Centrifugal machine intelligent operation parameter self-adaptive adjustment software system

Abstract

The invention relates to the technical field of intelligent control, in particular to a centrifugal machine intelligent operation parameter self-adaptive adjustment software system which comprises a parameter sensing and collecting module, a self-adaptive adjustment algorithm module, a state monitoring and fault early warning module, a data storage and analysis module and a human-computer interaction interface module. Wherein the parameter sensing and collecting module is used for collecting key parameters during operation of the centrifugal machine through a sensor and is responsible for high-speed data processing and analysis through a DSP chip. By fusing fuzzy control, reinforcement learning and an LSTM algorithm, intelligent identification and dynamic parameter optimization in a process stage are realized, the separation efficiency and precision are improved, an intelligent model is constructed based on material characteristics and real-time data, early fault detection and graded response are realized in combination with DSP spectral analysis, safe and stable operation of equipment is guaranteed, and the method is suitable for large-scale industrial production. And the centrifugal machine can operate more intelligently, safely and efficiently.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, in particular to a centrifuge intelligent operating parameter adaptive adjustment software system. Background Art

[0002] Centrifuges are core equipment that utilize centrifugal force to separate, grade, or concentrate mixed liquids. Their performance directly impacts production efficiency and process stability in the biopharmaceutical, chemical, and food industries. With the increasing demand for industrial intelligence, centrifuges are evolving from "single mechanical separation" to "intelligent process control," and the technical bottlenecks of traditional control systems are becoming increasingly prominent. Existing technologies often suffer from fixed parameters that cannot adapt to material changes, large fluctuations in separation efficiency, delayed fault warnings, and high system upgrade and maintenance costs.

[0003] Based on this, the present invention provides a centrifuge intelligent operating parameter adaptive adjustment software system to solve the above-mentioned technical problems. Summary of the Invention

[0004] The objective of the present invention is to provide a centrifuge intelligent operating parameter adaptive adjustment software system. The present invention integrates fuzzy control, reinforcement learning and LSTM algorithm to realize intelligent identification of process stages and dynamic optimization of parameters, improve separation efficiency and precision, build intelligent models based on material characteristics and real-time data, and combine DSP spectrum analysis to achieve early fault detection and graded response, ensuring safe and stable operation of equipment. Through local storage, edge AI analysis and digital twin technology, it supports online model updates and remote system upgrades, continuously optimizes system performance, and makes centrifuge operation more intelligent, safe and efficient.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The present invention provides a centrifuge intelligent operating parameter adaptive adjustment software system, including a parameter perception and acquisition module, an adaptive adjustment algorithm module, a status monitoring and fault warning module, a data storage and analysis module, and a human-computer interaction interface module, wherein:

[0007] The parameter sensing and acquisition module is used to collect key parameters of the centrifuge during operation through sensors, and is responsible for high-speed data processing and analysis through the DSP chip;

[0008] The adaptive adjustment algorithm module: Based on the collected parameters and process objectives, it automatically identifies the process stage and dynamically optimizes the parameters through a hybrid intelligent algorithm that integrates fuzzy control, reinforcement learning and LSTM time series prediction;

[0009] The status monitoring and fault warning module builds an intelligent fault identification model based on material characteristics and operating status, combines DSP real-time spectrum analysis technology for early anomaly detection, and ensures safe and stable operation of equipment through graded warnings and automatic emergency response strategies;

[0010] The data storage and analysis module is used to locally store and intelligently analyze operating parameters and process result data, implement online model updates based on the edge AI inference engine, and support continuous optimization and iteration of system performance through process digital twins and remote OTA upgrade mechanisms;

[0011] The human-machine interaction interface module is used to provide an intuitive operation interface, allowing operators to set parameters, monitor processes and obtain system feedback.

