, and heavy machinery manufacturing sectors, used to produce high-performance rolls with composite structures, such as high-chromium cast iron rolls and high-speed steel rolls. Using centrifugal casting technology, molten metal is poured into a high-speed rotating mold, where centrifugal force is used to evenly distribute the molten metal and solidify it layer by layer, forming a composite roll with a high-hardness outer layer and a high-toughness inner layer. However, as industrial requirements for roll performance increase, the control technology of traditional roller centrifuges has gradually exposed the following key issues:

[0003] Existing roller centrifuges typically use independent control of the pouring speed and rotation speed, lacking a dynamic coordination mechanism. For example, when rotating at high speeds, if the pouring speed does not match the changes in centrifugal force in real time, the molten metal is prone to splashing or uneven distribution, resulting in large deviations in the thickness of the outer layer of the roller and significant fluctuations in hardness. In addition, the viscosity differences between different materials further exacerbate the complexity of process parameter adaptation, requiring repeated debugging based on manual experience, which is inefficient. When the roller centrifuge rotates at high speeds, vibrations caused by factors such as dynamic imbalance of the mold and impact of the molten metal flow can cause loose material or porosity defects in the inner layer. Traditional hydraulic or mechanical balancing systems have high response delays and cannot achieve real-time vibration compensation. In severe cases, they can even cause bearing damage or mold cracking, increasing equipment maintenance costs.

[0004] While existing technologies have proposed some improvements to address these issues, such as improving single-variable control accuracy by adjusting proportional, integral, and differential parameters, their effectiveness in collaborative control of multivariable coupled systems is limited. Furthermore, the use of rubber cushions or dampers to reduce vibration transmission fails to eliminate the vibration source and compromises the equipment's load capacity. Therefore, to overcome these limitations, the present invention proposes an adaptive control system and method for an intelligent roller centrifuge. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the purpose of the present invention is to provide an adaptive control system and method for an intelligent roller centrifuge, which solves the problem of how to achieve precise coordinated control of the casting speed and rotation speed of the roller centrifuge under complex working conditions, while suppressing the vibration interference caused by high-speed rotation, ensuring uniform distribution of molten metal and consistency of performance of the inner and outer layers of the roller.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The adaptive control system of the intelligent roller centrifuge includes:

[0008] Real-time collection of multi-source heterogeneous data during the operation of the roller centrifuge and transmission to the central control platform. The multi-source heterogeneous data includes mold inner wall pressure data, spindle vibration amplitude data, spindle vibration spectrum data, molten metal temperature data, and speed data.

[0009] Perform multi-dimensional correlation analysis based on the material property database, including: detecting the flow uniformity of the molten metal in the roller centrifuge according to the mold inner wall pressure data. If uneven flow of the molten metal is detected, the viscosity of the molten metal is obtained based on the viscosity-temperature curve of the molten metal material, and the correction step size of the roller centrifuge speed is calculated to generate the roller centrifuge speed instruction;

[0010] The speed command of the roller centrifuge is parsed to control the speed of the roller centrifuge. By establishing a dynamic relationship equation between the pouring speed and the speed, the pouring flow rate of the molten metal is synchronously controlled. During the speed control process of the roller centrifuge, abnormal vibration is monitored based on the vibration spectrum data of the roller centrifuge main shaft. If potential abnormal vibration is detected, the eccentric phase angle is identified and the abnormal vibration is corrected by gradually adding counterweights to the roller centrifuge.

[0011] In addition, multi-dimensional fault warnings are carried out in real time during the operation of the roller centrifuge, including abnormal pressure warnings, abnormal metal liquid warnings, abnormal speed warnings and abnormal vibration warnings, and the multi-dimensional fault warning information is displayed through the central control platform.

[0012] Specifically, the specific steps of multidimensional correlation analysis include:

[0013] When the central control platform receives a set of multi-source heterogeneous data, it encodes and marks the multi-source heterogeneous data according to the sensor ID and timestamps the multi-source heterogeneous data according to the collection time;

[0014] The labeled multi-source heterogeneous data are grouped according to data type, and outlier detection is performed on each group of multi-source heterogeneous data. The detected outliers are marked, and the mean of each group of data is used to correct the outliers.

[0015] Obtain the mold inner wall pressure data after outlier correction, extract its statistical features, and configure a variance threshold. If the variance of the mold inner wall pressure data after outlier correction is greater than the variance threshold, it is determined that there is a potential pressure anomaly in the roller centrifuge and the flow uniformity of the metal liquid in the roller centrifuge is tested. Otherwise, no action is taken.

[0016] When it is determined that the roller centrifuge metal liquid flow uniformity test fails, the metal liquid temperature data in the roller centrifuge is obtained, and the viscosity-temperature curve of the corresponding metal liquid material is retrieved from the material property database according to the characteristic code of the metal liquid material; based on the metal liquid temperature data in the roller centrifuge, the current metal liquid viscosity is calculated using an interpolation algorithm.

[0017] Specifically, the specific steps of multi-dimensional correlation analysis also include:

[0018] Determine the critical speed formula and combine it with the current viscosity of the molten metal to obtain the critical speed of the roller centrifuge ;

[0019] Setting limit thresholds , combined with the critical speed of the roller centrifuge, the safe speed range of the roller centrifuge is limited to [0, ];

[0020] Obtain the current roller centrifuge speed data to determine whether it is within the safe speed range of the roller centrifuge. If it is within the safe speed range of the roller centrifuge, obtain the speed correction step size based on the target viscosity of the molten metal and gradually correct the roller centrifuge speed.

[0021] Generate a step-by-step correction instruction for the speed of the roller centrifuge according to the current speed data of the roller centrifuge and the speed correction step size;

[0022] Otherwise, a safety deceleration time is set, and based on the current roller centrifuge speed data, a speed stop step is calculated to generate a roller centrifuge speed stop instruction to terminate the operation of the roller centrifuge.

[0023] Specifically, the specific steps of performing flow uniformity detection on the molten metal in the roller centrifuge include:

[0024] Divide the inner wall of the roller centrifuge mold into regions, mark each region with a number, and obtain the coordinates of the center point of each numbered region;

[0025] According to the coding marks of the piezoelectric sensors, the area where the piezoelectric sensors are located is determined, the piezoelectric sensors in each numbered area are grouped, and the mold inner wall pressure data collected by each group of piezoelectric sensors is obtained;

[0026] According to the distance between the position coordinates of the piezoelectric sensor in each numbered area and the coordinates of the center point of the area, an inverse proportional function is used to determine the position weighting coefficient;

[0027] For each area, the mold inner wall pressure mean data of each area is calculated by weighted averaging based on the mold inner wall pressure data collected by the piezoelectric sensor in the area and the corresponding position weighting coefficient.

[0028] Specifically, the specific steps of performing flow uniformity detection on the molten metal in the roller centrifuge also include:

[0029] Define the neighborhood range of the region. For each region, calculate the pressure difference between it and the average pressure data of the mold inner wall of each region in the neighborhood. Record the value of each pressure difference and correspond it to the region number in the neighborhood.

[0030] Configure a pressure difference threshold. If the pressure difference between a region and the average mold inner wall pressure data of all neighboring regions is greater than the pressure difference threshold, the region is marked as an abnormal region and is judged to have failed the roller centrifuge metal liquid flow uniformity test. Otherwise, the roller centrifuge metal liquid flow uniformity test is judged to have passed.

