Centrifuge rotor identification and control method and system

Abstract

The invention provides a centrifugal machine rotor identification and control method and system, and relates to the field of centrifugal machines. The method comprises the steps that experimental data are input, sensing data are collected for comparison and verification, centrifugal parameters are generated, and the centrifugal parameters comprise the type, the volume, the mass and the concentration of a centrifugal tube and a solution; scanning a rotor ID and calling historical data, constructing a rotating speed-centrifugal parameter curve model, and determining a target rotating speed according to centrifugal parameters; real-time state data of a main shaft in the double-bearing structure of the rotor are monitored, implicit association between the real-time state data is analyzed, a constraint rule is generated, and the real-time state data comprise the physical state of the main shaft, the pawl-torsion spring mechanism state, the state of the double-bearing structure and the wind speed; and the real-time rotating speed of the rotor is obtained, and the real-time rotating speed of the rotor is adjusted in real time based on the constraint rule The technical problem that the centrifugal machine is low in precision, safety and efficiency due to the fact that the rotor rotating speed is preset depending on fixed single data is solved.

Description

Technical Field

[0001] This application relates to the field of centrifuges, and more particularly to a centrifuge rotor identification and control method and system. Background Technology

[0002] Centrifuges, as core equipment in fields such as biomedicine and chemical industry, rely heavily on rotor recognition technology, which directly impacts separation efficiency, equipment safety, and sample integrity. Currently, rotor speeds are primarily set manually by operators based on experience or pre-stored templates. This approach fails to fully capture the physicochemical properties of the centrifuge tubes and solutions, resulting in incomplete parameter acquisition. Consequently, the set speed may not match the actual needs of the solution, leading to insufficient separation of light and heavy components, or denaturation or rupture of sensitive samples (such as cells and proteins) due to overload at high speeds. Furthermore, pre-stored templates and operator-preset speeds not only fail to adapt to sudden changes in centrifugation parameters (such as uneven solution distribution or rotor imbalance), potentially causing spindle or bearing overload, but also struggle to optimize transient forward and reverse rotation curves during high-speed centrifugation. This can lead to rotor vibration or disengagement, increasing the probability of sudden equipment shutdown or component damage. Summary of the Invention

[0003] This application provides a centrifuge rotor identification and control method and system, which solves the technical problem that the existing technology relies on a fixed single data preset rotor speed, which reduces the accuracy, safety and efficiency of the centrifuge.

[0004] To achieve the above objectives, this application adopts the following technical solution: In a first aspect, a centrifuge rotor identification and control method includes: inputting experimental data and collecting sensor data for comparison and verification, generating centrifugation parameters, including the type, volume, mass, and concentration of centrifuge tubes and solutions; scanning the rotor ID and calling historical data to construct a speed-centrifugation parameter curve model, and determining the target speed based on the centrifugation parameters; monitoring the real-time status data of the main shaft in the rotor's dual-bearing structure, analyzing the implicit correlations between the real-time status data, generating constraint rules, the real-time status data including the physical state of the main shaft, the state of the pawl-torsion spring mechanism, the state of the dual-bearing structure, and the wind speed; acquiring the real-time rotor speed, and adjusting the real-time rotor speed in real time based on the constraint rules.

[0005] Based on the above technical solution, the centrifuge rotor identification and control method provided in this application integrates a sensor network and a user input interface to collect the physicochemical parameters of the centrifuge tubes and solutions in real time. This allows for comprehensive quantification of centrifugation parameters, enabling personalized settings for the centrifugation process based on solution characteristics. This improves separation accuracy and consistency, avoiding incomplete separation or sample damage due to missing parameters. Simultaneously, combining the speed-centrifugation parameter curve with the real-time status data of the main shaft in the rotor's dual-bearing structure forms a closed-loop architecture. The speed-centrifugation parameter curve can accurately match the actual centrifugation parameter conditions of the rotor, preventing energy waste caused by excessively high or low speeds and preventing overload of the main shaft or bearings due to sudden changes in centrifugation parameters. Furthermore, based on the real-time status data of the main shaft in the rotor's dual-bearing structure, key components (such as the pawl-torsion spring mechanism) can be comprehensively monitored, uncovering implicit correlations between various factors. By optimizing the speed curve through rules, the stability of the centrifuge during forward and reverse rotation switching is ensured. Continuously collected status data enables a shift from passive maintenance to proactive maintenance, reducing downtime and preventing equipment shutdowns or safety accidents caused by sudden malfunctions.

[0006] In conjunction with the first aspect mentioned above, one possible implementation involves scanning the rotor ID and retrieving historical data to construct a speed-centrifugation parameter curve. Specifically, this includes: scanning the unique identifier on the rotor to obtain the rotor model, and retrieving the corresponding rotor's historical operating data from the historical database. The historical operating data includes historical centrifugation parameters and historical actual speed records from multiple centrifugation tasks; using a multinomial regression curve fitting algorithm, with historical centrifugation parameters as independent variables and historical actual speed as dependent variables, constructing a multidimensional speed-centrifugation parameter curve model for each rotor ID, creating a personalized curve; and interpolating the target speed from the speed-centrifugation parameter curve model based on the centrifugation parameters.

[0007] In conjunction with the first aspect mentioned above, in one possible implementation, the process of monitoring the real-time status data of the main shaft in a rotor dual-bearing structure specifically includes: deploying a multi-source sensor network, which includes vibration sensors, temperature sensors, and deformation sensors for monitoring the dual-bearing structure; position sensors and pressure sensors for monitoring the pawl-torsion spring mechanism; and an anemometer for measuring the heat dissipation wind speed; collecting vibration characteristic data of the main shaft and the pawl-torsion spring mechanism in the dual-bearing structure, and calculating the overall vibration characteristics through a data fusion algorithm, wherein the overall vibration characteristics are the comprehensive vibration spectrum after coupling the dual-bearing structure and the pawl-torsion spring mechanism; collecting mechanical performance parameters, load distribution data, surface condition analysis data, and heat dissipation resistance of the dual-bearing structure to construct a multi-dimensional real-time status dataset, wherein the surface condition analysis data includes the wear degree, corrosion degree, and heat dissipation resistance of the dual-bearing structure; and preprocessing the multi-dimensional real-time status dataset through a dedicated signal processing chip integrated into an edge computing device to obtain standardized real-time status data.

[0008] In conjunction with the first aspect mentioned above, one possible implementation involves collecting mechanical performance parameters, load distribution data, surface condition analysis data, and heat dissipation resistance data of a dual-bearing structure to construct a multi-dimensional real-time state dataset. Specifically, this includes: acquiring surface image data of the spindle in the dual-bearing structure and using a dedicated image processing chip integrated into an edge computing device to extract wear and corrosion on the spindle surface; cross-validating the wear and corrosion with the mechanical performance parameters and load distribution data and calculating the error rate; triggering re-collection when the error rate exceeds a preset error threshold; and constructing a five-dimensional real-time state dataset composed of mechanical performance parameters, load distribution data, wear, corrosion, and heat dissipation resistance data, based on the real-time collected heat dissipation resistance data.

[0009] In conjunction with the first aspect mentioned above, one possible implementation involves analyzing the implicit relationships between real-time state data and generating constraint rules. Specifically, this includes: preprocessing the real-time state dataset into industrial time-series data, dividing it into continuous and state-related data; discretizing the continuous data into different levels; and binary encoding the state-related data. The Apriori algorithm is used to analyze and calculate the weighted support of each frequent itemset to construct hierarchical itemsets. These hierarchical itemsets include first-level frequent itemsets containing basic factors, second-level frequent itemsets containing coupling factors, and third-level frequent itemsets containing control target association items. Dynamic pruning of the hierarchical itemsets is performed in conjunction with the industrial scenario to generate association rules. The association rules are then filtered using a dual threshold of confidence and causality coefficients and mapped to a predefined fine-tuning parameter library to generate constraint rules.

[0010] In conjunction with the first aspect mentioned above, in one possible implementation, the process of acquiring the real-time rotor speed and adjusting it in real-time based on constraint rules specifically includes: real-time acquisition of rotor speed data to obtain a real-time speed sequence; construction of a real-time adjustment and control platform including a data input layer, a rule matching layer, and a control output layer, for receiving the real-time speed sequence and constraint rules; the rule matching layer parsing the constraint rules, extracting the speed limit conditions and fine-tuning parameters from the rules, and matching them with the real-time speed sequence to obtain a matching result. The speed limit conditions include an upper speed limit, a lower speed limit, and a dynamic tolerance range. The fine-tuning parameters include speed offset and current adjustment ratio; the control output layer, based on the matching result, calling a predefined fine-tuning parameter library, calculating the speed adjustment value, and generating a control signal; and the motor drive unit executing the control signal to adjust the rotor speed in real time.

[0011] Secondly, a centrifuge rotor identification and control system is provided, comprising: a communication unit, a processing unit, a dedicated integrated module, a hierarchical control platform, and a multi-source sensor network; the communication unit is used to input experimental data and collect sensor data for comparison and verification, generating centrifugation parameters, including the type, volume, mass, and concentration of centrifuge tubes and solutions; the processing unit is used to scan the rotor ID and retrieve historical data, construct a speed-centrifugation parameter curve model, and determine the target speed based on the centrifugation parameters; monitor the real-time status data of the main shaft in the rotor's dual-bearing structure, analyze the implicit correlation between real-time status data, and generate constraint rules; acquire the real-time rotor speed, and adjust the real-time rotor speed in real time based on the constraint rules; the dedicated integrated module includes an edge computing device integrating a dedicated signal processing chip and a dedicated image processing chip. The system employs a chip for preprocessing and feature extraction of real-time status data. Edge computing devices are dynamically deployed and remotely upgraded to adapt to different centrifuge operating conditions. The hierarchical control platform includes a data input layer, a rule matching layer, and a control output layer. The data input layer receives real-time speed sequences and constraint rules. The rule matching layer parses the rules and extracts speed limit conditions and fine-tuning parameters. The control output layer generates control signals based on the matching results and performs real-time adjustments through the motor drive unit. A multi-source sensor network is deployed on the rotor's dual-bearing main shaft, including vibration sensors, temperature sensors, deformation sensors, position sensors, pressure sensors, and anemometers. This network collects data on the main shaft's physical state, the pawl-torsion spring mechanism's state, the dual-bearing structure's state, and wind speed, and calculates the overall vibration characteristics through a data fusion algorithm.