[0012] The parameter sensing and acquisition module includes a sensor unit, a signal processing unit, and a data analysis unit, wherein:

[0013] The sensor unit is used to deploy multiple types of sensors to collect key operating parameters of the centrifuge, such as speed, temperature, and vibration;

[0014] The signal processing unit is used to perform high-speed filtering, amplification and analog-to-digital conversion on the sensor collected signal through the DSP chip;

[0015] The data parsing unit is used to parse the processed data through the computing power of the DSP chip and respond in microseconds.

[0016] The adaptive adjustment algorithm module includes a hybrid algorithm unit, a process identification unit, and a parameter optimization unit, wherein:

[0017] The hybrid algorithm unit is used to integrate fuzzy control, reinforcement learning and LSTM time series prediction algorithms to provide adaptive adjustment core computing capabilities;

[0018] The process identification unit automatically identifies different process stages of centrifuge operation based on collected parameters and hybrid algorithms;

[0019] The parameter optimization unit is used to dynamically optimize the parameters of the centrifuge speed and operating time according to the process stage and process objectives.

[0020] The process objectives include cell recovery rate, protein crystal purity, blood component separation accuracy, and nanoparticle size control.

[0021] The hybrid algorithm unit integrates fuzzy control, reinforcement learning, and LSTM time series prediction algorithms to provide adaptive core computing capabilities. The specific operations are as follows:

[0022] A1: Load the preset rule base through fuzzy control to generate initial control parameters;

[0023] A2: Analyze the time series characteristics of historical operation data through LSTM prediction and output the predicted value of the turning point of the process stage;

[0024] A3: Calculate reward values ​​based on real-time process performance through reinforcement learning and dynamically adjust fuzzy rule weights;

[0025] A4: The fusion decision performs a weighted summation on the output results of the three algorithms, generates the final control instruction, and sends it to the centrifuge actuator.

[0026] The state monitoring and fault warning module includes a model building unit, an anomaly detection unit, and an early warning response unit, wherein:

[0027] The model building unit is used to combine material characteristics and operating status data to build an intelligent fault identification model;

[0028] The abnormality detection unit is used to perform early abnormality detection on the centrifuge operation status by using DSP real-time spectrum analysis technology;

[0029] The early warning response unit is used to execute graded early warnings and initiate automatic emergency response strategies based on the abnormality detection results to ensure the safe operation of the equipment.

[0030] The model building unit combines material characteristics and operating status data to build an intelligent fault identification model. The specific operations are as follows:

[0031] B1: Data collection and cleaning: Obtain key parameters of the centrifuge's speed, vibration, temperature, and characteristic data of material density and viscosity during operation, and remove outliers through sliding average filtering;

[0032] B2: Feature engineering step: perform wavelet packet decomposition on key operating parameters to extract frequency domain feature vector F, normalize material characteristic data to generate vector M, and combine them to obtain input feature matrix X = [F; M];

[0033] B3: Model architecture selection: Select a deep learning architecture based on the fault type, where:

[0034] ① For vibration faults, 1D convolutional neural network is used;

[0035] ②For temperature anomalies, use LSTM network;

[0036] B4: Model training: Take the feature matrix X and the corresponding fault label Y as the training set and use the Adam optimizer to minimize the following loss function:

[0037]

[0038] Where N is the number of samples, y i is the true label, P i Predict probabilities for the model;

[0039] B5: Model Validation and Optimization: Use cross-validation methods to evaluate model accuracy and optimize model performance by adjusting network hyperparameters.

[0040] The data storage and analysis module includes a data storage unit, an intelligent analysis unit, and an upgrade and iteration unit, wherein:

[0041] The data storage unit is used to locally store operating parameters and process result data to ensure data traceability;

[0042] The intelligent analysis unit: based on the edge AI inference engine, performs intelligent analysis on the stored data and mines optimization rules;

[0043] The upgrade and iteration unit is used to continuously optimize and iterate system performance through process digital twins and remote OTA upgrade mechanisms.