[0031] Specifically, the specific steps of synchronously controlling the metal liquid pouring flow rate include:

[0032] Based on the mold volume, molten metal viscosity, molten metal density and the coupling coefficient between pouring speed and rotation speed of the roller centrifuge, a dynamic relationship equation between pouring speed and rotation speed is established;

[0033] Based on the type of molten metal, the historical multi-source heterogeneous data in the central control platform is screened to fit the coupling coefficient of the pouring speed and rotation speed in the dynamic relationship equation between the pouring speed and rotation speed;

[0034] receiving a speed command for a roller centrifuge, and if it is a speed step-by-step correction command, parsing the speed step-by-step correction command to obtain a speed correction step, a target speed, and an adjustment time interval;

[0035] According to the speed correction step and the adjustment time interval, the speed of the roller centrifuge is gradually updated in real time, and the metal liquid pouring flow rate is obtained according to the dynamic relationship equation between the pouring speed and the speed after fitting, and the pouring flow rate instruction is generated;

[0036] According to the pouring flow instruction, the PID controller adjusts the pouring valve opening, changes the flow cross-sectional area of ​​the molten metal, and controls the change of the molten metal pouring flow of the roller centrifuge;

[0037] If it is a speed stop command, a stop pouring command is generated and the pouring valve is closed.

[0038] Specifically, the specific steps of abnormal vibration monitoring include:

[0039] When receiving the speed instruction of the roller centrifuge and controlling the speed of the roller centrifuge, abnormal vibration monitoring is started;

[0040] According to the type of molten metal, the vibration amplitude data of the roller centrifuge spindle when the roller centrifuge speed is not regulated is filtered from the central control platform, and the data is cleaned to remove abnormal data points and establish a historical vibration amplitude database;

[0041] Perform fast Fourier transform on the spindle vibration amplitude data in the historical vibration amplitude database to obtain the main frequency component, amplitude and harmonic components of the corresponding spindle vibration spectrum data;

[0042] According to the mean and variance of the amplitude of the main shaft vibration spectrum data in the historical vibration amplitude database, the amplitude threshold is set, including the upper amplitude threshold and the lower amplitude threshold;

[0043] Perform fast Fourier transform on the real-time collected spindle vibration amplitude data to obtain the real-time main frequency component, real-time amplitude and real-time harmonic component of the real-time spindle vibration spectrum data;

[0044] If the real-time amplitude is greater than the upper amplitude threshold or less than the lower amplitude threshold, it is determined that there is potential abnormal vibration and abnormal vibration correction is performed; otherwise, no processing is performed.

[0045] Specifically, the specific steps for correcting abnormal vibration include:

[0046] When it is determined that there is potential abnormal vibration, the theoretical rotation fundamental frequency is calculated based on the current speed of the roller centrifuge;

[0047] Configure an eccentricity threshold. If the absolute value of the difference between the real-time main frequency component of the main shaft vibration spectrum data and the theoretical rotation fundamental frequency is greater than the eccentricity threshold, it is determined that an eccentric vibration source main frequency exists. The phase difference of each axis at the theoretical rotation fundamental frequency is obtained to determine the eccentricity phase angle. Otherwise, no processing is performed.

[0048] In the flat angle direction of the eccentric phase angle, the roller centrifuge counterweights are gradually added, and based on the real-time amplitude of the main shaft vibration spectrum data after the counterweights are added, the change in the real-time amplitude before and after the addition of the counterweights is calculated to obtain the attenuation rate of the real-time amplitude;

[0049] If the attenuation rate of the real-time amplitude increases after adding the counterweight of the roller centrifuge, the abnormal vibration monitoring should be carried out again;

[0050] Otherwise, the attenuation rate of the real-time amplitude after the counterweights of the roller centrifuge are gradually added is monitored, and an attenuation threshold is configured. When the attenuation rate of the real-time amplitude of the main shaft vibration spectrum data is less than the attenuation threshold, the addition of the counterweights is stopped;

[0051] Configure the counterweight correction threshold, calculate the total weight of the counterweight blocks added to the roller centrifuge, and if it is greater than the counterweight correction threshold, stop the abnormal vibration correction and restart the abnormal vibration monitoring.

[0052] Specifically, the specific steps of multi-dimensional fault warning include:

[0053] When the data analysis module determines that there is a potential pressure anomaly in the roller centrifuge, a pressure anomaly warning is triggered. The pressure anomaly warning information includes the time when the pressure anomaly occurred, the sensor ID involved, and the range of the abnormal pressure data;

[0054] When the data analysis module determines that the molten metal flow uniformity test fails, a molten metal abnormality warning is triggered. The molten metal abnormality warning information includes the number of the abnormal area and the pressure difference value;

[0055] When the adaptive control module detects that the speed data of the roller centrifuge is not within the safe speed range, it triggers a speed abnormality warning. The speed abnormality warning information includes the roller centrifuge speed and the safe speed range;

[0056] When the adaptive control module determines that there is potential abnormal vibration, it triggers a vibration abnormality warning. The vibration abnormality warning information includes the time when the vibration abnormality occurs, the real-time amplitude, and the amplitude threshold range.

[0057] The adaptive control method of the intelligent roller centrifuge comprises the following steps:

[0058] Step S1: acquiring multi-source heterogeneous data in real time during the operation of the roller centrifuge, including mold inner wall pressure data, spindle vibration amplitude data, spindle vibration spectrum data, molten metal temperature data, and speed data, collecting data through sensors, and transmitting the collected multi-source heterogeneous data to the central control platform;

[0059] Step S2: performing a multi-dimensional correlation analysis based on the material property database, and detecting the flow uniformity of the molten metal in the roller centrifuge according to the mold inner wall pressure data. When uneven flow of the molten metal is detected, the viscosity of the molten metal is obtained according to the viscosity-temperature curve of the molten metal material, and the correction step length of the roller centrifuge speed is calculated to generate a roller centrifuge speed instruction, including a speed step correction instruction and a speed stop instruction.

[0060] Step S3: parsing the speed command of the roller centrifuge and controlling the speed of the roller centrifuge. By establishing a dynamic relationship equation between the pouring speed and the speed, the pouring flow rate of the molten metal is synchronously controlled. During the speed control process of the roller centrifuge, abnormal vibration monitoring is performed based on the vibration spectrum data of the roller centrifuge main shaft. If potential abnormal vibration is detected, the eccentric phase angle is identified and the abnormal vibration is corrected by gradually adding counterweights to the roller centrifuge.

[0061] Step S4: Perform multi-dimensional fault warning, including abnormal pressure warning, abnormal metal liquid warning, abnormal speed warning and abnormal vibration warning, and display the multi-dimensional fault warning information through the central control platform.