[0012] In conjunction with the second aspect mentioned above, in one possible implementation, the processing unit further includes an association rule mining module integrated on an edge computing device within a dedicated integrated module. This module analyzes implicit associations between real-time state data to generate constraint rules. The association rule mining module includes: a data preprocessing unit, used to perform industrial time-series data preprocessing on the real-time state dataset, discretizing continuous data into multiple levels and binary encoding state data; an itemset construction unit, used to analyze and calculate the weighted support of each frequent itemset using the Apriori algorithm to construct hierarchical itemsets, including first-level frequent itemsets containing basic factors, second-level frequent itemsets containing coupling factors, and third-level frequent itemsets containing control target association items; a rule generation unit, used to dynamically prune the hierarchical itemsets in conjunction with the industrial scenario to generate association rules, and uses confidence and causality coefficient dual thresholds for filtering; and a control mapping unit, used to map association rules to a predefined fine-tuning parameter library to generate constraint rules.

[0013] In conjunction with the second aspect mentioned above, in one possible implementation, the process of dynamically deploying and remotely upgrading the association rule mining module specifically includes: real-time acquisition of the centrifuge's current operating condition data, which includes the rotor's real-time speed, main shaft physical state data, pawl-torsion spring mechanism state data, dual bearing structure state data, and wind speed data; based on the current operating condition data, calculating the matching degree between the current operating condition and the preset operating condition model using the Euclidean distance formula; when the matching degree is lower than a preset threshold, automatically generating a module update request; based on the module update request, reconfiguring the operating parameters of the data preprocessing unit, itemset construction unit, rule generation unit, and control mapping unit of the association rule mining module through hardware acceleration using an integrated dedicated deployment chip; dynamically loading the update package, which includes optimized algorithm parameters, rule base, and weight configuration.

[0014] In conjunction with the second aspect mentioned above, in one possible implementation, the association rule mining module further includes an adaptive learning unit, used to dynamically update the weights and support thresholds of association rule mining based on historical operating data and real-time feedback through an online learning algorithm. Specifically, this includes: a data stream processing submodule, used to receive real-time data on the physical state of the main shaft, the pawl-torsion spring mechanism state, the dual-bearing structure state, and wind speed data collected by a multi-source sensor network, and perform time-series alignment and noise filtering; a weight update submodule, which uses a sliding window mechanism to analyze the frequency of frequent itemsets in historical operating data, calculates the weight adjustment amount using a gradient descent algorithm, and dynamically updates the weighted support weights in the itemset construction unit; a threshold optimization submodule, which adaptively adjusts the confidence threshold and causality coefficient threshold using a recursive least squares method based on the historical error rate of the confidence and causality coefficients; and a deployment interface submodule, used to load the optimized algorithm parameters via remote upgrade.

[0015] This application provides a centrifuge rotor identification and control method and system. By integrating a sensor network and a user input interface, it collects the physicochemical parameters of centrifuge tubes and solutions in real time, enabling comprehensive quantification of centrifugation parameters. This allows for personalized settings of the centrifugation process based on solution characteristics, improving separation accuracy and consistency and avoiding incomplete separation or sample damage due to missing parameters. Simultaneously, by combining the speed-centrifugation parameter curve with the real-time status data of the main shaft in the rotor's dual-bearing structure, a closed-loop architecture is formed. The speed-centrifugation parameter curve can accurately match the actual centrifugation parameter conditions of the rotor, avoiding energy waste caused by excessively high or low speeds and preventing overload of the main shaft or bearings due to sudden changes in centrifugation parameters. Furthermore, based on the real-time status data of the main shaft in the rotor's dual-bearing structure, key components (such as the pawl-torsion spring mechanism) can be comprehensively monitored, uncovering implicit correlations between various factors. By optimizing the speed curve through rules, the stability of the centrifuge during forward and reverse rotation switching is ensured. Through continuous collection of status data, a shift from passive maintenance to proactive maintenance is achieved, reducing downtime and preventing equipment shutdowns or safety accidents caused by sudden malfunctions.

[0016] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0017] Figure 1 A system architecture diagram of a centrifuge rotor identification and control method provided in this application embodiment; Figure 2 A flowchart illustrating a centrifuge rotor identification and control method provided in an embodiment of this application; Figure 3 A flowchart illustrating a centrifuge rotor identification and control method provided in an embodiment of this application; Figure 4 A flowchart illustrating a centrifuge rotor identification and control method provided in an embodiment of this application; Figure 5A flowchart illustrating a centrifuge rotor identification and control method provided in an embodiment of this application; Figure 6 A flowchart illustrating a centrifuge rotor identification and control method provided in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of a centrifuge rotor identification and control system provided in an embodiment of this application; Figure 8 This is a schematic diagram of a centrifuge rotor identification and control system provided in an embodiment of this application. Detailed Implementation

[0018] In the description of this application, unless otherwise stated, "" means "or," for example, A / B can mean A or B. "And / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, "at least one" means one or more, and "multiple" means two or more. The words "first," "second," etc., do not limit the quantity or order of execution, and "first," "second," etc., do not necessarily imply differences.

[0019] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0020] To address the issue that current centrifuge rotor speeds primarily rely on manual settings by operators based on experience or pre-stored templates, which cannot fully capture the physicochemical properties of centrifuge tubes and solutions, resulting in incomplete centrifugation parameter acquisition and mismatches between the set speed and the actual needs of the solution, leading to insufficient separation of light and heavy components, or denaturation and breakage of sensitive samples such as cells and proteins due to overload at high speeds. Furthermore, pre-stored templates and experience-based preset speeds lack adaptive adjustment capabilities. When centrifugation parameters change suddenly (such as uneven solution distribution or rotor imbalance), they cannot respond promptly, easily leading to overload of the spindle or bearings. Meanwhile, during high-speed centrifugation, it is difficult to optimize the transient curves of forward and reverse rotation, which can easily lead to rotor vibration or disengagement, increasing the probability of sudden equipment shutdown and component damage. Furthermore, it is impossible to dynamically adjust based on the real-time status of key components, and it is impossible to proactively avoid safety hazards caused by component wear and corrosion. This affects the centrifuge's separation accuracy and operating efficiency, reduces the safety and stability of equipment operation, and may also lead to prolonged downtime due to sudden failures, affecting experimental or production progress. This application provides a centrifuge rotor identification and control method. This method integrates a sensor network and a user input interface to collect the physicochemical parameters of centrifuge tubes and solutions in real time, enabling comprehensive quantification of centrifugation parameters. This allows for personalized settings of the centrifugation process based on solution characteristics, thereby improving separation accuracy and consistency and avoiding incomplete separation or sample damage due to missing parameters. Simultaneously, combining the speed-centrifugation parameter curve with the real-time status data of the main shaft in the rotor's dual-bearing structure forms a closed-loop architecture. The speed-centrifugation parameter curve can accurately match the actual centrifugation parameter conditions of the rotor, not only avoiding energy waste caused by excessively high or low speeds but also preventing overload of the main shaft or bearings due to sudden changes in centrifugation parameters. Meanwhile, based on the real-time status data of the main shaft in the rotor dual-bearing structure, key components (such as the pawl-torsion spring mechanism) can be fully monitored, the implicit correlation between various factors can be explored, and the speed curve can be optimized by rules to ensure the stability of the centrifuge when switching between forward and reverse rotation. Furthermore, by continuously collecting status data, the transformation from passive maintenance to proactive maintenance can be achieved, reducing downtime and avoiding equipment shutdowns or safety accidents caused by sudden failures.

[0021] like Figure 1 As shown in the embodiment of this application, a centrifuge rotor identification and control method includes: Step 101: Input experimental data and collect sensor data for comparison and verification, and generate centrifugation parameters, including the type, volume, mass and concentration of centrifuge tubes and solutions.

[0022] Before use, the material identification of the centrifuge tube and the type, volume, mass, and concentration of the solution are directly input through the client interface, providing data support for subsequent speed settings. The solution type includes homogeneous or heterogeneous, viscosity, solute particle size, and density, which determine the basis for speed adaptation. Heterogeneous solutions (containing suspended particles) need to be matched to particle sedimentation requirements; higher viscosity or smaller particle density difference with the solvent requires a higher speed to achieve effective separation.

[0023] After manually inputting the centrifugation parameters, centrifuge tubes containing the solution can be placed directly into the centrifuge. The centrifuge then activates its integrated sensor, enabling parallel data acquisition through its image recognition, volume measurement, weighing, and concentration analysis channels. The image recognition channel automatically identifies the material of the centrifuge tube and the type of solution; the volume measurement channel uses a volume measuring device (such as an optical sensor or graduations) to read the volume of the solution within the centrifuge tube; the weighing module channel uses a weighing module to measure the total mass of the centrifuge tube and solution in real time; and the concentration analysis channel uses a concentration analysis instrument (such as a UV-Vis spectrophotometer or densitometer) within the integrated sensor to calculate the solution concentration.

[0024] The sensor data and the input data from the client port are cross-validated item by item. If the data match, the centrifugation parameters are generated directly; if they do not match, a remeasurement or alarm is triggered. During cross-validation, for categorical parameters, string matching or image feature similarity calculation (such as cosine similarity) is used to verify consistency; for numerical parameters such as volume, mass, and concentration, the relative error is calculated (e.g., (input value - sensor value) / sensor value × 100%). If the error is below a preset threshold (e.g., 5%), the data is considered consistent; if the error exceeds the limit, a remeasurement process is automatically triggered: sensor data is used first, and the user interface prompts for verification. After verification, the data is marked as "verified" and stored in the historical database.

[0025] For example, when separating protein solutions in a biological laboratory, if the user manually inputs the centrifuge tube type as glass, the solution type as homogeneous, the volume as 10ml, the mass as 15g, and the concentration as 5%, the centrifuge tube containing the solution can be directly placed inside the centrifuge, and data can be collected synchronously through integrated sensors. If the image recognition shows that the centrifuge tube is made of glass and the solution is transparent (homogeneous characteristics), the volume measurement result is 10.2ml, the weighing mass is 15.1g, and the concentration analysis shows 4.9%, the volume error (|10-10.2| / 10.2×100%≈2%), mass error (|15-15.1| / 15.1×100%≈0.7%), and concentration error (|5-4.9| / 4.9×100%≈2%) can be directly calculated. Since they are all below the 5% threshold, the verification is considered successful. The input values ​​are then fine-tuned using sensor data to ensure that the speed setting accurately matches the actual solution characteristics.

[0026] Step 102: Scan the rotor ID and call up historical data to build a speed-centrifugation parameter curve model, and determine the target speed based on the centrifugation parameters.

[0027] The rotor ID is a unique identifier on the rotor used to identify its type and specifications. Historical data refers to the operating parameters and performance data related to a specific rotor accumulated during the centrifuge's past operations. The speed-centrifugation parameter curve is a mathematical model or data chart describing the optimal speed relationship of the centrifuge rotor under different centrifugation parameter conditions. The target speed refers to the most suitable centrifuge operating speed value calculated based on the current centrifugation parameters and rotor characteristics.