[0044] The intelligent analysis unit uses the edge AI inference engine to perform intelligent analysis on stored data and explore optimization rules. The specific operations are as follows:

[0045] C1: Sliding window normalization of centrifuge historical operating parameters, with a window length of 10s and an overlap rate of 50%;

[0046] C2: Feature extraction step: DSP is used to accelerate the calculation of the rotational speed fluctuation rate, temperature gradient change rate and time-frequency domain joint features;

[0047] C3: Model inference step: deploy the pruned and quantized LSTM network to predict the optimal speed-temperature parameter combination;

[0048] C4: Optimization suggestion generation step: Based on the prediction results and process goals, output parameter adjustment plan.

[0049] The human-computer interaction interface module includes an operation setting unit, a process monitoring unit, and a feedback interaction unit, wherein:

[0050] The operation setting unit is used to provide an intuitive interface for the operator to conveniently set the centrifuge operating parameters;

[0051] The process monitoring unit is used to display the centrifuge operation process and status in real time;

[0052] The feedback interaction unit is used to receive system feedback information and provide the operator with equipment operation results and abnormal prompts.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] The present invention integrates fuzzy control, reinforcement learning and LSTM algorithms to achieve intelligent identification of process stages and dynamic optimization of parameters, thereby improving separation efficiency and accuracy. It builds an intelligent model based on material characteristics and real-time data, and combines DSP spectrum analysis to achieve early fault detection and graded response, ensuring the safe and stable operation of the equipment. Through local storage, edge AI analysis and digital twin technology, it supports online model updates and remote system upgrades, continuously optimizes system performance, and makes the centrifuge operation more intelligent, safe and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a system diagram of a centrifuge intelligent operating parameter adaptive adjustment software system according to the present invention.

[0056] Figure 2 This is a flow chart of an adaptive adjustment algorithm in a centrifuge intelligent operating parameter adaptive adjustment software system of the present invention.

[0057] Figure 3 This is a flow chart of fault warning processing in a centrifuge intelligent operating parameter adaptive adjustment software system of the present invention.

[0058] Figure 4 This is a flow chart of edge AI analysis in a centrifuge intelligent operating parameter adaptive adjustment software system of the present invention.

[0059] Description of Figure Numbers:

[0060] 100. Parameter perception and acquisition module; 101. Sensor unit; 102. Signal processing unit; 103. Data analysis unit; 200. Adaptive adjustment algorithm module; 201. Hybrid algorithm unit; 202. Process identification unit; 203. Parameter optimization unit; 300. Status monitoring and fault warning module; 301. Model building unit; 302. Anomaly detection unit; 303. Warning response unit; 400. Data storage and analysis module; 401. Data storage unit; 402. Intelligent analysis unit; 403. Upgrade and iteration unit; 500. Human-computer interaction interface module; 501. Operation setting unit; 502. Process monitoring unit; 503. Feedback interaction unit. DETAILED DESCRIPTION

[0061] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0062] Example:

[0063] like Figures 1-4 As shown, this embodiment provides a centrifuge intelligent operating parameter adaptive adjustment software system, including a parameter perception and acquisition module 100, an adaptive adjustment algorithm module 200, a state monitoring and fault warning module 300, a data storage and analysis module 400, and a human-computer interaction interface module 500, wherein: the parameter perception and acquisition module 100: is used to collect key parameters of the centrifuge during operation through sensors, and is responsible for high-speed data processing and analysis through a DSP chip; the adaptive adjustment algorithm module 200: is based on the collected parameters and process goals, and performs automatic process stage identification and dynamic parameter optimization through a hybrid intelligent algorithm that integrates fuzzy control, reinforcement learning and LSTM time series prediction. ization; Status monitoring and fault warning module 300: Builds an intelligent fault identification model based on material characteristics and operating status, combines DSP real-time spectrum analysis technology for early anomaly detection, and ensures safe and stable operation of equipment through graded warning and automatic emergency response strategies; Data storage and analysis module 400: Used for local storage and intelligent analysis of operating parameters and process result data, realizes online model update based on edge AI inference engine, and supports continuous optimization and iteration of system performance through process digital twin and remote OTA upgrade mechanism; Human-computer interaction interface module 500: Used to provide an intuitive operation interface, allowing operators to set parameters, monitor processes and obtain system feedback.