[0062] Beneficial effects of the present invention:

[0063] The adaptive control system of the intelligent roller centrifuge acquires multi-source heterogeneous data in real time, providing a comprehensive basis for data analysis. Through multi-dimensional correlation analysis, it detects the uniformity of molten metal flow and generates appropriate speed commands based on the viscosity-temperature curve, effectively ensuring the flow state of the molten metal. By establishing a dynamic relationship between pouring speed and speed, it synchronously controls the pouring flow rate of the molten metal. During speed control, it monitors the main shaft vibration spectrum data to promptly detect and correct abnormal vibrations, ensuring stable equipment operation. The fault warning module provides multi-dimensional fault warnings covering pressure, molten metal, speed, and vibration, and displays warning information on the central control platform, facilitating timely countermeasures and reducing the risk of failure. Overall, the adaptive control system and method of the intelligent roller centrifuge solve the problem of achieving precise coordinated control of the pouring speed and rotation speed of the roller centrifuge under complex operating conditions, while suppressing vibration interference caused by high-speed rotation, ensuring uniform distribution of molten metal and consistent performance of the inner and outer layers of the roller. This effectively improves the operational stability and product quality of the roller centrifuge, increases production efficiency, and reduces equipment failure rate and maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 Schematic diagram of the structure of the adaptive control system of the intelligent roller centrifuge of the present invention;

[0065] Figure 2 A flowchart of the specific steps of the multi-dimensional correlation analysis of the present invention;

[0066] Figure 3 This is a flow chart of the specific steps of the present invention for detecting the flow uniformity of the molten metal in the roller centrifuge;

[0067] Figure 4 This is a flow chart of the specific steps of synchronously controlling the pouring flow rate of molten metal according to the present invention;

[0068] Figure 5 A flowchart of specific steps for performing abnormal vibration correction according to the present invention;

[0069] Figure 6 This is a flow chart of the adaptive control method of the intelligent roller centrifuge of the present invention. DETAILED DESCRIPTION

[0070] Example 1:

[0071] See also Figure 1 ,This embodiment introduces the adaptive control system of the intelligent roller centrifuge, including a data acquisition module, a data analysis module, an adaptive control module and a fault warning module;

[0072] The data acquisition module is used to obtain multi-source heterogeneous data in the operation of the roller centrifuge in real time, including mold inner wall pressure data, spindle vibration amplitude data, spindle vibration spectrum data, molten metal temperature data and speed data. It collects data through sensors and transmits the collected multi-source heterogeneous data to the central control platform.

[0073] The sensors include: piezoelectric sensor: installed in the low-temperature area outside the mold near the inner wall, and the pressure of the inner wall of the mold is transmitted to the sensor through a specially designed conduction structure, so as to measure the pressure data of the inner wall of the mold in a high-temperature environment; three-axis MEMS accelerometer: installed on the main shaft support seat away from the high-temperature area of ​​the roller centrifuge, and indirectly captures the main shaft vibration amplitude data by monitoring the vibration of the main shaft support seat; infrared thermal imager: placed on the side of the roller centrifuge at a certain safe distance from the molten metal and at a lower temperature. Through a special optical reflection device, the thermal radiation of the molten metal is reflected into the detection range of the infrared thermal imager, thereby realizing non-contact measurement, monitoring the temperature distribution of the molten metal, and obtaining the temperature data of the molten metal; magnetoelectric encoder: installed on the side of the transmission chain of the roller centrifuge drum away from the high-temperature source, and accurately collects the speed data by monitoring the movement of the transmission chain.

[0074] In this embodiment, when the roller centrifuge is started, the data acquisition module begins collecting data from each sensor in real time at a set sampling frequency, acquiring multi-source, heterogeneous data from the roller centrifuge's operation. During the acquisition process, the collected data undergoes preliminary filtering and preprocessing to remove noise and outliers and ensure data validity. For example, a digital filtering algorithm is used to smooth pressure and vibration data to remove high-frequency noise; bad pixel detection and correction are performed on temperature data to ensure the integrity of the temperature field data. The preprocessed data is transmitted to the central control platform via wired or wireless transmission to ensure stable and real-time data transmission. During transmission, data verification and encryption technologies are used to ensure data accuracy and security, preventing data tampering or loss during transmission.

[0075] The data analysis module is used to perform multi-dimensional correlation analysis based on the material property database, and to detect the flow uniformity of the molten metal in the roller centrifuge according to the mold inner wall pressure data. When uneven flow of the molten metal is detected, the viscosity of the molten metal is obtained according to the viscosity-temperature curve of the molten metal material, and the correction step size of the roller centrifuge speed is calculated to generate the roller centrifuge speed instruction, including the speed step correction instruction and the speed stop instruction.

[0076] The material property database stores the physical, chemical and mechanical properties of various processed metal materials, including density, melting point, thermal expansion coefficient, viscosity-temperature curve, yield strength, tensile strength, hardness, and material feature coding library.

[0077] In this embodiment, the data analysis module relies on the material property database to conduct multi-dimensional correlation analysis on multi-source heterogeneous data from the central control platform, and uses the collected mold inner wall pressure data to detect the uniformity of the molten metal flow in the roller centrifuge. The pressure data can intuitively reflect the distribution state of the molten metal in the mold. By analyzing the changes and distribution of the pressure, it is determined whether the molten metal flow is uniform. If the detection finds that the molten metal flow is uneven, the viscosity-temperature curve of the corresponding molten metal material is retrieved from the material property database, and the current viscosity of the molten metal is selected in combination with the real-time collected molten metal temperature data. Based on the influence of the molten metal viscosity on the speed of the roller centrifuge, the speed of the roller centrifuge is corrected to ensure that the speed matches the characteristics of the molten metal and improve the flow state of the molten metal. Taking into account multiple factors such as the molten metal characteristics, the mold structure, and the operating parameters of the roller centrifuge, the pouring speed of the roller centrifuge is corrected to further ensure the uniform flow of the molten metal in the roller centrifuge.

[0078] See also Figure 2 Preferably, the specific steps of multi-dimensional correlation analysis include:

[0079] When the central control platform receives a set of multi-source heterogeneous data, it encodes and labels the data according to the sensor ID and timestamps the data according to the time of collection. Each sensor has a unique ID that represents its installation location and the type of data collected. For example, pressure sensors installed at different locations on the inner wall of a mold have different IDs, and the coding labels can clearly distinguish the pressure data at different locations. This facilitates subsequent analysis of data characteristics at different locations and accurately determines the distribution of the molten metal within the mold. The timestamp records the collection time of each data point, giving the data information a time dimension. When analyzing the flow state of molten metal, time series data can reflect the changing trend of the flow state over time, such as the pressure changes of the molten metal during the start-up, acceleration, and stable operation of the roller centrifuge.

[0080] The labeled multi-source heterogeneous data are grouped according to data type, and outlier detection is performed on each group of multi-source heterogeneous data. The detected outliers are marked, and the mean of each group of data is used to correct the outliers.

[0081] The mold inner wall pressure data after outlier correction is obtained, and its statistical features, including mean and variance, are extracted. A variance threshold is configured based on the normal operating parameter range, historical data, and production process requirements of the roller centrifuge. If the variance of the mold inner wall pressure data after outlier correction is greater than the variance threshold, it is determined that there is a potential pressure anomaly in the roller centrifuge, and the flow uniformity of the metal liquid in the roller centrifuge is tested. Otherwise, no processing is performed.