[0028] Before use, the rotor ID can be scanned to identify the currently installed rotor model and specifications. Historical operating data for that rotor model can then be retrieved from the stored historical database. Based on this retrieved historical data, a curve relating rotational speed to centrifugation parameters can be directly constructed. During actual use, once the integrated sensor network and user input interface collect real-time information on the centrifugation parameters of the current centrifugation task, including the type, volume, mass, and concentration of the centrifuge tubes and solutions, the corresponding data can be directly selected from the constructed rotational speed-centrifugation parameter curve based on the current centrifugation parameters to calculate and determine the most suitable target rotational speed value, thus achieving personalized and precise centrifugation operation.

[0029] It should be noted that each centrifugation data will be recorded in the historical database, thereby continuously collecting status data. This allows for better updating and maintenance of the speed-centrifugation parameter curve, improving the accuracy of each calculation and analysis. It also enables a shift from passive to proactive maintenance, reducing downtime and preventing equipment shutdowns or safety accidents caused by sudden malfunctions.

[0030] Taking DNA extraction in a biological laboratory as an example, when researchers need to centrifuge a batch of blood samples, they first place the centrifuge tubes containing the blood samples into the centrifuge. At this time, the rotor ID is automatically scanned and identified as a suitable angle rotor for blood separation. Then, the historical data of the angle rotor is retrieved, and it is found that its optimal speed range for processing similar blood samples is 3000-5000 rpm. Based on the sensor detection that the current centrifuge tubes are 1.5ml in size, a total of 12 tubes, each containing about 1ml of blood, and other centrifugation parameters, combined with the historical speed-centrifugation parameter curve, the target speed for the current task is calculated to be 4200 rpm, achieving efficient and safe separation of blood components.

[0031] Step 103: Monitor the real-time status data of the main shaft in the rotor dual-bearing structure, analyze the implicit correlation between the real-time status data, and generate constraint rules. The real-time status data includes the physical status of the main shaft, the status of the pawl-torsion spring mechanism, the status of the dual-bearing structure, and the wind speed.

[0032] In the rotor dual-bearing structure, the main shaft is the core transmission component driving the rotor rotation in the centrifuge. Real-time status data refers to physical quantity data reflecting the current operating condition of the equipment, acquired in real time during centrifuge operation. Implicit correlation refers to deep interrelationships between different state parameters that are difficult to discover directly through simple observation. Constraint rules are conditions and criteria generated based on data analysis results, used to limit and guide the safe and efficient operation of the equipment. Main shaft physical state refers to the physical characteristics of the main shaft during operation, such as vibration amplitude, temperature change, and deformation degree. The pawl-torsion spring mechanism is a mechanical locking and releasing device used to control the forward and reverse rotation switching of the centrifuge. The dual-bearing structure refers to a mechanical arrangement that uses two bearings to support the main shaft to enhance stability. Air velocity refers to the speed of airflow within the centrifuge chamber, which affects heat dissipation and air resistance.

[0033] The physical state of the main shaft is continuously monitored by vibration sensors, temperature sensors, and deformation detection units deployed on the main shaft within the rotor's dual-bearing structure. Simultaneously, position and pressure sensors collect data on the engagement state of the pawl-torsion spring mechanism and the torsion spring pressure. A sensor array mounted on the bearing housing acquires information on the temperature, vibration, and lubrication status of the dual-bearing structure. Finally, an anemometer measures the wind speed within the cavity in real time. This allows for the direct analysis of multi-dimensional real-time status data, including the main shaft's physical state, the pawl-torsion spring mechanism's state, the dual-bearing structure's state, and wind speed. Association rule mining algorithms can then be used to analyze the implicit correlations between these parameters (e.g., identifying the correspondence between specific vibration modes under high wind speed conditions and bearing temperature increases). Based on these implicit correlations, specific operational constraint rules are generated (e.g., when main shaft vibration exceeds limits and is accompanied by a specific wind speed pattern). These rules can then be used to limit the maximum speed or trigger a speed reduction command, thereby achieving real-time safety constraints on equipment operation.

[0034] For example, when the vibration amplitude of the main shaft increases to the warning value, and the temperature of one of the bearings in the dual-bearing structure shows an abnormal upward trend, while the cavity wind speed data indicates that the heat dissipation conditions are normal, the vibration mode and the temperature rise can be strongly correlated through association rule mining. This indicates that there is a potential risk of bearing damage. A constraint rule is then generated to immediately limit the current speed below the safety threshold and prompt that the bearing needs to be checked, thereby avoiding serious equipment failure that may be caused by bearing failure.

[0035] Step 104: Obtain the real-time rotor speed and adjust the real-time rotor speed in real time based on constraint rules.

[0036] In some implementations, a speed sensor installed on the centrifuge drive acquires the rotor's current rotational speed in real time. This speed is then compared with pre-defined constraint rules based on data analysis of the spindle's physical state, the pawl-torsion spring mechanism's state, the dual-bearing structure's state, and wind speed. If the real-time speed meets the rule requirements, the current operating state is maintained. If the real-time speed reaches a limit set in the rules (such as a speed limit, acceleration limit, or a speed range that does not match the centrifugation parameters), the processing unit immediately generates a control signal to drive the actuator (such as a motor driver) to dynamically adjust the rotor speed (such as implementing deceleration, acceleration, or smooth transition operations), thereby achieving rule-based real-time closed-loop regulation.

[0037] For example, if the real-time rotor speed is 5000 rpm, and the constraint rule stipulates that the speed should not exceed 4800 rpm under the current operating conditions of high dual bearing temperature and low wind speed, the speed over-limit will be immediately identified, and the motor output will be adjusted in real time to reduce the speed to below 4800 rpm, thereby avoiding bearing overheating and damage.

[0038] Based on the above technical solution, by integrating a sensor network and a user input interface, the physicochemical parameters of centrifuge tubes and solutions can be collected in real time, enabling comprehensive quantification of centrifugation parameters. This allows for personalized settings of the centrifugation process based on solution characteristics, thereby improving separation accuracy and consistency and avoiding incomplete separation or sample damage due to missing parameters. Simultaneously, combining the speed-centrifugation parameter curve with the real-time status data of the main shaft in the rotor's dual-bearing structure forms a closed-loop architecture. The speed-centrifugation parameter curve can accurately match the actual centrifugation parameter conditions of the rotor, avoiding energy waste caused by excessively high or low speeds and preventing overload of the main shaft or bearings due to sudden changes in centrifugation parameters. Furthermore, based on the real-time status data of the main shaft in the rotor's dual-bearing structure, key components (such as the pawl-torsion spring mechanism) can be comprehensively monitored, uncovering implicit correlations between various factors. By optimizing the speed curve through rules, the stability of the centrifuge during forward and reverse rotation switching is ensured. Continuously collected status data enables a shift from passive maintenance to proactive maintenance, reducing downtime and preventing equipment shutdowns or safety accidents caused by sudden malfunctions.

[0039] In one possible implementation of the embodiments of this application, combined with Figure 1-2 As shown, scanning the rotor ID and retrieving historical data to construct the speed-centrifugal parameter curve can be achieved through steps 201 to 205, which are explained in detail below: Step 201: Scan the unique identifier on the rotor to obtain the rotor model, and retrieve the historical operating data of the corresponding rotor from the historical database. The historical operating data includes historical centrifugation parameters and historical actual speed records for multiple centrifugation tasks.

[0040] Before starting the centrifuge, the operator installs the rotor onto the centrifuge and automatically activates the RFID scanner, enabling the scanner to emit radio frequency signals to scan the RFID tag on the rotor, read the unique identifier, and decode it to obtain the rotor model (e.g., identified as "Angle Rotor - Model B"). Based on the rotor model obtained from the scan, the operator can access a historical database via network or local interface to query and retrieve historical operating data for multiple centrifugation tasks for the corresponding rotor, including historical centrifugation parameters and historical actual rotation speed records. The historical centrifugation parameters record detailed information such as centrifuge tube material type, solution type, volume value, mass value, and concentration value, while the historical actual rotation speed records correspond to the actual rotation speed value of the centrifuge during each task.

[0041] In some implementations, after the operator places the centrifuge tubes into the centrifuge, the system automatically activates the RFID scanner to scan the rotor ID and obtain the rotor model, such as "angle rotor - model A". Then, the processing unit accesses the historical database and retrieves the past 50 running records for that rotor model, including the centrifugation parameters for each run, such as the centrifuge tube material being glass, the solution type being blood, the volume range being 5-20 ml, the mass range being 10-30 g, the concentration range being 1%-10%, and the corresponding actual rotation speed being 3000-5000 rpm.

[0042] Step 202: Using a multinomial regression curve fitting algorithm, with centrifugal parameters as independent variables and actual rotational speed as dependent variable, construct a multi-dimensional rotational speed-centrifugal parameter curve model to create a personalized curve for each rotor ID.

[0043] By retrieving historical operating data for a specific rotor ID from the historical database, polynomial regression can be selected as the curve fitting algorithm. This allows for the use of centrifugation parameters as combinations of independent variables (e.g., encoding type as numerical values, volume, mass, and concentration as multi-dimensional inputs), with the actual rotational speed as the dependent variable. The polynomial order (e.g., quadratic or cubic) can be set to avoid overfitting. The polynomial coefficients can then be directly calculated using the least squares method, generating a multi-dimensional rotational speed-centrifugation parameter curve model. This model is stored using data encryption and backup mechanisms, with the storage format including polynomial coefficients, centrifugation parameter ranges (e.g., minimum and maximum values ​​for type, volume, mass, and concentration), and rotor ID association information to ensure rapid retrieval. This allows for the creation of personalized curves for each rotor ID based on different combinations of centrifugation parameters (e.g., high volume and low concentration or low mass and high type).

[0044] Step 203: Based on the centrifugation parameters, interpolate and calculate the target rotational speed from the rotational speed-centrifugation parameter curve model.

[0045] In some implementations, when a user initiates a centrifugation task, new centrifugation parameters are received directly through an input interface (such as a sensor or user interface). The format and range validity of these new parameters are checked. After ensuring that the volume value is positive and the concentration is within a reasonable range, a personalized curve model is retrieved from the stored model based on the rotor ID corresponding to the new centrifugation parameters. The new centrifugation parameters are then mapped to a combination of independent variables in the model (e.g., the type is encoded as a numerical value (e.g., glass material is encoded as 1), and volume, mass, and concentration are used as multi-dimensional input vectors). An interpolation algorithm (such as linear interpolation or numerical interpolation based on multinomial regression) is then used to calculate the predicted rotational speed value used to control the centrifuge rotor. This predicted rotational speed value is then fed back to the user interface or log for operator confirmation and adjustment.