[0064] Among them, it should be noted that the parameter perception and acquisition module 100 collects data and, after DSP processing, transmits it to the adaptive adjustment algorithm module 200 for process identification and parameter optimization. At the same time, the status monitoring and fault warning module 300 monitors the equipment status in real time based on the data. The data storage and analysis module 400 analyzes the data of the entire process and supports model upgrades. The human-computer interaction interface module 500 realizes the command interaction and information feedback between the operator and each module.

[0065] In this embodiment, it should also be noted that the parameter sensing and acquisition module 100 includes a sensor unit 101, a signal processing unit 102, and a data analysis unit 103, wherein: the sensor unit 101 is used to deploy multiple types of sensors to collect key operating parameters such as centrifuge speed, temperature, and vibration; the signal processing unit 102 is used to perform high-speed filtering, amplification, and analog-to-digital conversion on the sensor acquisition signals through a DSP chip; the data analysis unit 103 is used to analyze the processed data through the computing power of the DSP chip and respond in microseconds.

[0066] It should be noted that the sensor unit 101 is responsible for collecting basic data, the signal processing unit 102 pre-processes the original signal, and the data analysis unit 103 performs in-depth analysis on the processed data.

[0067] Furthermore, it should be noted that the speed acquisition adopts a magnetoelectric speed sensor (model: HCTL-2022, accuracy ±0.1%); the vibration monitoring adopts a three-axis accelerometer (model: ADXL345, range ±16g, resolution 0.001g); the temperature measurement adopts a Pt100 thermal resistor (model: PT100-PLA, temperature measurement range -200℃~850℃, accuracy ±0.5℃); the signal processing unit 102 uses DSP's fixed-point operation to implement IIR filtering, with a processing speed of 100MIPS; the data analysis unit 103 realizes zero-copy data transmission through the DSP's DMA channel, with specific indicators of microsecond response, such as analysis delay ≤10μs.

[0068] In this embodiment, it should also be noted that the adaptive adjustment algorithm module 200 includes a hybrid algorithm unit 201, a process identification unit 202, and a parameter optimization unit 203. The hybrid algorithm unit 201 is used to integrate fuzzy control, reinforcement learning, and LSTM time series prediction algorithms to provide core adaptive adjustment computing capabilities. The specific operations are as follows: A1: Using fuzzy control, a preset rule library is loaded to generate initial control parameters. A2: Using LSTM prediction, the temporal characteristics of historical operation data are analyzed to output predicted values ​​for process stage turning points. A3: Using reinforcement learning, a reward value is calculated based on real-time process performance, dynamically adjusting the fuzzy rule weights. A4: The fusion decision-making process performs a weighted summation of the outputs of the three algorithms to generate a final control instruction, which is then issued to the centrifuge actuator. The process identification unit 202 automatically identifies the different process stages of the centrifuge operation based on collected parameters and the hybrid algorithm. The parameter optimization unit 203 is used to dynamically optimize centrifuge speed and run time parameters based on the process stage and process objectives. Process objectives include cell recovery rate, protein crystal purity, blood component separation accuracy, and nanoparticle size control.

[0069] It should be noted that the hybrid algorithm unit 201 provides core computing capabilities by integrating three types of algorithms, providing support for the process identification unit 202 to identify the process stage based on the collected parameters and algorithm results. The results of the process identification unit 202 serve as the basis for the parameter optimization unit 203 to dynamically optimize parameters in combination with the process goals.