[0082] See also Figure 3 Preferably, the specific steps of performing flow uniformity detection on the metal liquid of the roller centrifuge include:

[0083] Based on the geometric shape of the roller centrifuge mold, the inner wall of the roller centrifuge mold is divided into regions, each region is numbered and labeled, and the coordinates of the center point of each numbered region are obtained to identify the data of each region. For cylindrical molds, the division can be performed according to the circumferential and axial directions. For example, the inner wall of the mold can be divided into several small regions by evenly dividing the circumferential direction into several sectors and the axial direction into several intervals of equal length. For the circumferential sector regions, their center point coordinates are calculated based on the geometric properties of the sectors. For the axial interval, since the division is along the mold axis, its center point coordinates coincide with the line connecting the mold center in the circumferential direction and are the midpoint of the interval in the axial direction.

[0084] According to the coding marks of the piezoelectric sensors, the area where the piezoelectric sensors are located is determined, the piezoelectric sensors in each numbered area are grouped, and the mold inner wall pressure data collected by each group of piezoelectric sensors is obtained;

[0085] According to the distance between the position coordinates of the piezoelectric sensor in each numbered area and the coordinates of the area center, the position weighting coefficient is determined using an inverse proportional function. The closer the sensor is to the area center, the greater the contribution of its measurement data to the average pressure value of the area, and the higher the position weighting coefficient; the farther the sensor is from, the lower the position weighting coefficient. When the distance between the piezoelectric sensor and the area center is When , the position weight coefficient for:

[0086]

[0087] in, is a very small constant used to avoid the denominator being zero;

[0088] For each area, the mean pressure data of the mold inner wall in the area is calculated by weighted averaging based on the mold inner wall pressure data collected by the piezoelectric sensor in the area and the corresponding position weighting coefficient, that is:

[0089]

[0090] in, is the average pressure data of the inner wall of the mold in the area, It is the first The pressure data of the inner wall of the mold collected by the piezoelectric sensor, It is the first The position weighting coefficient corresponding to the mold inner wall pressure data collected by the piezoelectric sensor, The value range is {1,2,3,..., }, is the total number of piezoelectric sensors in the area;

[0091] Based on the structure of the mold and the characteristics of the molten metal flow, the neighborhood range of the area is defined. For the area in the circumferential direction, the two adjacent left and right fan-shaped areas can be used as neighborhoods; for the area in the axial direction, the two upper and lower adjacent intervals can be used as neighborhoods. By defining the neighborhood range, the correlation relationship between the areas is established, so that when analyzing the uniformity of the molten metal flow, it is possible to expand from the local area to its surrounding related areas, more comprehensively consider the changes in the pressure distribution of the molten metal in the mold, and provide a more reference data range for the subsequent accurate judgment of whether the molten metal flow is uniform.

[0092] For each area, the pressure difference between it and the average pressure data of the mold inner wall of each area in the neighborhood is calculated in turn, the value of each pressure difference is recorded, and it is corresponding to the area number in the neighborhood; the pressure difference between different areas is quantified, and the pressure changes between each area are clearly and intuitively presented through the pressure difference value and the corresponding area number.

[0093] Configure a pressure difference threshold. If the pressure difference between a region and the average mold inner wall pressure data of all neighboring regions is greater than the pressure difference threshold, the region is marked as an abnormal region and is judged to have failed the roller centrifuge metal liquid flow uniformity test. Otherwise, the roller centrifuge metal liquid flow uniformity test is judged to have passed.

[0094] When it is determined that the roller centrifuge metal liquid flow uniformity test has failed, the metal liquid temperature data in the roller centrifuge is obtained, and the viscosity-temperature curve of the corresponding metal liquid material is retrieved from the material property database based on the characteristic code of the metal liquid material. The current viscosity of the metal liquid is calculated using an interpolation algorithm based on the metal liquid temperature data in the roller centrifuge. The greater the viscosity, the greater the flow resistance of the metal liquid, and the roller centrifuge speed needs to be increased to ensure its fluidity. Conversely, the smaller the viscosity, the lower the speed can be reduced to avoid problems such as uneven distribution of the metal liquid due to excessive centrifugal force.

[0095] According to the structural characteristics of the roller centrifuge, the critical speed formula is determined, and combined with the current viscosity of the molten metal, the critical speed of the roller centrifuge is obtained, that is:

[0096]

[0097] in, is the critical speed of the roller centrifuge, is the acceleration due to gravity, is the current viscosity of the molten metal, is the current density of the molten metal, is the roller centrifuge die volume;

[0098] Setting limit thresholds , combined with the critical speed of the roller centrifuge, the safe speed range of the roller centrifuge is limited to [0, The limit threshold value range is [0.7, 0.9] to avoid the vibration of the roller centrifuge caused by excessive speed, which will lead to increased uneven flow of molten metal;

[0099] The current roller centrifuge speed data is obtained to determine whether it is within the safe speed range of the roller centrifuge. If it is within the safe speed range of the roller centrifuge, the speed correction step size is obtained according to the target viscosity of the molten metal, and the roller centrifuge speed is gradually corrected, that is:

[0100]

[0101] in, is the speed correction step size, is the target viscosity of the molten metal, is the proportionality factor related to the roller centrifuge model;

[0102] Generate a step-by-step correction instruction for the speed of the roller centrifuge according to the current speed data of the roller centrifuge and the speed correction step size;

[0103] Otherwise, the safety deceleration time is set, and the speed stop step is calculated based on the current roller centrifuge speed data to generate the roller centrifuge speed stop instruction to terminate the roller centrifuge operation and avoid the risk of equipment damage or metal liquid splashing due to overspeed.

[0104] The adaptive control module is used to parse the speed command of the roller centrifuge and control the speed of the roller centrifuge. By establishing a dynamic relationship equation between the pouring speed and the speed, the metal liquid pouring flow is synchronously controlled. During the speed regulation of the roller centrifuge, abnormal vibration is monitored based on the vibration spectrum data of the roller centrifuge main shaft. If potential abnormal vibration is detected, the eccentric phase angle is identified and the abnormal vibration is corrected by gradually adding counterweight blocks to the roller centrifuge.

[0105] In this embodiment, the adaptive control module accurately interprets and executes the speed command for the roller centrifuge. First, the speed command is deeply analyzed to extract key parameters such as the speed correction step size, target speed, and adjustment interval. Based on a deep understanding of the flow characteristics of molten metal, mold structure, and production process, a dynamic coupling model for pouring speed and speed is constructed. This model senses speed changes in real time and, based on this coupling relationship, synchronously and precisely controls the pouring flow rate of the molten metal. This ensures that the pouring process and the speed of the roller centrifuge are optimally matched under various operating conditions, effectively improving product quality and production efficiency. Throughout the entire control process, the vibration spectrum data of the roller centrifuge's main shaft is continuously monitored in real time. The collected data is then compared and analyzed to enable rapid and accurate detection of abnormal vibrations. Once abnormal vibration is detected, the dominant frequency component of the eccentric vibration source is identified. Based on the identified eccentric phase angle, counterweights are added to the corresponding positions of the roller centrifuge, effectively correcting the abnormal vibration and ensuring the stable and safe operation of the roller centrifuge.