[0046] During interpolation, weight values ​​can be dynamically obtained from a preset rule base based on the rotor ID (e.g., for high-viscosity solution rotors, the concentration weight is set to a higher value, such as 0.4, while the volume weight is 0.3). Material values, such as glass or plastic, are converted to numerical values ​​through type encoding, allowing the application of weighted formulas to calculate centrifugation parameter values. This enables the use of a linear interpolation algorithm to find the nearest data point based on the centrifugation parameter values ​​from the stored speed-centrifugation parameter curve model, calculate the target speed, and feed it back to the centrifuge control system for real-time adjustment.

[0047] For example, the current centrifugation parameters are: type glass (coded as 1), volume 10ml, mass 15g, concentration 5%, and rotor ID "angle rotor-A". The system dynamically adjusts the weights based on this ID to type 0.2, volume 0.3, mass 0.3, and concentration 0.2, and calculates the centrifugation parameter value = 0.2×1 + 0.3×10 + 0.3×15 + 0.2×5 = 8.7. Then, it interpolates from the speed-centrifugation parameter curve model. When the centrifugation parameter value of 8.7 corresponds to a speed of 4000rpm, the target speed of 4000rpm is output.

[0048] Based on the above technical solution, a rotor identification mechanism is established by scanning the rotor's unique identifier. This precisely binds historical operating data with a specific rotor ID, enabling full lifecycle data tracking of the rotor and providing a data foundation for personalized models. This solves the problem of inaccurate speed setting caused by rotor generalization in existing technologies. Simultaneously, a multinomial regression algorithm is used to comprehensively fit multi-dimensional centrifugation parameters (type, volume, mass, concentration), effectively quantifying the complex relationship between speed and centrifugation parameters. This allows for better adaptation to different load conditions at the preset speed. Finally, the target speed is intelligently interpolated from the personalized curve model based on real-time centrifugation parameters, ensuring precise matching between the speed setting and real-time operating conditions. This achieves precise and adaptive speed setting, solving the adaptability problem of fixed speed settings under load changes.

[0049] In one possible implementation of the embodiments of this application, combined with Figure 1-3 As shown, the process of monitoring the real-time status data of the spindle in the rotor dual-bearing structure can be achieved through the following steps 301 to 304, which are explained in detail below: Step 301: Deploy a multi-source sensor network, which includes vibration sensors, temperature sensors, and deformation sensors for monitoring the dual-bearing structure; position sensors and pressure sensors for monitoring the pawl-torsion spring mechanism; and an anemometer for measuring the cooling wind speed.

[0050] The dual-bearing structure refers to a mechanical component in a centrifuge that uses two bearings to jointly support the main shaft to enhance stability. Vibration sensors are devices used to detect the frequency and amplitude of mechanical vibrations, such as piezoelectric accelerometers. Temperature sensors are devices used to measure the temperature of an object's surface or the environment, such as thermocouples or PT100 resistance thermometers. Deformation sensors are devices used to monitor minute deformations of mechanical parts, such as strain gauges or fiber optic grating sensors. The pawl-torsion spring mechanism refers to a mechanical locking and releasing device in a centrifuge used to control the forward and reverse rotation of the rotor, consisting of a pawl and a torsion spring. Position sensors are devices used to detect the position or movement of mechanical parts, such as linear encoders or Hall effect sensors. Pressure sensors are devices used to measure changes in the pressure of fluids or solids, such as piezoresistive sensors. Anemometers are devices used to measure the speed of airflow, such as hot-wire anemometers or impeller anemometers.

[0051] In some implementations, vibration sensors, temperature sensors, and deformation sensors are deployed near the centrifuge's dual-bearing structure. Vibration sensors can be magnetically or threadedly mounted to the spindle end to directly acquire vibration spectrum data, while temperature sensors are attached to the outer bearing surface to monitor temperature changes. Simultaneously, deformation sensors are bonded to key parts of the spindle via strain gauges to measure radial deformation. Next, position and pressure sensors are deployed around the pawl-torsion spring mechanism. Position sensors are mounted on the pawl's rotating shaft to detect engagement angles and displacement, while pressure sensors are integrated into the torsion spring attachment to record pressure fluctuations. Finally, an anemometer is installed at the centrifuge chamber inlet to measure the cooling airflow velocity and obtain cooling resistance in real time. All sensors can be directly connected via shielded cables or wireless modules, and their collected data can be aggregated to form a multi-source sensor network, enabling synchronous data acquisition and real-time transmission.

[0052] Step 302: Collect the vibration characteristic data of the spindle in the dual-bearing structure and the vibration characteristic data of the pawl-torsion spring mechanism, and calculate the overall vibration characteristics through the data fusion algorithm. The overall vibration characteristics are the comprehensive vibration spectrum after the coupling of the dual-bearing structure and the pawl-torsion spring mechanism.

[0053] Vibration characteristic data refers to mechanical vibration parameters collected by sensors, such as frequency, amplitude, and spectrum. Data fusion algorithms are computational methods used to integrate multi-source data to generate comprehensive information, such as weighted averaging or spectral superposition. The comprehensive vibration spectrum refers to the vibration energy distribution map obtained through frequency analysis.

[0054] In some implementations, during centrifuge operation, vibration frequency and amplitude data are collected by vibration sensors deployed on the dual-bearing main shaft, while vibration parameters such as meshing vibration frequency and torsion spring fluctuation amplitude are collected by vibration sensors near the pawl-torsion spring mechanism. The collected vibration characteristic data of the dual-bearing main shaft and the mechanical vibration characteristic data of the pawl-torsion spring are then directly input into a data fusion algorithm. At this point, weights are dynamically assigned based on the mechanical contribution of each component in the centrifuge, allowing the data fusion algorithm to use a spectral superposition method to weightedly fuse the vibration spectrum of the dual-bearing main shaft and the vibration spectrum of the pawl-torsion spring mechanism. The algorithm then identifies the main resonant frequency bands and energy distribution characteristics in the coupled integrated vibration spectrum, outputting the overall vibration characteristics.

[0055] For example, if a dual-bearing spindle vibration sensor collects data with a vibration frequency of 120 Hz and an amplitude of 6 μm when separating a protein solution in a centrifuge, while a pawl-torsion spring mechanism vibration sensor collects data with a frequency of 180 Hz and an amplitude of 4 μm, the data can be directly weighted and superimposed using a data fusion algorithm (if the weights are 0.6 and 0.4 respectively). The overall vibration characteristics can then be directly calculated, showing that the comprehensive vibration frequency has a main peak at 140 Hz. This indicates that the coupled vibration of the system is within a safe range.

[0056] Step 303: Collect mechanical performance parameters, load distribution data, surface condition analysis data, and heat dissipation resistance of the dual bearing structure to construct a multi-dimensional real-time condition dataset. The surface condition analysis data includes the wear degree, corrosion degree, and heat dissipation resistance of the dual bearing structure.

[0057] Among them, mechanical performance parameters refer to the mechanical properties of bearing materials such as elastic modulus, fatigue strength and hardness; load distribution data refers to the magnitude and direction information of the radial and axial forces borne by the bearing during centrifuge operation; surface condition analysis data refers to the physical condition indicators of the bearing surface obtained through testing.

[0058] In some implementations, real-time load distribution data is first acquired using strain gauges mounted on a dual-bearing structure. Simultaneously, pre-stored mechanical property parameters of the bearing material (including elastic modulus and fatigue strength) are accessed. Image sensors then directly capture images of the bearing surface, and image processing algorithms quantify wear and corrosion. This data is then combined with airflow data measured by an anemometer to calculate the heat dissipation resistance. This effectively aligns the five types of data—mechanical property parameters, load distribution data, wear, corrosion, and heat dissipation resistance—by timestamps. After data standardization to eliminate dimensional differences, the data is integrated into a multi-dimensional real-time state dataset in matrix form.

[0059] When a centrifuge processes a blood sample at 4200 r / min, if the strain gauge acquires a radial load of 800 N on the dual bearings, and the pre-stored bearing steel elastic modulus is 210 GPa, image analysis shows a wear depth of 0.05 mm (wear grade 2) and a corrosion area ratio of 1.5% (corrosion grade 1), and the anemometer measures an airflow velocity of 3 m / s corresponding to a wind resistance value of 15 Pa, a real-time status dataset containing five dimensions [210 GPa, 800 N, 2, 1, 15 Pa] can be directly generated.

[0060] Step 304: Preprocess the multi-dimensional real-time state dataset by integrating a dedicated signal processing chip into the edge computing device to obtain standardized real-time state data.

[0061] In some implementations, multi-dimensional real-time state datasets are transmitted to edge computing devices via data interfaces. This allows the dedicated signal processing chip integrated in the edge computing device to receive the data and directly perform data cleaning. After removing outliers caused by sensor noise or transmission errors, a normalization algorithm is applied to scale the data of each dimension to a standard range (such as 0 to 1). Then, the feature extraction module can calculate statistical features (such as mean and variance) to generate a standardized state feature vector as output.

[0062] Based on the above technical solution, by deploying a multi-source sensor network, multiple physical parameters of the dual-bearing spindle and the pawl-torsion spring mechanism can be synchronously acquired directly, solving the monitoring blind spot problem caused by single sensors and isolated data in existing technologies, thereby improving the integrity of the overall data. Simultaneously, by using a data fusion algorithm, the vibration characteristics of the spindle and the pawl-torsion spring mechanism are coupled into a comprehensive vibration spectrum, achieving precise quantification of the overall vibration characteristics. This overcomes the shortcomings of traditional methods, such as one-sided vibration analysis and inability to reflect component interactions, thus enhancing the accuracy of fault early warning. Secondly, by combining mechanical performance parameters, load distribution data, and surface condition analysis data to construct a multi-dimensional real-time state dataset, the state assessment can be made more in-depth and structured, solving the problems of parameter limitations and superficial assessment in existing technologies, thereby supporting long-term operational prediction. Finally, by integrating a dedicated signal processing chip into an edge computing device for real-time preprocessing to generate standardized state feature vectors, data processing can be effectively made low-latency and high-efficiency, solving the response sluggishness problem caused by traditional reliance on central servers, thus ensuring the real-time control safety of the centrifuge during high-speed operation.