[0070] Furthermore, it should be noted that the reinforcement learning reward function is designed to design a multi-objective reward function for different process objectives:

[0071] Cell recycling scenario:

[0072] Nanoparticle synthesis scenario:

[0073] LSTM model structure: Input layer: receives a 10-dimensional feature vector (including time series data such as current speed, temperature, and vibration); Hidden layer: 2 layers of LSTM, 128 neurons per layer, a dropout rate of 0.2, and an activation function of tanh; Output layer: single-step prediction (such as predicting the speed fluctuation value in the next 10 seconds), using a linear activation function.

[0074] Fusion decision weight calculation: weighted summation formula and dynamic weight adjustment mechanism, ΔP = ω1·ΔP 模糊 +ω2·ΔP RL +ω3·ΔP LSTM , where the weight ω1+ω2+ω3=1, and is dynamically adjusted according to the process stage: acceleration stage: ω1=0.7 (fuzzy control dominated), ω2=0.2, ω3=0.1; constant speed stage: ω2=0.6 (reinforcement learning dominated), ω1=0.3, ω3=0.1; deceleration stage: ω3=0.5 (LSTM prediction dominated), ω1=0.3, ω2=0.2.

[0075] Cell recovery rate: used for whole cell recovery after high-density mammalian / insect cell culture; protein crystallization purity: suitable for inclusion body washing and protein precipitation / crystallization processes; blood component separation accuracy: for the efficient separation of plasma, red blood cells, albumin and other components in blood separation applications; nanoparticle size control: used for centrifugal classification after the synthesis of non-metallic or metallic nanoparticles.

[0076] In this embodiment, it should also be noted that the state monitoring and fault warning module 300 includes a model construction unit 301, an anomaly detection unit 302, and an early warning response unit 303, wherein: the model construction unit 301 is used to combine material characteristics and operating status data to build an intelligent fault identification model; the specific operations are as follows: B1: data acquisition and cleaning: obtain the key parameters of the centrifuge's speed, vibration, temperature and the characteristic data of the material density and viscosity during operation, and remove outliers through sliding average filtering; B2: feature engineering step: perform wavelet packet decomposition on the key operating parameters, extract the frequency domain feature vector F, normalize the material characteristic data to generate a vector M, and merge to obtain the input feature matrix X = [F; M]; B3: model architecture selection: select a deep learning architecture according to the fault type, wherein: ① for vibration faults, a 1D convolutional neural network is used; ② for temperature anomalies, an LSTM network is used; B4: model training: use the feature matrix X and the corresponding fault label Y as the training set, and use the Adam optimizer to minimize the following loss function:

[0077]

[0078] Where N is the number of samples, y i is the true label, P iModel prediction probability; B5: Model Validation and Optimization: Uses cross-validation to evaluate model accuracy and optimizes model performance by adjusting network hyperparameters. Anomaly Detection Unit 302: Utilizes DSP real-time spectrum analysis technology to perform early anomaly detection on the centrifuge's operating status. Early Warning Response Unit 303: Based on the anomaly detection results, executes graded early warnings and initiates automated emergency response strategies to ensure equipment safety.

[0079] Among them, it should be noted that the model building unit 301 constructs an intelligent fault recognition model by collecting and cleaning data, feature engineering processing, architecture selection, model training and verification optimization, etc., combining material characteristics and operating status data, to provide a basis for the anomaly detection unit 302 to use DSP real-time spectrum analysis technology for early anomaly detection, and the detection results output by the anomaly detection unit 302 serve as the key input for the early warning response unit 303 to execute hierarchical early warning and automatic emergency response strategies to ensure the safe operation of equipment.