[0106] See also Figure 4 Preferably, the specific steps of synchronously controlling the metal liquid pouring flow rate include:

[0107] Based on the mold volume, molten metal viscosity, molten metal density, and the coupling coefficient between pouring speed and rotation speed of the roller centrifuge, the dynamic relationship equation between pouring speed and rotation speed is established, which is as follows:

[0108]

[0109] in, is the pouring speed, is the coupling coefficient between pouring speed and rotation speed, The temperature of the molten metal is The viscosity of the molten metal at is the roller centrifuge speed, is the current density of the molten metal, is the roller centrifuge die volume;

[0110] Based on the type of molten metal, the historical multi-source heterogeneous data in the central control platform is screened to fit the coupling coefficient of the pouring speed and rotation speed in the dynamic relationship equation between the pouring speed and rotation speed;

[0111] receiving a speed command for a roller centrifuge, and if it is a speed step-by-step correction command, parsing the speed step-by-step correction command to obtain a speed correction step, a target speed, and an adjustment time interval;

[0112] According to the speed correction step and the adjustment time interval, the speed of the roller centrifuge is gradually updated in real time, and the metal liquid pouring flow rate is obtained according to the dynamic relationship equation between the pouring speed and the speed after fitting, and the pouring flow rate instruction is generated;

[0113] According to the pouring flow instruction, the PID controller adjusts the pouring valve opening to change the flow cross-sectional area of ​​the molten metal, thereby controlling the change in the molten metal pouring flow of the roller centrifuge; ensuring that the pouring process of the molten metal and the speed of the roller centrifuge always maintain the best matching state.

[0114] If it is a speed stop command, a stop pouring command is generated, the pouring valve is closed, and the pouring of molten metal is stopped to prevent leakage of molten metal or other safety problems during the shutdown process.

[0115] Preferably, the specific steps of abnormal vibration monitoring include:

[0116] When receiving the speed instruction of the roller centrifuge and controlling the speed of the roller centrifuge, abnormal vibration monitoring is started;

[0117] Based on the type of molten metal, the vibration amplitude data of the roller centrifuge spindle when the roller centrifuge speed is not regulated is filtered from the central control platform, and the data is cleaned to remove abnormal data points to establish a historical vibration amplitude database;

[0118] The spindle vibration amplitude data in the historical vibration amplitude database is subjected to fast Fourier transform to obtain the main frequency component, amplitude and harmonic components of the corresponding spindle vibration spectrum data; the main frequency component, which reflects the main frequency component of the vibration; the amplitude, which reflects the intensity of the vibration; and the harmonic component, which are used to analyze the complexity and stability of the vibration.

[0119] According to the mean and variance of the spindle vibration spectrum data amplitude in the historical vibration amplitude database, according to the statistical principle, the Principle: Set amplitude thresholds, including upper and lower amplitude thresholds;

[0120] The real-time collected spindle vibration amplitude data is immediately subjected to fast Fourier transform to obtain the real-time main frequency component, real-time amplitude and real-time harmonic component of the real-time spindle vibration spectrum data;

[0121] If the real-time amplitude is greater than the upper amplitude threshold or less than the lower amplitude threshold, it is determined that there is potential abnormal vibration and abnormal vibration correction is performed; otherwise, no processing is performed.

[0122] See also Figure 5 Preferably, the specific steps of performing abnormal vibration correction include:

[0123] When it is determined that there is potential abnormal vibration, according to the current speed of the roller centrifuge , calculate the theoretical rotation fundamental frequency ,Right now: ;

[0124] An eccentricity threshold is configured based on the equipment accuracy, operating conditions, and historical data analysis results of the roller centrifuge. If the absolute value of the difference between the real-time main frequency component of the main shaft vibration spectrum data and the theoretical rotation fundamental frequency is greater than the eccentricity threshold, it is determined that an eccentric vibration source main frequency exists. The phase difference of each axis at the theoretical rotation fundamental frequency is obtained at this time to determine the eccentric phase angle; otherwise, no processing is performed. The main shaft vibration amplitude data is captured by the three-axis MEMS accelerometer, and the main frequency component of the main shaft vibration spectrum data of each axis is obtained to calculate the phase difference. The eccentric phase angle is obtained after coordinate system transformation.

[0125] In the flat angle direction of the eccentric phase angle, the roller centrifuge counterweights are gradually added. Based on the real-time amplitude of the main shaft vibration spectrum data after adding the counterweights, the change in the real-time amplitude before and after adding the counterweights is calculated to obtain the attenuation rate of the real-time amplitude, that is:

[0126]

[0127] in, is the decay rate of the real-time amplitude, is the amplitude of the main shaft vibration spectrum data before adding the roller centrifuge counterweight, is the amplitude of the main shaft vibration spectrum data after adding the roller centrifuge counterweight;

[0128] If the real-time amplitude attenuation rate increases after adding counterweights to the roller centrifuge, re-initiate abnormal vibration monitoring. This indicates that the current counterweight operation may have caused a new abnormality, requiring the restart of the abnormal vibration monitoring process to thoroughly investigate the cause of the vibration anomaly. Otherwise, monitor the real-time amplitude attenuation rate after gradually adding counterweights to the roller centrifuge, configure an attenuation threshold, and stop adding counterweights when the real-time amplitude attenuation rate of the spindle vibration spectrum data falls below the threshold. If the real-time amplitude attenuation rate increases after a counterweight is added, for example, after adding the third counterweight, the attenuation rate changes from 12.5% ​​to 15%, this indicates that the current counterweight operation may have caused a new abnormality, requiring the restart of the abnormal vibration monitoring process to thoroughly investigate the cause of the vibration anomaly, such as checking whether the counterweights are securely installed and whether there are other potential mechanical faults. If the total weight of the added counterweights exceeds the counterweight correction threshold, the current abnormal vibration correction process must be stopped and abnormal vibration monitoring re-initiated.

[0129] Configure a counterweight correction threshold and calculate the total weight of the added counterweights to the roller centrifuge. If the total weight exceeds the counterweight correction threshold, stop abnormal vibration correction and restart abnormal vibration monitoring. Further troubleshoot the problem to prevent the negative impact of unreasonable counterweight on equipment performance.

[0130] The fault warning module is responsible for multi-dimensional fault warning, including abnormal pressure warning, abnormal metal liquid warning, abnormal speed warning and abnormal vibration warning, and displays multi-dimensional fault warning information through the central control platform;

[0131] In this embodiment, the data analysis module receives mold inner wall pressure data processed by the data analysis module. When the variance of the mold inner wall pressure data after outlier correction exceeds a pre-configured variance threshold, the roller centrifuge is determined to have a potential pressure anomaly, and a pressure anomaly warning is immediately triggered. The pressure warning information includes the time the pressure anomaly occurred, the sensor ID involved, and the range of the abnormal pressure data. If, during the roller centrifuge metal flow uniformity test, the pressure difference between a region and the average mold inner wall pressure data of all neighboring regions exceeds the pressure difference threshold, the region is marked as an abnormal region, the metal flow uniformity test is determined to have failed, and a metal anomaly warning is triggered. The pressure warning information specifies the abnormal region number, the pressure difference value, and a comparison analysis with normal conditions. The adaptive control module obtains roller centrifuge speed data. When the roller centrifuge speed is not within the safe speed range, a speed anomaly warning is immediately issued. The warning information should include the current speed, the safe speed range, and the degree of deviation from the safe range. If a speed shutdown command is received, the fault warning module simultaneously issues a speed shutdown warning, reminding the operator to pay attention to the equipment status and avoid equipment damage or metal splashing due to overspeed. During abnormal vibration monitoring in the adaptive control module, if the real-time spindle vibration amplitude data collected through a fast Fourier transform (FFT) shows that the real-time amplitude exceeds the upper threshold or falls below the lower threshold, a potential abnormal vibration is detected, triggering a vibration anomaly warning. The warning information includes the time of the abnormal vibration occurrence, the real-time amplitude, and the amplitude threshold range. During abnormal vibration correction, if the attenuation rate of the real-time amplitude increases after adding counterweights, or if the total weight of the added counterweights exceeds the counterweight correction threshold, abnormal vibration monitoring is restarted and a vibration correction anomaly warning is issued, prompting the operator to re-investigate the cause of the vibration anomaly. Warning information is intuitively displayed on the central control platform's human-machine interface, and detailed information viewing is provided. Operators can click on a warning to view a detailed description of the fault, the time of occurrence, and the sensor data involved. When a warning is triggered, relevant personnel are notified via various means, including text messages, emails, and internal system messages. Detailed records of each warning are stored in a database, creating a comprehensive warning history archive.