[0063] In one possible implementation of the embodiments of this application, combined with Figure 1-4 As shown, the mechanical performance parameters, load distribution data, surface condition analysis data, and heat dissipation resistance of the dual-bearing structure are collected to construct a multi-dimensional real-time state dataset. This can be achieved through the following steps 401 to 405, which are explained in detail below: Step 401: Obtain surface image data of the spindle in the dual-bearing structure, and use a dedicated image processing chip integrated on the edge computing device to extract surface feature parameters of the spindle in the dual-bearing structure, including wear and corrosion.

[0064] Among them, dedicated image processing chips refer to hardware accelerators, such as FPGAs or ASICs, integrated into edge computing devices to efficiently execute image processing algorithms.

[0065] Before use, the sampling frequency of a high-resolution image sensor (such as an industrial camera) is dynamically adjusted according to the centrifuge's operating status, and its image resolution is set to 1080p or higher. This allows for real-time acquisition of surface image data of the spindle in the dual-bearing structure, ensuring full coverage of the spindle surface. The acquired image data can then be directly transmitted to an edge computing device (such as the NVIDIA Jetson series) via a high-speed interface. The edge computing device integrates a dedicated image processing chip (such as Intel MyriadX, with a pre-loaded image recognition model based on convolutional neural networks). This allows the dedicated image processing chip (such as Intel MyriadX) to be trained on a large number of bearing surface images. The input image undergoes preprocessing, including grayscale conversion, noise filtering, and contrast enhancement, to highlight surface details. Texture and morphological features are then extracted again through multi-layer convolutional operations of the convolutional neural network, outputting wear values ​​(in micrometers, calculated by analyzing the variance of surface roughness changes) and corrosion percentages (calculated by identifying the ratio of pitted pixel area to the total area).

[0066] Step 402: Cross-validate the wear and corrosion with mechanical performance parameters and load distribution data and calculate the error rate. If the error rate exceeds the preset error threshold, trigger re-acquisition.

[0067] In some implementations, the extracted wear and corrosion values ​​are used as measured values ​​and cross-validated with mechanical performance parameters (such as elastic modulus) retrieved from the database and real-time load distribution data (such as radial force). A finite element analysis model is then used to calculate theoretical wear and corrosion values ​​based on the mechanical performance parameters and load data. The error rate formula (error rate = |measured value - theoretical value| / theoretical value × 100%) is then used to calculate the error rate between the measured and theoretical values. If the measured wear is 0.02 mm and the theoretical value is 0.019 mm, the error rate is approximately |0.02 - 0.019| / 0.019 × 100% ≈ 5.26%. The calculated error rate can then be directly compared with a preset error threshold (such as 5%). If the error rate exceeds the threshold, the data is considered inconsistent, automatically triggering a re-acquisition command to restart the image sensor and data acquisition process. If the error rate is below the threshold, the data is considered valid and directly output.

[0068] Cross-validation primarily includes a triple validation mechanism: First, time-series synchronization validation ensures that the timestamp deviation of all data acquisitions is less than 0.1 seconds. Second, physical consistency validation checks whether the wear and corrosion distribution patterns match the load direction. Finally, magnitude-based validation eliminates abnormal data exceeding material performance limits. Simultaneously, the finite element model used during validation automatically matches the parameter library based on the bearing type; for example, angular contact bearings are calculated using Hertzian contact theory, while deep groove ball bearings are calculated using the Pamgren life model.

[0069] It should be noted that the triggering conditions for cross-validation have adaptive characteristics. A lenient threshold (8%) is used during the centrifuge acceleration phase, a strict threshold (5%) is used during the steady-state operation phase, and validation is suspended during the emergency braking phase. Simultaneously, the re-acquisition command will prioritize activating the backup sensor array. If the master and slave sensor data are inconsistent three times consecutively, a calibration procedure is automatically triggered and an anomaly log is recorded. Finally, the validated data will also be stored in the encrypted database along with a timestamp, checksum, and version number.

[0070] Taking the processing of blood samples in a high-speed centrifuge as an example, if image analysis shows that the bearing wear is 0.025 mm, the finite element model, based on the current radial load of 600 N and axial load of 200 N, calculates a theoretical wear threshold of 0.023 mm, with an error rate of 8.7%, exceeding the threshold. At this point, a re-acquisition process will be immediately initiated. The second measurement result will be 0.024 mm, and when the error rate drops to 4.3%, the data will be validated. This effectively prevents measurement distortion caused by temporary oil film rupture, ensuring the accuracy of the condition assessment.

[0071] Step 403: Combine the real-time collected heat dissipation resistance data to construct a five-dimensional real-time status dataset consisting of mechanical performance parameters, load distribution data, wear degree, corrosion degree, and heat dissipation resistance.

[0072] In some implementations, real-time heat dissipation resistance data is collected using an anemometer (e.g., measuring airflow velocity within a centrifuge cavity using a hot-wire anemometer). This data is then aggregated with cross-validated mechanical performance parameters (e.g., elastic modulus retrieved from a database), load distribution data (e.g., real-time collected radial force), wear (e.g., values ​​extracted through image processing), and corrosion (e.g., pitting area percentage), aligning them to a unified timestamp to ensure data synchronization. The data processing module on the edge computing device standardizes the five dimensions of data (e.g., normalizing values ​​to the 0-1 range using a Min-Max scaling algorithm). This results in a direct integration and output in a five-dimensional matrix format. Each row of the five-dimensional real-time state dataset represents a state record at a specific point in time and is stored in an encrypted database.

[0073] Based on the above technical solution, wear and corrosion are extracted from the surface image data of the spindle to address the issues of high subjectivity and low efficiency caused by traditional reliance on manual visual inspection. Furthermore, the extracted wear and corrosion are cross-validated with pre-stored mechanical performance parameters and real-time load distribution data, effectively ensuring data reliability and consistency and avoiding errors from a single data source. Combined with real-time acquired heat dissipation resistance data, a five-dimensional real-time status dataset is constructed, consisting of mechanical performance parameters, load distribution data, wear, corrosion, and heat dissipation resistance. This achieves deep fusion and structured storage of multiple physical quantities, providing comprehensive input for implicit correlation analysis and real-time control of the centrifuge rotor. This effectively solves the technical problems of isolated data, incomplete monitoring, and poor adaptive capabilities in existing technologies, improving the accuracy, safety, and efficiency of the centrifuge.

[0074] In one possible implementation of the embodiments of this application, combined with Figure 1-5 As shown, the process of analyzing the implicit relationships between real-time state data and generating constraint rules can be achieved through the following steps 501 to 504, which are explained in detail below: Step 501: Perform industrial time series data preprocessing on the real-time status dataset. The dataset is divided into continuous data and status data. The continuous data is discretized into different levels, and the status data is binary encoded.

[0075] Industrial time-series data preprocessing involves cleaning, transforming, and standardizing time-series data from industrial equipment. Continuous data refers to parameters that change continuously in value, such as wind speed or vibration frequency. State-based data refers to discrete parameters that represent the state or category of equipment, such as damage level or mechanism status.

[0076] In some implementations, the real-time status dataset can first undergo data cleaning and standardized industrial time-series data preprocessing to ensure data quality. This involves dividing the preprocessed data into continuous and state-specific data. Continuous data (such as wind speed or temperature values) can be discretized into three levels—low, medium, and high—using preset industrial thresholds (e.g., dividing wind speed values ​​into low (below threshold), medium (threshold range), and high (above threshold)). State-specific data (such as the degree of spindle damage) can be converted into numerical form using binary encoding (e.g., 0 for no damage and 1 for minor damage). After discretization and encoding, the data is standardized into a format that the algorithm can process and recorded as a basic factor.

[0077] Step 502: Calculate the weighted support of each frequent itemset using the Apriori algorithm to construct a hierarchical itemset. The hierarchical itemset includes a first-level frequent itemset containing basic factors, a second-level frequent itemset containing coupling factors, and a third-level frequent itemset containing control target related items.

[0078] Hierarchical itemsets refer to sets of itemsets divided according to a hierarchical structure, primarily used to systematically organize elements in association rule mining. First-level frequent itemsets are itemsets composed of basic factors. Basic factors are single key parameters extracted from real-time state data, such as wind speed level or spindle damage status. Second-level frequent itemsets are itemsets composed of coupling factors. Coupling factors are interactive combinations of multiple basic factors, such as the correlation between wind speed and spindle vibration. Third-level frequent itemsets are itemsets composed of control target related items. Control target related items are combinations of factors directly related to driving fine-tuning parameters, such as a speed adjustment command corresponding to a specific state condition. Weights are importance values ​​assigned to itemsets, used to amplify the significance of high-risk items when calculating support.

[0079] Each discretized level label is considered as an independent basic factor (e.g., low wind speed, medium wind speed, and high wind speed after wind speed discretization; low vibration, medium vibration, and "high vibration" after vibration frequency discretization). At the same time, each state value after binary encoding is also considered as an independent basic factor. For example, no damage (0), slight damage (1), and severe damage (2) after encoding spindle surface damage; normal (0) and stuck (1) after encoding pawl-torsion spring mechanism state.

[0080] Before use, a minimum support threshold can be set directly based on historical data to determine the frequency of an itemset (e.g., setting the minimum support to 5% means that an itemset must appear at least 5% of all data records (e.g., samples from a centrifuge's operating time window) to be considered a frequent itemset). Then, the system calls a pre-processed dataset to prepare the input data. In this scenario, each data point represents the set of all basic factors that appear within a specific time window (e.g., every 10 minutes). For example, a time window might contain basic factors such as {medium wind speed, high vibration, medium bearing temperature, no damage}.

[0081] Therefore, the Apriori algorithm can be directly used to first scan the entire dataset and calculate the support of each basic factor (such as medium wind speed, high vibration, etc.) appearing individually. All individual factors with support not lower than a minimum support threshold (e.g., 5%) are selected to form first-level frequent itemsets. From these first-level frequent itemsets, all possible two-item combinations can be generated through self-joins, i.e., candidate second-level frequent itemsets (e.g., {medium wind speed, high vibration}, {medium wind speed, medium bearing temperature}, etc.). The dataset is then scanned again to calculate the support of each candidate second-level frequent itemset, and combinations with support exceeding a preset support threshold are retained to form second-level frequent itemsets. These second-level frequent itemsets are the initial coupling factors, effectively revealing the frequent co-occurrence relationship between two basic factors.

[0082] The process continues iterating directly from the second-level frequent itemsets, generating all possible combinations of two items through self-connection to produce third-level frequent itemsets. Simultaneously, during the construction of each frequent itemset, a weight is preset for each itemset; for example, a weight of 3 is assigned to "severe spindle damage," and a weight of 1 is assigned to the normal state.

[0083] Step 503: Perform dynamic pruning on the hierarchical itemsets in conjunction with the industrial scenario to generate association rules.