[0080] Furthermore, it should be noted that material properties such as viscosity and density, and operating status data such as vibration spectrum and current fluctuations. Real-time spectrum analysis technology: DSP implements specific FFT parameters: sampling frequency: 10kHz (satisfies the Nyquist sampling theorem for vibration signals); window function: Hanning window (reduces spectral leakage); frequency resolution: 10Hz (calculated via a 1024-point FFT); feature extraction: calculate the energy distribution within the 10-1000Hz frequency band and extract the peak frequency and amplitude. Hierarchical warning strategy: defines three levels of warning thresholds and response actions, Level I warning (yellow): the vibration amplitude exceeds the baseline value by 1.5σ, the trigger condition is "5 consecutive sampling points exceed the threshold", the system prompts "abnormal vibration, it is recommended to check the load balance" on the human-machine interface; Level II alarm (orange): the vibration amplitude exceeds the baseline value by 2.5σ, the trigger condition is "3 consecutive sampling points exceed the threshold", the system automatically reduces the speed by 15% and records the abnormal event to the log; Level III shutdown (red): the vibration amplitude exceeds the baseline value by 3.5σ, or the temperature exceeds the material thermal sensitivity threshold, immediately triggering an emergency shutdown and cutting off the power supply, and sending a text message to notify the operation and maintenance personnel.

[0081] In this embodiment, it should also be noted that the data storage and analysis module 400 includes a data storage unit 401, an intelligent analysis unit 402, and an upgrade and iteration unit 403. The data storage unit 401 is used to locally store operating parameters and process result data to ensure data traceability. The intelligent analysis unit 402 uses an edge AI inference engine to intelligently analyze the stored data and discover optimization patterns. The specific operations are as follows: C1: Sliding window normalization of the centrifuge's historical operating parameters, with a window length of 10 seconds and a 50% overlap rate. C2: Feature extraction step: DSP-accelerated calculation of speed fluctuation rate, temperature gradient change rate, and time-frequency domain joint features. C3: Model inference step: Deployment of a pruned and quantized LSTM network to predict the optimal speed-temperature parameter combination. C4: Optimization suggestion generation step: Outputting parameter adjustment plans based on the predicted results and process objectives. The upgrade and iteration unit 403 is used to continuously optimize and iterate system performance through a process digital twin and remote OTA upgrade mechanism.

[0082] Among them, it should be noted that the data storage unit 401 locally stores the operating parameters and process result data, providing a data basis for the intelligent analysis unit 402 to perform standardization, feature extraction, model reasoning and optimization suggestion generation based on the edge AI inference engine and with the help of DSP acceleration processing. The optimization rules and generated solutions discovered by the intelligent analysis unit 402 serve as an important basis for the upgrade and iteration unit 403 to continuously optimize and iterate the system performance using process digital twins and remote OTA mechanisms.

[0083] Furthermore, it should be noted that the local storage solution uses industrial-grade SSDs (Samsung 870QVO, 2TB capacity, 500MB / s read / write speed) with RAID 1 data redundancy support. Data storage uses binary and CSV formats for dual backup, with a 100Hz sampling rate and a 365-day storage period. The edge AI inference engine reduces the LSTM network model size from 12MB to 5MB through pruning (removing connections with an absolute weight value less than 0.01, resulting in a 40% compression rate) and 8-bit fixed-point quantization (less than 5% loss of accuracy). Inference speed on the DSP is ≤5ms per data entry, with power consumption ≤3W.

[0084] In this embodiment, it should also be noted that the human-computer interaction interface module 500 includes an operation setting unit 501, a process monitoring unit 502, and a feedback interaction unit 503, wherein: the operation setting unit 501 is used to provide an intuitive interface for the operator to conveniently set the centrifuge operating parameters; the process monitoring unit 502 is used to display the centrifuge operating process and status in real time; the feedback interaction unit 503 is used to receive system feedback information and provide the operator with equipment operation results and abnormal prompts.

[0085] Among them, it should be noted that the operation setting unit 501 provides the operator with a parameter setting entry, transmits instructions to the system to drive the centrifuge operation, the process monitoring unit 502 presents the operation dynamics in real time, and the feedback interaction unit 503 collects the system operation results and abnormal information and feeds back to the operator.