[0132] Preferably, the specific steps of multi-dimensional fault warning include:

[0133] When the data analysis module determines that there is a potential pressure anomaly in the roller centrifuge, a pressure anomaly warning is triggered. The pressure anomaly warning information includes the time when the pressure anomaly occurred, the sensor ID involved, and the range of the abnormal pressure data;

[0134] When the data analysis module determines that the molten metal flow uniformity test fails, a molten metal abnormality warning is triggered. The molten metal abnormality warning information includes the number of the abnormal area and the pressure difference value;

[0135] When the adaptive control module detects that the speed data of the roller centrifuge is not within the safe speed range, it triggers a speed abnormality warning. The speed abnormality warning information includes the roller centrifuge speed and the safe speed range;

[0136] When the adaptive control module determines that there is potential abnormal vibration, it triggers a vibration abnormality warning. The vibration abnormality warning information includes the time when the vibration abnormality occurs, the real-time amplitude, and the amplitude threshold range.

[0137] Example 2

[0138] See also Figure 6 This embodiment introduces an adaptive control method for an intelligent roller centrifuge, comprising the following steps:

[0139] Step S1: acquiring multi-source heterogeneous data in real time during the operation of the roller centrifuge, including mold inner wall pressure data, spindle vibration amplitude data, spindle vibration spectrum data, molten metal temperature data, and speed data, collecting data through sensors, and transmitting the collected multi-source heterogeneous data to the central control platform;

[0140] Step S2: performing a multi-dimensional correlation analysis based on the material property database, and detecting the flow uniformity of the molten metal in the roller centrifuge according to the mold inner wall pressure data. When uneven flow of the molten metal is detected, the viscosity of the molten metal is obtained according to the viscosity-temperature curve of the molten metal material to calculate the correction step size of the roller centrifuge speed, and generate a roller centrifuge speed instruction, including a speed step correction instruction and a speed stop instruction.

[0141] Step S3: parsing the speed command of the roller centrifuge and controlling the speed of the roller centrifuge. By establishing a dynamic relationship equation between the pouring speed and the speed, the pouring flow rate of the molten metal is synchronously controlled. During the speed control process of the roller centrifuge, abnormal vibration monitoring is performed based on the vibration spectrum data of the roller centrifuge main shaft. If potential abnormal vibration is detected, the eccentric phase angle is identified and the abnormal vibration is corrected by gradually adding counterweights to the roller centrifuge.

[0142] Step S4: Perform multi-dimensional fault warning, including abnormal pressure warning, abnormal metal liquid warning, abnormal speed warning and abnormal vibration warning, and display the multi-dimensional fault warning information through the central control platform.

[0143] Preferably, the specific steps of multi-dimensional correlation analysis include:

[0144] When the central control platform receives a set of multi-source heterogeneous data, it encodes and marks the multi-source heterogeneous data according to the sensor ID and timestamps the multi-source heterogeneous data according to the collection time;

[0145] The labeled multi-source heterogeneous data are grouped according to data type, and outlier detection is performed on each group of multi-source heterogeneous data. The detected outliers are marked, and the mean of each group of data is used to correct the outliers.

[0146] Obtain the mold inner wall pressure data after outlier correction, extract its statistical features, and configure a variance threshold. If the variance of the mold inner wall pressure data after outlier correction is greater than the variance threshold, it is determined that there is a potential pressure anomaly in the roller centrifuge and the flow uniformity of the metal liquid in the roller centrifuge is tested. Otherwise, no action is taken.

[0147] When it is determined that the roller centrifuge metal liquid flow uniformity test has failed, the metal liquid temperature data in the roller centrifuge is obtained, and the viscosity-temperature curve of the corresponding metal liquid material is retrieved from the material property database according to the characteristic code of the metal liquid material; the current metal liquid viscosity is calculated using an interpolation algorithm based on the metal liquid temperature data in the roller centrifuge;

[0148] Determine the critical speed formula and combine it with the current viscosity of the molten metal to obtain the critical speed of the roller centrifuge ;

[0149] Setting limit thresholds , combined with the critical speed of the roller centrifuge, the safe speed range of the roller centrifuge is limited to [0, ];

[0150] Obtain the current roller centrifuge speed data to determine whether it is within the safe speed range of the roller centrifuge. If it is within the safe speed range of the roller centrifuge, obtain the speed correction step size based on the target viscosity of the molten metal and gradually correct the roller centrifuge speed.

[0151] Generate a step-by-step correction instruction for the speed of the roller centrifuge according to the current speed data of the roller centrifuge and the speed correction step size;

[0152] Otherwise, a safety deceleration time is set, and based on the current roller centrifuge speed data, a speed stop step is calculated to generate a roller centrifuge speed stop instruction to terminate the operation of the roller centrifuge.

[0153] Preferably, the specific steps of performing flow uniformity detection on the molten metal of the roller centrifuge include:

[0154] Divide the inner wall of the roller centrifuge mold into regions, mark each region with a number, and obtain the coordinates of the center point of each numbered region;

[0155] According to the coding marks of the piezoelectric sensors, the area where the piezoelectric sensors are located is determined, the piezoelectric sensors in each numbered area are grouped, and the mold inner wall pressure data collected by each group of piezoelectric sensors is obtained;

[0156] According to the distance between the position coordinates of the piezoelectric sensor in each numbered area and the coordinates of the center point of the area, an inverse proportional function is used to determine the position weighting coefficient;

[0157] For each area, the mold inner wall pressure data collected by the piezoelectric sensor in the area and the corresponding position weighting coefficient are used to calculate the average mold inner wall pressure data of each area through weighted averaging;

[0158] Define the neighborhood range of the region. For each region, calculate the pressure difference between it and the average pressure data of the mold inner wall of each region in the neighborhood. Record the value of each pressure difference and correspond it to the region number in the neighborhood.

[0159] Configure a pressure difference threshold. If the pressure difference between a region and the average mold inner wall pressure data of all neighboring regions is greater than the pressure difference threshold, the region is marked as an abnormal region and is judged to have failed the roller centrifuge metal liquid flow uniformity test. Otherwise, the roller centrifuge metal liquid flow uniformity test is judged to have passed.