[0084] After constructing hierarchical itemsets, the mechanical working principle of the centrifuge can be directly analyzed (such as the physical constraints between wind speed level and the condition of key components (such as mechanical damage to the main shaft, pitting of dual bearings, and pawl jamming)) to determine which combinations have extremely low probabilities in reality (e.g., severe mechanical damage to the main shaft is almost impossible under low wind speed conditions because low wind speed usually corresponds to low load, making it difficult to cause severe damage). Thus, combinations with extremely low probabilities are directly transformed into constraints (e.g., defining "low wind speed + severe mechanical damage to the main shaft" as an invalid itemset and automatically removing such combinations during the candidate set generation stage of the Apriori algorithm). This effectively ensures that subsequent frequent itemset mining only focuses on physically reasonable associations, reducing computational redundancy.

[0085] Then, a dynamic support adjustment strategy is used to set a lower minimum support (e.g., 1%) for high-risk state itemsets (such as pitting in dual bearings and pawl jamming), and a higher support (e.g., 5%) for regular state itemsets, thereby ensuring that key fault associations are prioritized for discovery. After pruning, the Apriori algorithm can be used again to generate association rules from the remaining frequent itemsets to directly reflect the implicit relationships between factors in the industrial scenario (such as the coupling relationship between wind speed and component status).

[0086] Step 504: Use confidence and causality coefficient as dual thresholds to filter association rules and map them to a predefined fine-tuning parameter library to generate constraint rules.

[0087] In association rules, confidence level is the probability of the consequent occurring given the occurrence of the antecedent; it measures the reliability of the rule. Causality coefficient is a causal correlation index between variables calculated using methods such as mutual information; it distinguishes between genuine causality and spurious associations. Dual thresholds refer to setting minimum limits for both confidence level and causality coefficient; only rules that satisfy both thresholds are retained.

[0088] After generating association rules in the hierarchical itemsets, the confidence of each rule is directly calculated. For example, the support of rule A→B is the number of occurrences of itemset A∪B divided by the total number of transactions. Therefore, the confidence of each rule can be directly calculated using the confidence formula: Confidence (A→B) = Support (A∪B) / Support (A), where Support (A) is the frequency of occurrence of the antecedent A alone. A confidence threshold is also set (e.g., not less than 80%), and only rules with a confidence exceeding this threshold can proceed to the next step.

[0089] By introducing a causality coefficient and calculating the causal correlation between the antecedent and consequent in the rule through mutual information, and setting a causality coefficient threshold (such as not less than 0.5), we can effectively eliminate pseudo-association rules with high confidence but weak causality, thus ensuring that the rules have actual causal significance.

[0090] Finally, the rules for dual-threshold filtering are mapped to a predefined fine-tuning parameter library (such as reducing the drive current by 5% or adjusting the speed by ±2 r / min). For example, the rule "high wind speed and spindle damage" is mapped to the parameter library. Specific constraint rules, such as speed limits or current adjustment commands, can be generated directly based on the mapping results.

[0091] Based on the above technical solutions, a closed-loop structure can be effectively formed through structured preprocessing of industrial time-series data, hierarchical weighted itemset construction, dynamic pruning of industrial scenarios, and dual-threshold screening and control mapping. This enables precise correlation and adaptive regulation between the centrifuge rotor's operating status and speed control. In the data preprocessing stage, continuous data is discretized and hierarchically classified according to industrial thresholds and associated with equipment status thresholds. State-type data is binary encoded, and time-series windows can be divided according to the operating cycle to strengthen the time-series correlation of factors, solving the problems of messy industrial time-series data and strong randomness of single-time-point data in traditional data processing. In the itemset construction stage, a hierarchical itemset is introduced, consisting of basic factors, coupling factors, and control target related items. Weighted support is calculated for key damage states, effectively overcoming the limitations of the traditional Apriori algorithm's generalized itemset definition and failure to highlight high-risk factors in industrial scenarios. By incorporating domain knowledge pruning and dynamic support adjustment strategies based on the centrifuge's mechanical principles, invalid itemsets are effectively eliminated, and key fault correlations are avoided, solving the problems of computational redundancy and missing high-risk rules caused by the lack of industrial constraints in traditional pruning strategies. Finally, a dual threshold screening method using confidence level and causality coefficient is employed to eliminate false association rules, and valid rules are mapped to a predefined fine-tuning parameter library. Real-time control transformation is achieved by combining rule priority ranking. This effectively bridges the gap between state association mining and driver fine-tuning, overcoming the shortcomings of traditional algorithms being disconnected from industrial control processes and having insufficient rule applicability. The overall solution enables precise mining of implicit associations among multiple dimensions of factors, such as the physical state of the centrifuge rotor shaft, the pawl-torsion spring mechanism, the dual-bearing structure, and wind speed. The generated constraint rules can guide real-time adaptive adjustment of the rotor speed, effectively solving the problem in existing technologies where relying on fixed single data preset speeds fails to cope with parameter mutations and component state changes, leading to reduced centrifuge accuracy, safety, and efficiency. This improves the accuracy of fault warnings and the timeliness of control response.

[0092] In one possible implementation of the embodiments of this application, combined with Figure 1-6 As shown, the process of obtaining the real-time rotor speed and adjusting it in real time based on constraint rules can be achieved through steps 601 to 605, which are explained in detail below: Step 601: Collect rotor speed data in real time to obtain a real-time speed sequence.

[0093] In some implementations, the rotor's rotational speed is monitored in real time by a speed sensor mounted on the centrifuge drive unit at fixed time intervals (e.g., multiple samples per second), continuously generating speed signals. These signals are then converted into digital signals by an analog-to-digital converter to ensure the data can be processed by a microprocessor. This allows for the direct recording of the speed value at each point in time, which is then stored in a buffer or memory in timestamp order to form a real-time speed sequence. This real-time speed sequence will contain timestamps and corresponding speed values.

[0094] For example, when centrifuging DNA samples in a biological laboratory, if the target rotor speed is 4200 rpm, the speed sensor will collect speed data every 100 milliseconds. For example, if the collected sequence is [4195 rpm, 4202 rpm, 4198 rpm] and the corresponding timestamp [t1, t2, t3], a real-time speed sequence is formed to monitor speed fluctuations.

[0095] Step 602: Construct a real-time adjustment and control platform including a data input layer, a rule matching layer, and a control output layer, for receiving real-time speed sequences and constraint rules.

[0096] The real-time control platform consists of three layers: a data input layer (interface responsible for receiving and inputting data, including real-time speed sequences and constraint rules), a rule matching layer (logic processing layer for parsing constraint rules and matching them with the real-time speed sequence), and a control output layer (output layer for generating control signals based on the matching results). The platform integrates these layers to achieve real-time speed regulation. The real-time speed sequence is a sequence of rotor speed data arranged chronologically. Constraint rules are speed limits and fine-tuning parameter rules generated based on the spindle status data.

[0097] In some implementations, a real-time adjustment and control platform is constructed, comprising a data input layer, a rule matching layer, and a control output layer. During operation, the data input layer receives real-time speed sequences and pre-generated constraint rule data from a speed sensor via a communication interface. After ensuring data format consistency, the received data is transmitted to the rule matching layer in real time. The rule matching layer parses the received constraint rules, extracting speed limit conditions such as upper and lower speed limits and dynamic tolerance ranges, as well as fine-tuning parameters such as speed offset and current adjustment ratio. These conditions are then matched item by item against the real-time speed sequence, and a comparison algorithm is used to obtain the matching result, such as identifying whether the speed exceeds the limit or requires fine-tuning. The control output layer, based on the matching result from the rule matching layer, calls a predefined fine-tuning parameter library to calculate the specific speed adjustment value and generates the corresponding control signal. This control signal is then sent to the motor drive unit through the platform's output interface, completing the real-time adjustment of the rotor speed.

[0098] Step 603: The rule matching layer parses the constraint rules, extracts the speed limit conditions and fine-tuning parameters in the rules, and matches them with the real-time speed sequence to obtain the matching results. The speed limit conditions include the upper limit speed, the lower limit speed, and the dynamic tolerance range. The fine-tuning parameters include the speed offset and the current adjustment ratio.

[0099] In some implementations, the rule matching layer analyzes real-time state data (such as the physical state of the spindle, the state of the pawl-torsion spring mechanism, etc.) using the Apriori algorithm to generate a set of associated rules stored in a structured format, containing conditional and action items. Then, a parsing algorithm is initiated to read each constraint rule entry one by one, identifying the conditional part (such as logical conditions defined based on factors like wind speed and bearing condition) and the action part (such as the corresponding speed limit conditions and fine-tuning parameters). Through string matching or key-value extraction, the specific values ​​of the speed limit conditions are extracted from the conditional part; for example, the upper speed limit might be expressed as "≤4800rpm", the lower speed limit as "≥3000rpm", and the dynamic tolerance range as "±50rpm". Simultaneously, fine-tuning parameters are parsed from the action part, such as speed offset (e.g., "-100rpm" indicates a speed reduction offset) and current adjustment ratio (e.g., "0.95" indicates a 5% current reduction). The parsed parameters can then be directly converted into a standardized data format and compared in real-time with the real-time speed sequence to ensure that the extracted values ​​conform to the mapping relationships in the predefined fine-tuning parameter library. The matching degree can then be calculated using comparison logic (such as checking whether the real-time speed exceeds the upper or lower limit, or whether it is within the dynamic tolerance range). Finally, a matching result is generated based on the comparison result. For example, if the real-time speed is 5100 rpm and the upper limit is 5000 rpm, the matching result is marked as "out of limit" and associated with the corresponding fine-tuning parameters.

[0100] For example, when a centrifuge processes a protein solution, if the constraint rule specifies that "when the spindle vibration exceeds the limit, the upper limit speed is set to 4800 rpm, the speed offset is -200 rpm, and the current adjustment ratio is 0.95", the rule matching layer will parse the rule and extract the speed limit conditions (upper limit 4800 rpm, lower limit not specified, default, dynamic tolerance range ±50 rpm) and fine-tuning parameters (speed offset -200 rpm, current adjustment ratio 0.95), and then match them with the real-time speed sequence (such as the current speed of 4900 rpm). The matching result is identified as "exceeding the limit", and a fine-tuning command is triggered.

[0101] Step 604: Based on the matching results, the control output layer calls the predefined fine-tuning parameter library to calculate the speed adjustment value and generate a control signal.

[0102] The predefined fine-tuning parameter library refers to a pre-stored database containing parameters such as speed offset and current adjustment ratio.