[0086] Furthermore, it should be noted that the main interface is divided into three parts: ① Parameter setting area (upper left): contains input boxes for speed (default value 5000rpm, range 3000-10000rpm), temperature (default value 25℃, range 4-60℃), and time (default value 30min, range 5-120min); ② Real-time monitoring area (center): dynamic curves display speed, temperature, and vibration trends (sampling period 1 second), and historical data can view the records of the last 72 hours; ③ Alarm information area (lower right): scrolls and displays the last 10 warning / alarm records, including timestamp, level, and detailed description.

[0087] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0088] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to the specific embodiments described. Obviously, many modifications and variations are possible based on the content of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A centrifuge intelligent operating parameter adaptive adjustment software system, characterized in that: The system comprises a parameter sensing and acquisition module (100), an adaptive adjustment algorithm module (200), a state monitoring and fault warning module (300), a data storage and analysis module (400), and a human-computer interaction interface module (500), wherein: The parameter sensing and acquisition module (100) is used to collect key parameters of the centrifuge during operation through sensors, and is responsible for high-speed data processing and analysis through a DSP chip; The adaptive adjustment algorithm module (200) performs automatic process stage identification and dynamic parameter optimization based on the collected parameters and process targets through a hybrid intelligent algorithm integrating fuzzy control, reinforcement learning and LSTM time series prediction; The state monitoring and fault warning module (300) builds an intelligent fault identification model based on material characteristics and operating status, combines DSP real-time spectrum analysis technology to perform early abnormality detection, and ensures safe and stable operation of the equipment through hierarchical warning and automatic emergency response strategies; The data storage and analysis module (400) is used to perform local storage and intelligent analysis of operating parameters and process result data, realize online model updates based on the edge AI inference engine, and support continuous optimization and iteration of system performance through process digital twins and remote OTA upgrade mechanisms; The human-machine interaction interface module (500) is used to provide an intuitive operating interface, allowing the operator to set parameters, monitor the process, and obtain system feedback.

2. A centrifuge intelligent operating parameter adaptive adjustment software system according to claim 1, characterized in that: The parameter sensing and acquisition module (100) comprises a sensor unit (101), a signal processing unit (102), and a data analysis unit (103), wherein: The sensor unit (101) is used to deploy multiple types of sensors to collect key operating parameters of the centrifuge, such as speed, temperature, and vibration; The signal processing unit (102) is used to perform high-speed filtering, amplification and analog-to-digital conversion processing on the sensor collected signal through a DSP chip; The data parsing unit (103) is used to parse the processed data through the computing power of the DSP chip and perform microsecond-level response.

3. The centrifuge intelligent operating parameter adaptive adjustment software system according to claim 1, characterized in that: The adaptive adjustment algorithm module (200) includes a hybrid algorithm unit (201), a process identification unit (202), and a parameter optimization unit (203), wherein: The hybrid algorithm unit (201) is used to integrate fuzzy control, reinforcement learning and LSTM time series prediction algorithms, and provide adaptive adjustment core computing capabilities; The process identification unit (202) automatically identifies different process stages of the centrifuge operation based on the collected parameters and the hybrid algorithm; The parameter optimization unit (203) is used to dynamically optimize the parameters of the centrifuge speed and operating time according to the process stage and process target.

4. A centrifuge intelligent operating parameter adaptive adjustment software system according to claim 3, characterized in that: The process objectives include cell recovery rate, protein crystal purity, blood component separation accuracy, and nanoparticle size control.

5. A centrifuge intelligent operating parameter adaptive adjustment software system according to claim 4, characterized in that: The hybrid algorithm unit (201) integrates fuzzy control, reinforcement learning and LSTM time series prediction algorithm to provide adaptive adjustment core computing capabilities. The specific operations are as follows: A1: Load the preset rule base through fuzzy control to generate initial control parameters; A2: Analyze the time series characteristics of historical operation data through LSTM prediction and output the predicted value of the turning point of the process stage; A3: Calculate reward values ​​based on real-time process performance through reinforcement learning and dynamically adjust fuzzy rule weights; A4: The fusion decision performs a weighted summation on the output results of the three algorithms, generates the final control instruction, and sends it to the centrifuge actuator.