[0160] Preferably, the specific steps of abnormal vibration monitoring include:

[0161] When receiving the speed instruction of the roller centrifuge and controlling the speed of the roller centrifuge, abnormal vibration monitoring is started;

[0162] According to the type of molten metal, the vibration amplitude data of the roller centrifuge spindle when the roller centrifuge speed is not regulated is filtered from the central control platform, and the data is cleaned to remove abnormal data points and establish a historical vibration amplitude database;

[0163] Perform fast Fourier transform on the spindle vibration amplitude data in the historical vibration amplitude database to obtain the main frequency component, amplitude and harmonic components of the corresponding spindle vibration spectrum data;

[0164] According to the mean and variance of the amplitude of the main shaft vibration spectrum data in the historical vibration amplitude database, the amplitude threshold is set, including the upper amplitude threshold and the lower amplitude threshold;

[0165] The real-time collected spindle vibration amplitude data is immediately subjected to fast Fourier transform to obtain the real-time main frequency component, real-time amplitude and real-time harmonic component of the real-time spindle vibration spectrum data;

[0166] If the real-time amplitude is greater than the upper amplitude threshold or less than the lower amplitude threshold, it is determined that there is potential abnormal vibration and abnormal vibration correction is performed; otherwise, no processing is performed.

[0167] Working principle and its effect:

[0168] When the adaptive control system of the intelligent roller centrifuge is operating, the data acquisition module uses sensors to collect real-time, multi-source heterogeneous data such as mold inner wall pressure, spindle vibration amplitude and spectrum, molten metal temperature, and speed, and transmits it to the central control platform. After receiving the data, the data analysis module first encodes and timestamps it according to the sensor ID and acquisition time, then groups it to detect and correct outliers. Statistical features such as the mean and variance are extracted from the mold inner wall pressure data. When the variance exceeds a threshold, a potential pressure anomaly is determined and the flow uniformity of the molten metal is tested. The mold inner wall is divided into regions, and position weighting coefficients are determined based on sensor locations to calculate the mean pressure of each region. The flow uniformity is then determined by comparing the pressure difference in adjacent regions with the threshold. If the flow is uneven, the viscosity-temperature curve in the material properties database is retrieved to calculate the current molten metal viscosity. The safe speed range is determined by combining the critical speed formula and the limit threshold, thereby generating a speed command to ensure uniform molten metal flow and improve product quality.

[0169] The adaptive control module analyzes the speed command to control the speed. It establishes a dynamic relationship equation between pouring speed and speed based on factors such as mold volume and molten metal viscosity. By fitting the coupling coefficient, it updates the speed in real time according to the speed command and synchronously controls the molten metal pouring flow rate to ensure that the two match. At the same time, during speed control, the spindle vibration spectrum data is monitored, a historical vibration amplitude database is established, and an amplitude threshold is set. The real-time amplitude is compared with the threshold to determine whether there is potential abnormal vibration. If so, the theoretical rotational fundamental frequency is calculated to determine whether there is an eccentric vibration source main frequency and the eccentric phase angle. A counterweight is added in the flat angle direction and the counterweight is adjusted according to the real-time amplitude attenuation rate to ensure stable operation of the equipment and reduce the risk of failure.

[0170] When the data analysis module determines that there is a potential pressure anomaly, the metal liquid flow uniformity test fails, or the adaptive control module detects that the speed is not within the safe speed range or there is potential abnormal vibration, the fault warning module triggers pressure, metal liquid, speed and vibration abnormality warnings respectively, and displays the warning content containing detailed abnormal information on the central control platform so that operators can take timely measures to reduce equipment damage and production accidents, and improve production safety and reliability.

[0171] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the concept of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. The adaptive control system of the intelligent roller centrifuge is characterized by: include: Real-time collection of multi-source heterogeneous data during the operation of the roller centrifuge and transmission to the central control platform. The multi-source heterogeneous data includes mold inner wall pressure data, spindle vibration amplitude data, spindle vibration spectrum data, molten metal temperature data, and speed data. Performing a multi-dimensional correlation analysis based on a material property database, including: detecting the flow uniformity of the molten metal in the roller centrifuge according to the mold inner wall pressure data; when uneven flow of the molten metal is detected, obtaining the viscosity of the molten metal according to the viscosity-temperature curve of the molten metal material, calculating a correction step length for the speed of the roller centrifuge, and generating a speed instruction for the roller centrifuge; The speed command of the roller centrifuge is parsed to control the speed of the roller centrifuge. By establishing a dynamic relationship equation between the pouring speed and the speed, the pouring flow rate of the molten metal is synchronously controlled. During the speed control process of the roller centrifuge, abnormal vibration monitoring is performed based on the vibration spectrum data of the roller centrifuge main shaft. If potential abnormal vibration is detected, the eccentric phase angle is identified and the abnormal vibration is corrected by gradually adding counterweight blocks to the roller centrifuge. In addition, multi-dimensional fault warnings are carried out in real time during the operation of the roller centrifuge, including abnormal pressure warnings, abnormal metal liquid warnings, abnormal speed warnings and abnormal vibration warnings, and the multi-dimensional fault warning information is displayed through the central control platform.

2. The adaptive control system of the intelligent roller centrifuge according to claim 1, characterized in that: The specific steps of the multi-dimensional association analysis include: When the central control platform receives a set of multi-source heterogeneous data, it encodes and marks the multi-source heterogeneous data according to the sensor ID and timestamps the multi-source heterogeneous data according to the collection time; The labeled multi-source heterogeneous data are grouped according to data type, and outlier detection is performed on each group of multi-source heterogeneous data. The detected outliers are marked, and the mean of each group of data is used to correct the outliers. Obtain the mold inner wall pressure data after outlier correction, extract its statistical features, and configure a variance threshold. If the variance of the mold inner wall pressure data after outlier correction is greater than the variance threshold, it is determined that there is a potential pressure anomaly in the roller centrifuge and the flow uniformity of the metal liquid in the roller centrifuge is tested. Otherwise, no action is taken. When it is determined that the roller centrifuge metal liquid flow uniformity test fails, the metal liquid temperature data in the roller centrifuge is obtained, and the viscosity-temperature curve of the corresponding metal liquid material is retrieved from the material property database according to the characteristic code of the metal liquid material; based on the metal liquid temperature data in the roller centrifuge, the current metal liquid viscosity is calculated using an interpolation algorithm.

3. The adaptive control system of the intelligent roller centrifuge according to claim 2, characterized in that: The specific steps of the multi-dimensional association analysis also include: Determine the critical speed formula and combine it with the current viscosity of the molten metal to obtain the critical speed of the roller centrifuge ; Setting limit thresholds , combined with the critical speed of the roller centrifuge, the safe speed range of the roller centrifuge is limited to [0, ]; Obtain the current roller centrifuge speed data to determine whether it is within the safe speed range of the roller centrifuge. If it is within the safe speed range of the roller centrifuge, obtain the speed correction step size based on the target viscosity of the molten metal and gradually correct the roller centrifuge speed. Generate a step-by-step correction instruction for the speed of the roller centrifuge according to the current speed data of the roller centrifuge and the speed correction step size; Otherwise, a safety deceleration time is set, and based on the current roller centrifuge speed data, a speed stop step is calculated to generate a roller centrifuge speed stop instruction to terminate the operation of the roller centrifuge.