[0103] In some implementations, the control output layer can directly identify (e.g., over-limit type or normal state) the matching results received from the rule matching layer, which include the real-time speed sequence and speed limit conditions (such as upper limit speed, lower limit speed, and dynamic tolerance range). It then calls a predefined fine-tuning parameter library and retrieves the corresponding fine-tuning parameters through a query interface, such as speed offset (positive or negative values ​​indicate increase or decrease) and current adjustment ratio (ratio value indicates current change). Based on the retrieved fine-tuning parameters, the speed adjustment value can be calculated (e.g., by adding or subtracting the speed offset from the current real-time speed using arithmetic operations). This calculated speed adjustment value can then be converted into a standardized control signal, ensuring it can be directly recognized and executed by the motor drive unit, thus completing the control signal generation process.

[0104] For example, if the matching result indicates that the real-time speed of 5100 rpm exceeds the upper limit speed of 5000 rpm, the control output layer calls the fine-tuning parameter library to obtain the speed offset of -150 rpm and the current adjustment ratio of 0.9, calculates the speed adjustment value of 4950 rpm, and generates the corresponding speed reduction control signal to send to the motor unit.

[0105] Step 605: The motor drive unit executes the control signal to adjust the rotor speed in real time.

[0106] The motor drive unit refers to the hardware component in the centrifuge responsible for receiving and executing control signals, which is used to drive the motor and adjust the output.

[0107] In some implementations, after receiving a control signal containing speed adjustment values ​​via a communication interface, the motor drive unit can directly parse the instructions in the control signal (such as identifying speed offset or current adjustment ratio). Then, through internal circuitry or a microprocessor, the instructions are converted into executable drive parameters for the motor (such as adjusting voltage, current, or frequency). The motor's power output can then be adjusted in real time based on these executable drive parameters (e.g., changing the motor speed via pulse width modulation (PWM)). This effectively feeds back the adjusted speed from the motor drive unit to the rotor system, allowing the rotor speed to quickly match the target value.

[0108] For example, if the control signal indicates that the rotor speed should be reduced from 5000 rpm to 4800 rpm, the motor drive unit will analyze the speed offset of -200 rpm in the signal and reduce the motor drive current accordingly, so that the rotor speed will smoothly decrease to the target value within seconds.

[0109] Based on the above technical solution, by constructing a three-layer control platform comprising a data input layer, a rule matching layer, and a control output layer, and specifically adapting it for efficient interaction between speed sequences and constraint rules, the problems of loose structure and inefficient data and rule matching in traditional regulation systems can be effectively solved. Simultaneously, the rule matching layer deeply analyzes constraint rules and extracts limit conditions such as upper limit speed and dynamic tolerance range, as well as fine-tuning parameters such as speed offset, and then accurately compares them with the real-time speed sequence. This effectively overcomes the limitations of existing technologies that simply compare speed thresholds and ignore dynamic operating condition adaptation. Furthermore, the control output layer calculates adjustment values ​​and generates control signals based on the matching results, calling a predefined fine-tuning parameter library, which are then executed instantly by the motor drive unit. This effectively establishes a real-time response channel across the entire chain of data acquisition, rule matching, and execution adjustment, overcoming the shortcomings of traditional speed regulation such as lag in response and lack of precise basis for adjustment parameters. This allows the overall solution to address the problems in existing technologies, such as reliance on manual intervention for speed adjustment, inability to adapt to sudden changes in centrifugation parameters and equipment status, leading to spindle overload or poor separation performance, through structured and refined control logic. This improves the stability of centrifuge operation, the accuracy of speed control, and the safety and efficiency of equipment operation.

[0110] The above primarily describes the solutions of the embodiments of this application from the perspective of device implementation. It is understood that each device, such as a centrifuge rotor identification and control system, includes at least one of the hardware structures and software modules corresponding to the execution of each function in order to achieve the above-mentioned functions. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in a hardware-driven or software-driven manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0111] This application embodiment can divide a centrifuge rotor identification and control system into functional units based on the above method example. For example, each function can be divided into separate functional units, or two or more functions can be integrated into the same processing unit. The integrated unit can be implemented in hardware or as a software functional unit. It should be noted that the unit division in this application embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0112] When using integrated units, such as Figure 7-8As shown, the above embodiment illustrates a possible structural diagram of a centrifuge rotor identification and control system. The centrifuge rotor identification and control system includes a communication unit, a processing unit, a dedicated integrated module, a hierarchical control platform, and a multi-source sensor network. The communication unit is used to input experimental data and collect sensor data for comparison and verification, generating centrifugation parameters, including the type, volume, mass, and concentration of centrifuge tubes and solutions. The processing unit scans the rotor ID and retrieves historical data, constructs a speed-centrifugation parameter curve model, and determines the target speed based on the centrifugation parameters. It monitors the real-time status data of the main shaft in the rotor's dual-bearing structure, analyzes the implicit correlations between real-time status data, and generates constraint rules. It acquires the rotor's real-time speed and adjusts it in real-time based on the constraint rules. The dedicated integrated module includes an edge computing device integrating a dedicated signal processing chip and a dedicated image processing chip for preprocessing and feature extraction of real-time status data. The system adapts to different centrifuge operating conditions through dynamic deployment and remote upgrades. The hierarchical control platform includes a data input layer, a rule matching layer, and a control output layer. The data input layer receives real-time speed sequences and constraint rules; the rule matching layer parses the rules and extracts speed limit conditions and fine-tuning parameters; and the control output layer generates control signals based on the matching results and executes real-time adjustments through a motor drive unit. A multi-source sensor network is deployed on the rotor's dual-bearing main shaft, including vibration sensors, temperature sensors, deformation sensors, position sensors, pressure sensors, and anemometers. This network collects data on the main shaft's physical state, the pawl-torsion spring mechanism's state, the dual-bearing structure's state, and wind speed, and calculates the overall vibration characteristics through a data fusion algorithm.

[0113] In one possible implementation, the processing unit further includes an association rule mining module integrated on an edge computing device within a dedicated integrated module. This module analyzes implicit associations between real-time state data to generate constraint rules. The association rule mining module includes: a data preprocessing unit, used to perform industrial time-series data preprocessing on the real-time state dataset, discretizing continuous data into multiple levels and binary encoding state data; an itemset construction unit, used to analyze and calculate the weighted support of each frequent itemset using the Apriori algorithm to construct hierarchical itemsets, including first-level frequent itemsets containing basic factors, second-level frequent itemsets containing coupling factors, and third-level frequent itemsets containing control target association items; a rule generation unit, used to dynamically prune the hierarchical itemsets in conjunction with the industrial scenario to generate association rules, and uses confidence and causality coefficient dual thresholds for filtering; and a control mapping unit, used to map association rules to a predefined fine-tuning parameter library to generate constraint rules.

[0114] In one possible implementation, the process of dynamically deploying and remotely upgrading the association rule mining module also includes: real-time acquisition of the centrifuge's current operating condition data, which includes the rotor's real-time speed, main shaft physical status data, pawl-torsion spring mechanism status data, dual bearing structure status data, and wind speed data; based on the current operating condition data, calculating the matching degree between the current operating condition and the preset operating condition model using the Euclidean distance formula; when the matching degree is lower than a preset threshold, automatically generating a module update request; based on the module update request, reconfiguring the operating parameters of the data preprocessing unit, itemset construction unit, rule generation unit, and control mapping unit of the association rule mining module using a dedicated deployment chip via hardware acceleration; dynamically loading the update package, which includes optimized algorithm parameters, rule base, and weight configuration.

[0115] In one possible implementation, the association rule mining module also includes an adaptive learning unit, which dynamically updates the weights and support thresholds of association rule mining based on historical operating data and real-time feedback using an online learning algorithm. Specifically, this includes: a data stream processing submodule, used to receive real-time data on the physical state of the main shaft, the pawl-torsion spring mechanism state, the dual-bearing structure state, and wind speed collected by a multi-source sensor network, and perform time-series alignment and noise filtering; a weight update submodule, which uses a sliding window mechanism to analyze the frequency of frequent itemsets in historical operating data, calculates the weight adjustment amount using a gradient descent algorithm, and dynamically updates the weighted support weights in the itemset construction unit; a threshold optimization submodule, which adaptively adjusts the confidence threshold and causality coefficient threshold using a recursive least squares method based on the historical error rate of the confidence and causality coefficients; and a deployment interface submodule, used to load the optimized algorithm parameters via remote upgrade.

[0116] During its operation, the data stream processing submodule receives real-time data on the physical state of the main shaft, the pawl-torsion spring mechanism, the dual-bearing structure, and wind speed from a multi-source sensor network. This allows for time-series alignment through timestamp synchronization, while filtering algorithms are applied to filter noise and improve data quality. This activates the weight update submodule, which analyzes the frequency of frequent itemsets in historical operational data using a sliding window mechanism. Specifically, it counts itemset frequencies within a fixed time window and calculates weight adjustments using a gradient descent algorithm, dynamically updating the weighted support weights in the itemset construction units to ensure adaptive changes in weights based on operational status.

[0117] The threshold optimization submodule, based on the historical error rate of confidence and causality coefficients (i.e., statistically analyzing past rule application error data), adaptively adjusts the confidence and causality coefficient thresholds using a recursive least squares method to optimize rule filtering accuracy. Finally, the deployment interface submodule loads the optimized algorithm parameters via remote upgrade, completing the entire update process.

[0118] For example, when the centrifuge is running at high speed, the data stream processing submodule receives abnormal spindle vibration data and increased wind speed. After time-series alignment and noise filtering, the weight update submodule analyzes historical frequent itemsets through a sliding window. It finds that the frequency of high wind speed is associated with increased vibration, and uses gradient descent to calculate the weight adjustment amount and update the weighted support weights. The threshold optimization submodule can then be called to adjust the confidence threshold based on historical error rates, making the rules more stringent. Finally, the deployment interface submodule remotely loads new parameters to improve system response speed.

[0119] Figure 7-8 If the integrated units in the process are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, in essence, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. Storage media for storing computer software products include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks.

[0120] In implementation, each step of the method provided in this embodiment can be completed by integrated logic circuits in the processor or by instructions in software form. The steps of the method disclosed in the embodiments of this application can be directly manifested as being executed by a hardware processor, or being executed by a combination of hardware and software modules in the processor.