6. The centrifuge intelligent operating parameter adaptive adjustment software system according to claim 1, characterized in that: The state monitoring and fault warning module (300) comprises a model building unit (301), an anomaly detection unit (302), and an early warning response unit (303), wherein: The model building unit (301) is used to combine material characteristics and operating status data to build an intelligent fault identification model; The abnormality detection unit (302) is used to perform early abnormality detection on the centrifuge operation state by using DSP real-time spectrum analysis technology; The early warning response unit (303) is used to execute graded early warning and start automatic emergency response strategy according to the abnormality detection result to ensure the safety of equipment operation.

7. A centrifuge intelligent operating parameter adaptive adjustment software system according to claim 6, characterized in that: The model building unit (301) combines material characteristics and operating status data to build an intelligent fault identification model, and the specific operations are as follows: B1: Data collection and cleaning: Obtain key parameters of the centrifuge's speed, vibration, temperature, and characteristic data of material density and viscosity during operation, and remove outliers through sliding average filtering; B2: Feature engineering step: perform wavelet packet decomposition on key operating parameters to extract frequency domain feature vector F, normalize material characteristic data to generate vector M, and combine them to obtain input feature matrix X = [F; M]； B3: Model architecture selection: Select a deep learning architecture based on the fault type, where: ① For vibration faults, 1D convolutional neural network is used; ②For temperature anomalies, use LSTM network; B4: Model training: Take the feature matrix X and the corresponding fault label Y as the training set and use the Adam optimizer to minimize the following loss function: Where N is the number of samples, y i is the true label, P i Predict probabilities for the model; B5: Model Validation and Optimization: Use cross-validation methods to evaluate model accuracy and optimize model performance by adjusting network hyperparameters.

8. The centrifuge intelligent operating parameter adaptive adjustment software system according to claim 1, characterized in that: The data storage and analysis module (400) includes a data storage unit (401), an intelligent analysis unit (402), and an upgrade iteration unit (403), wherein: The data storage unit (401) is used to locally store operating parameters and process result data to ensure data traceability; The intelligent analysis unit (402) performs intelligent analysis on the stored data based on the edge AI inference engine to mine optimization rules; The upgrade iteration unit (403) is used to continuously optimize and iterate system performance through process digital twin and remote OTA upgrade mechanism.

9. The centrifuge intelligent operating parameter adaptive adjustment software system according to claim 8, characterized in that: The intelligent analysis unit (402) performs intelligent analysis on the stored data based on the edge AI inference engine to mine optimization rules. The specific operations are as follows: C1: Sliding window normalization of centrifuge historical operating parameters, with a window length of 10s and an overlap rate of 50%; C2: Feature extraction step: DSP is used to accelerate the calculation of the rotational speed fluctuation rate, temperature gradient change rate and time-frequency domain joint features; C3: Model inference step: deploy the pruned and quantized LSTM network to predict the optimal speed-temperature parameter combination; C4: Optimization suggestion generation step: Based on the prediction results and process goals, output parameter adjustment plan.

10. The centrifuge intelligent operating parameter adaptive adjustment software system according to claim 1, characterized in that: The human-machine interaction interface module (500) includes an operation setting unit (501), a process monitoring unit (502), and a feedback interaction unit (503), wherein: The operation setting unit (501) is used to provide an intuitive interface for the operator to conveniently set the centrifuge operating parameters; The process monitoring unit (502) is used to display the centrifuge operation process and status in real time; The feedback interaction unit (503) is used to receive system feedback information and provide the operator with equipment operation results and abnormality prompts.

Images