4. The adaptive control system of the intelligent roller centrifuge according to claim 2, characterized in that: The specific steps of performing flow uniformity detection on the molten metal of the roller centrifuge include: Divide the inner wall of the roller centrifuge mold into regions, mark each region with a number, and obtain the coordinates of the center point of each numbered region; According to the coding marks of the piezoelectric sensors, the area where the piezoelectric sensors are located is determined, the piezoelectric sensors in each numbered area are grouped, and the mold inner wall pressure data collected by each group of piezoelectric sensors is obtained; According to the distance between the position coordinates of the piezoelectric sensor in each numbered area and the coordinates of the center point of the area, an inverse proportional function is used to determine the position weighting coefficient; For each area, the mold inner wall pressure mean data of each area is calculated by weighted averaging based on the mold inner wall pressure data collected by the piezoelectric sensor in the area and the corresponding position weighting coefficient.

5. The adaptive control system of the intelligent roller centrifuge according to claim 4, characterized in that: The specific steps of performing flow uniformity detection on the molten metal of the roller centrifuge also include: Define the neighborhood range of the region. For each region, calculate the pressure difference between it and the average pressure data of the mold inner wall of each region in the neighborhood. Record the value of each pressure difference and correspond it to the region number in the neighborhood. Configure a pressure difference threshold. If the pressure difference between a region and the average mold inner wall pressure data of all neighboring regions is greater than the pressure difference threshold, the region is marked as an abnormal region and is judged to have failed the roller centrifuge metal liquid flow uniformity test. Otherwise, the roller centrifuge metal liquid flow uniformity test is judged to have passed.

6. The adaptive control system of the intelligent roller centrifuge according to claim 1, characterized in that: The specific steps of synchronously controlling the metal liquid pouring flow rate include: Based on the mold volume, molten metal viscosity, molten metal density and the coupling coefficient between pouring speed and rotation speed of the roller centrifuge, a dynamic relationship equation between pouring speed and rotation speed is established; Based on the type of molten metal, the historical multi-source heterogeneous data in the central control platform is screened to fit the coupling coefficient of the pouring speed and rotation speed in the dynamic relationship equation between the pouring speed and rotation speed; receiving a speed command for a roller centrifuge, and if it is a speed step-by-step correction command, parsing the speed step-by-step correction command to obtain a speed correction step, a target speed, and an adjustment time interval; According to the speed correction step and the adjustment time interval, the speed of the roller centrifuge is gradually updated in real time, and the metal liquid pouring flow rate is obtained according to the dynamic relationship equation between the pouring speed and the speed after fitting, and the pouring flow rate instruction is generated; According to the pouring flow instruction, the PID controller adjusts the pouring valve opening, changes the flow cross-sectional area of ​​the molten metal, and controls the change of the molten metal pouring flow of the roller centrifuge; If it is a speed stop command, a stop pouring command is generated and the pouring valve is closed.

7. The adaptive control system of the intelligent roller centrifuge according to claim 1, characterized in that: The specific steps of abnormal vibration monitoring include: When receiving the speed instruction of the roller centrifuge and controlling the speed of the roller centrifuge, abnormal vibration monitoring is started; According to the type of molten metal, the vibration amplitude data of the roller centrifuge spindle when the roller centrifuge speed is not regulated is filtered from the central control platform, and the data is cleaned to remove abnormal data points and establish a historical vibration amplitude database; Perform fast Fourier transform on the spindle vibration amplitude data in the historical vibration amplitude database to obtain the main frequency component, amplitude and harmonic components of the corresponding spindle vibration spectrum data; According to the mean and variance of the amplitude of the main shaft vibration spectrum data in the historical vibration amplitude database, the amplitude threshold is set, including the upper amplitude threshold and the lower amplitude threshold; Perform fast Fourier transform on the real-time collected spindle vibration amplitude data to obtain the real-time main frequency component, real-time amplitude and real-time harmonic component of the real-time spindle vibration spectrum data; If the real-time amplitude is greater than the upper amplitude threshold or less than the lower amplitude threshold, it is determined that there is potential abnormal vibration and abnormal vibration correction is performed; otherwise, no processing is performed.

8. The adaptive control system of the intelligent roller centrifuge according to claim 7, characterized in that: The specific steps of performing abnormal vibration correction include: When it is determined that there is potential abnormal vibration, the theoretical rotation fundamental frequency is calculated based on the current speed of the roller centrifuge; Configure an eccentricity threshold. If the absolute value of the difference between the real-time main frequency component of the main shaft vibration spectrum data and the theoretical rotation fundamental frequency is greater than the eccentricity threshold, it is determined that an eccentric vibration source main frequency exists. The phase difference of each axis at the theoretical rotation fundamental frequency is obtained to determine the eccentricity phase angle. Otherwise, no processing is performed. In the flat angle direction of the eccentric phase angle, the roller centrifuge counterweights are gradually added, and based on the real-time amplitude of the main shaft vibration spectrum data after the counterweights are added, the change in the real-time amplitude before and after the addition of the counterweights is calculated to obtain the attenuation rate of the real-time amplitude; If the attenuation rate of the real-time amplitude increases after adding the counterweight of the roller centrifuge, the abnormal vibration monitoring should be carried out again; Otherwise, the attenuation rate of the real-time amplitude after the counterweights of the roller centrifuge are gradually added is monitored, and an attenuation threshold is configured. When the attenuation rate of the real-time amplitude of the main shaft vibration spectrum data is less than the attenuation threshold, the addition of the counterweights is stopped; Configure the counterweight correction threshold and calculate the total weight of the counterweight blocks added to the roller centrifuge. If the total weight is greater than the counterweight correction threshold, stop the abnormal vibration correction and restart the abnormal vibration monitoring.

9. An adaptive control method for an intelligent roller centrifuge, which is implemented based on the adaptive control system of the intelligent roller centrifuge according to any one of claims 1 to 8, characterized in that: The following steps are involved: Step S1: acquiring multi-source heterogeneous data in real time during the operation of the roller centrifuge, including mold inner wall pressure data, spindle vibration amplitude data, spindle vibration spectrum data, molten metal temperature data, and speed data, collecting data through sensors, and transmitting the collected multi-source heterogeneous data to the central control platform; Step S2: performing a multi-dimensional correlation analysis based on the material property database, and detecting the flow uniformity of the molten metal in the roller centrifuge according to the mold inner wall pressure data. When uneven flow of the molten metal is detected, the viscosity of the molten metal is obtained according to the viscosity-temperature curve of the molten metal material, and the correction step length of the roller centrifuge speed is calculated to generate a roller centrifuge speed instruction, including a speed step correction instruction and a speed stop instruction. Step S3: parsing the speed command of the roller centrifuge and controlling the speed of the roller centrifuge. By establishing a dynamic relationship equation between the pouring speed and the speed, the pouring flow rate of the molten metal is synchronously controlled. During the speed control process of the roller centrifuge, abnormal vibration monitoring is performed based on the vibration spectrum data of the roller centrifuge main shaft. If potential abnormal vibration is detected, the eccentric phase angle is identified and the abnormal vibration is corrected by gradually adding counterweights to the roller centrifuge. Step S4: Perform multi-dimensional fault warning, including abnormal pressure warning, abnormal metal liquid warning, abnormal speed warning and abnormal vibration warning, and display the multi-dimensional fault warning information through the central control platform.

Images