[0121] The processor in this application may include, but is not limited to, at least one of the following: a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a microcontroller unit (MCU), or an artificial intelligence processor, etc., and other computing devices that run software. Each computing device may include one or more cores for executing software instructions to perform calculations or processing. The processor may be a standalone semiconductor chip or integrated with other circuits into a single semiconductor chip. For example, it may form a System-on-a-Chip (SoC) with other circuits (such as encoding / decoding circuits, hardware acceleration circuits, or various bus and interface circuits), or it may be integrated as a built-in processor within an ASIC. The ASIC of this integrated processor may be packaged separately or together with other circuits. In addition to the cores for executing software instructions to perform calculations or processing, the processor may further include necessary hardware accelerators, such as field-programmable gate arrays (FPGAs), programmable logic devices (PLDs), or logic circuits that implement dedicated logic operations.

[0122] The memory in the embodiments of this application may include at least one of the following types: read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions; random access memory (RAM) or other types of dynamic storage devices capable of storing information and instructions; or electrically erasable programmable-only memory (EEPROM). In some scenarios, the memory may also be compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.

[0123] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software programs, implementation can be, in whole or in part, in the form of a computer program product. This computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device containing one or more servers, data centers, etc., that can be integrated with the medium. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

[0124] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, the disclosure, and the appended claims, can understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple instances. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0125] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely illustrative descriptions of the application as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from the spirit and scope of this application. Thus, if such modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is also intended to include such modifications and variations.

Claims

1. A method for identifying and controlling centrifuge rotors, characterized in that, include: Input experimental data and collect sensor data for comparison and verification, and generate centrifugation parameters, including the type, volume, mass and concentration of centrifuge tubes and solutions; Scan the rotor ID and retrieve historical data to construct a speed-centrifugation parameter curve model, and determine the target speed based on the centrifugation parameters; The real-time status data of the main shaft in the rotor dual-bearing structure is monitored, and the implicit correlation between the real-time status data is analyzed to generate constraint rules. The real-time status data includes the physical status of the main shaft, the status of the pawl-torsion spring mechanism, the status of the dual-bearing structure, and the wind speed. The real-time rotor speed is obtained, and the real-time rotor speed is adjusted in real time based on the constraint rules.

2. A centrifuge rotor identification and control method according to claim 1, characterized in that, Scan the rotor ID and retrieve historical data to construct a speed-centrifugation parameter curve, specifically including: Scan the unique identifier on the rotor to obtain the rotor model, and retrieve the historical operating data of the corresponding rotor from the historical database. The historical operating data includes historical centrifugation parameters and historical actual speed records for multiple centrifugation tasks. Using a multinomial regression curve fitting algorithm, with the historical centrifugation parameters as independent variables and the historical actual rotational speed as dependent variable, a multi-dimensional rotational speed-centrifugation parameter curve model is constructed to create a personalized curve for each rotor ID. Based on the centrifugation parameters, the target rotational speed is calculated by interpolation from the rotational speed-centrifugation parameter curve model.

3. A centrifuge rotor identification and control method according to claim 2, characterized in that, The process of monitoring the real-time status data of the spindle in a rotor dual-bearing structure specifically includes: Deploy a multi-source sensor network, which includes vibration sensors, temperature sensors, and deformation sensors for monitoring the dual-bearing structure; position sensors and pressure sensors for monitoring the pawl-torsion spring mechanism; and an anemometer for measuring the heat dissipation wind speed. Vibration characteristic data of the spindle in the dual-bearing structure and vibration characteristic data of the pawl-torsion spring mechanism are collected, and the overall vibration characteristics are calculated by a data fusion algorithm. The overall vibration characteristics are the comprehensive vibration spectrum after the dual-bearing structure and the pawl-torsion spring mechanism are coupled. A multi-dimensional real-time status dataset is constructed by collecting mechanical performance parameters, load distribution data, surface condition analysis data, and heat dissipation resistance of the dual-bearing structure. The surface condition analysis data includes the wear degree, corrosion degree, and heat dissipation resistance of the dual-bearing structure. By integrating a dedicated signal processing chip into an edge computing device, the multi-dimensional real-time state dataset is preprocessed to obtain standardized real-time state data.

4. A centrifuge rotor identification and control method according to claim 3, characterized in that, The process of collecting mechanical performance parameters, load distribution data, surface condition analysis data, and heat dissipation resistance of the dual-bearing structure to construct a multi-dimensional real-time condition dataset specifically includes: The surface image data of the spindle in the dual-bearing structure is obtained, and the wear and corrosion of the spindle surface in the dual-bearing structure are extracted using a dedicated image processing chip integrated on the edge computing device. The wear and corrosion rates are cross-validated with mechanical performance parameters and load distribution data, and the error rate is calculated. When the error rate exceeds a preset error threshold, re-collection is triggered. By combining real-time collected heat dissipation resistance data, a five-dimensional real-time status dataset is constructed, consisting of mechanical performance parameters, load distribution data, wear degree, corrosion degree, and heat dissipation resistance.

5. A centrifuge rotor identification and control method according to claim 4, characterized in that, The process of analyzing the implicit relationships between the real-time status data and generating constraint rules specifically includes: The real-time status dataset is preprocessed into industrial time-series data, which is divided into continuous data and status data. The continuous data is discretized into different levels, and the status data is binary encoded. The weighted support of each frequent itemset is calculated using the Apriori algorithm to construct a hierarchical itemset, which includes a first-level frequent itemset containing basic factors, a second-level frequent itemset containing coupling factors, and a third-level frequent itemset containing control target related items. Dynamically prune the hierarchical itemsets in conjunction with industrial scenarios to generate association rules; The association rules are filtered using a dual threshold of confidence level and causality coefficient, and then mapped to a predefined fine-tuning parameter library to generate constraint rules.

6. A centrifuge rotor identification and control method according to claim 5, characterized in that, The process of acquiring the real-time rotor speed and adjusting the real-time rotor speed in real time based on the constraint rules specifically includes: Real-time rotor speed data is collected to obtain a real-time speed sequence; A real-time adjustment and control platform is constructed, comprising a data input layer, a rule matching layer, and a control output layer, to receive real-time speed sequences and constraint rules; The rule matching layer parses the constraint rules, extracts the speed limit conditions and fine-tuning parameters in the rules, and matches them with the real-time speed sequence to obtain the matching result. The speed limit conditions include the upper limit speed, the lower limit speed, and the dynamic tolerance range. The fine-tuning parameters include the speed offset and the current adjustment ratio. Based on the matching result, the control output layer calls a predefined fine-tuning parameter library to calculate the speed adjustment value and generate a control signal. The motor drive unit executes the control signal to adjust the rotor speed in real time.

7. A centrifuge rotor identification and control system, characterized in that, Includes a communication unit, processing unit, dedicated integrated module, hierarchical control platform, and multi-source sensor network: The communication unit is used to input experimental data and collect sensor data for comparison and verification, and to generate centrifugation parameters, including the type, volume, mass and concentration of centrifuge tubes and solutions. The processing unit is used to scan the rotor ID and call historical data, construct a speed-centrifugal parameter curve model, and determine the target speed based on the centrifugal parameters; The system monitors the real-time status data of the spindle in the rotor dual-bearing structure, analyzes the implicit correlation between the real-time status data, and generates constraint rules; it obtains the real-time rotor speed and adjusts the real-time rotor speed in real time based on the constraint rules. The dedicated integrated module includes an edge computing device integrating a dedicated signal processing chip and a dedicated image processing chip, used for preprocessing and feature extraction of the real-time status data, wherein the edge computing device is adapted to different centrifuge operating conditions through dynamic deployment and remote upgrade. The hierarchical control platform includes a data input layer, a rule matching layer, and a control output layer. The data input layer is used to receive real-time speed sequences and constraint rules. The rule matching layer is used to parse the rules and extract speed limit conditions and fine-tuning parameters. The control output layer generates control signals based on the matching results and performs real-time adjustments through the motor drive unit. The multi-source sensor network is deployed on the rotor's dual-bearing main shaft and includes vibration sensors, temperature sensors, deformation sensors, position sensors, pressure sensors, and anemometers. It is used to collect data on the physical state of the main shaft, the state of the pawl-torsion spring mechanism, the state of the dual-bearing structure, and wind speed, and to calculate the overall vibration characteristics through a data fusion algorithm.

8. A centrifuge rotor identification and control system according to claim 7, characterized in that, The processing unit further includes an association rule mining module integrated on the edge computing device of the dedicated integrated module, used to analyze implicit associations between real-time state data to generate constraint rules. The association rule mining module includes: The data preprocessing unit is used to perform industrial time series data preprocessing on real-time status datasets, discretizing continuous data into multiple levels and encoding status data into binary. The itemset construction unit is used to analyze and calculate the weighted support of each frequent itemset using the Apriori algorithm to construct hierarchical itemsets. The hierarchical itemsets include first-level frequent itemsets containing basic factors, second-level frequent itemsets containing coupling factors, and third-level frequent itemsets containing control target related items. The rule generation unit is used to dynamically prune hierarchical itemsets in conjunction with industrial scenarios, generate association rules, and use confidence and causality coefficient dual thresholds for filtering. The control mapping unit is used to map association rules to a predefined fine-tuning parameter library to generate constraint rules.

9. A centrifuge rotor identification and control system according to claim 8, characterized in that, The process of dynamically deploying and remotely upgrading the association rule mining module specifically includes: Real-time acquisition of current operating condition data of the centrifuge, including real-time rotor speed, main shaft physical status data, pawl-torsion spring mechanism status data, dual bearing structure status data, and wind speed data; Based on the current operating condition data, the matching degree between the current operating condition and the preset operating condition model is calculated using the Euclidean distance formula. When the matching degree is lower than a preset threshold, a module update request is automatically generated. Based on the module update request, the operating parameters of the data preprocessing unit, itemset construction unit, rule generation unit, and control mapping unit of the association rule mining module are reconfigured through hardware acceleration using an integrated dedicated deployment chip, and an update package is dynamically loaded. The update package includes optimized algorithm parameters, rule base, and weight configuration.

10. A centrifuge rotor identification and control system according to claim 8, characterized in that, The association rule mining module also includes an adaptive learning unit, used to dynamically update the weights and support thresholds of association rule mining based on historical operating data and real-time feedback through an online learning algorithm, specifically including: The data stream processing submodule is used to receive real-time data on the physical state of the spindle, the pawl-torsion spring mechanism, the dual bearing structure, and wind speed from a multi-source sensor network, and to perform timing alignment and noise filtering. The weight update submodule uses a sliding window mechanism to analyze the frequency of frequent itemsets in historical running data, calculates the weight adjustment amount through the gradient descent algorithm, and dynamically updates the weighted support weights in the itemset building unit. The threshold optimization submodule adaptively adjusts the confidence threshold and the causality coefficient threshold using the recursive least squares method based on the historical error rate of the confidence level and the causality coefficient. The deployment interface submodule is used to load optimized algorithm parameters via remote upgrade.

Images