Multi-stage rotating speed coordination control method and device for three-stage magnetic suspension air compressor

Abstract

The invention discloses a multi-stage rotating speed coordination control method and device for a three-stage magnetic suspension air compressor, and the method comprises the following steps: obtaining the operation state data of each stage of compressor, and respectively installing a plurality of groups of sensors at the low-pressure stage, the medium-pressure stage and the high-pressure stage of the compressor, the plurality of groups of sensors comprise a pressure sensor, a temperature sensor, a vibration sensor, a rotating speed sensor and a power sensor. And each level of sensor monitors and records the gas pressure, temperature change, rotating speed fluctuation, vibration condition and power consumption of each level of the compressor in real time. According to the invention, through the combination of the quantum sensor and quantum calculation, the data acquisition and processing precision of the system is significantly improved, and especially in the aspects of high-frequency dynamic control and tiny disturbance monitoring, the accuracy of the system is greatly improved. The powerful parallel processing capacity of quantum computing can greatly reduce delay and errors in a traditional computing mode.

Description

Technical Field

[0001] This invention relates to the field of air compressor control technology, and in particular to a multi-stage speed coordination control device for a three-stage magnetic levitation air compressor. Background Technology

[0002] Magnetic levitation three-stage air compressors are widely used in new energy, precision manufacturing, and other fields due to their use of magnetic levitation bearings (no mechanical contact, low wear) and a three-stage compression structure (high energy efficiency, large air output). The core of the three-stage magnetic levitation air compressor is its magnetic levitation system, which uses magnetic force to eliminate physical contact between the rotor and stator, thereby significantly reducing friction loss and improving compressor efficiency. The three-stage system typically includes: a low-pressure stage (first stage): responsible for the initial compression of the gas; a medium-pressure stage (second stage): further compressing the gas; and a high-pressure stage (third stage): completing the final compression to achieve the required operating pressure. In this system, the speed control of each stage needs to be highly coordinated to ensure the optimal balance of gas flow, pressure, and energy efficiency.

[0003] Existing control methods often only control the speed of each stage individually, neglecting the coordination between different speed levels. This can easily lead to situations where some stages are overloaded while others are underloaded. The lack of coordination in speed control across different stages can cause a decrease in overall compressor efficiency and waste energy. This is especially true under conditions of large load fluctuations, where the speed of each stage cannot be optimized and adjusted according to real-time demand, resulting in low energy efficiency. Compressor load is often affected by external environmental factors and operating conditions, such as changes in gas demand and temperature, causing load fluctuations. Existing control methods typically cannot effectively distribute the load in real time, leading to overworking or inefficient operation of some stages. Different stages of the compressor need to have their load distributed according to demand. If the load distribution is unreasonable, some stages may frequently start and stop, increasing mechanical losses and energy consumption, and affecting system stability.

[0004] Existing control methods often rely on preset start-stop thresholds and cannot adaptively adjust to changes in actual load and operating conditions. This can lead to frequent start-stop cycles, causing mechanical damage and wasting energy. In multi-stage systems, overload at a single stage can degrade the performance of the entire system or damage equipment. In existing methods, overload protection mechanisms in multi-stage systems are often independently controlled, lacking global collaborative optimization and failing to effectively protect the operating states of each stage of the compressor across the entire system. In controlled systems, multi-stage compressor speed adjustments typically involve a certain response delay, which can lead to low energy efficiency or uneven system load, thus affecting operating efficiency. For rapidly changing operating environments (such as instantaneous load changes or gas pressure fluctuations), existing systems may not be able to achieve real-time rapid adjustments, resulting in a significant decrease in system efficiency. Speed ​​coordination in multi-stage compressor systems involves multiple variables, leading to high complexity in control algorithms. Existing methods often lack appropriate optimization, resulting in a heavy computational burden and an inability to process large amounts of dynamically changing data in real time. Traditional control algorithms may not be able to adaptively adjust to actual operating conditions, necessitating the consideration of more efficient scheduling and optimization strategies in algorithm design. Summary of the Invention

[0005] The present invention aims to at least partially solve one of the technical problems in the related art.

[0006] Therefore, the first objective of this invention is to propose a multi-stage speed coordination control method for a three-stage magnetic levitation air compressor.

[0007] Another objective of this invention is to provide a multi-stage speed coordination control device for a three-stage magnetic levitation air compressor.

[0008] The third objective of this invention is to provide a computer device.

[0009] The fourth objective of this invention is to provide a non-transitory computer-readable storage medium.

[0010] To achieve the above objectives, a first aspect of the present invention proposes a multi-stage speed coordination control method for a three-stage magnetic levitation air compressor, comprising: S1, acquire the operating status data of each stage of the compressor, install several sets of sensors in the low-pressure stage, medium-pressure stage and high-pressure stage of the compressor respectively, and transmit the operating status data of the compressor to the central control unit through the data bus, and perform data preprocessing, noise reduction, filtering and real-time analysis by the real-time computing platform; S2, based on a multimodal data fusion algorithm, the data from the several groups of sensors are synchronized in time, and feature extraction of real-time data is performed by combining a deep learning model; S3. Construct a reinforcement learning model, input real-time data, and optimize the multi-stage speed and load distribution of the compressor in real time. Based on the reward mechanism, the model integrates energy efficiency, equipment health status and load balance factors, and controls the speed ratio and start-stop timing of each stage of the compressor to automatically achieve energy-saving operation.

[0011] In one embodiment of the present invention, the plurality of sensors include a pressure sensor, a temperature sensor, a vibration sensor, a speed sensor, and a power sensor. The sensors are respectively arranged in the compressor components of the low-pressure stage, the medium-pressure stage, and the high-pressure stage. The compressor components include an air inlet, an air outlet, a rotor bearing, and a motor part, and monitor the parameters of the compressor's gas flow rate, pressure, vibration, and power consumption in real time.

[0012] In one embodiment of the present invention, a magnetic levitation bearing sensor is installed on the magnetic levitation bearing part of the compressor. The magnetic levitation bearing sensor monitors the changes between the rotor and the bearing. The data bus transmission is based on the data bus and uses a real-time communication protocol for transmission.

[0013] In one embodiment of the present invention, the deep learning model combines convolutional neural networks and long short-term memory networks to extract spatial features and temporal features respectively, and constructs a dynamic health assessment model for the operating status of each stage of the compressor.

[0014] In one embodiment of the present invention, the reinforcement learning model dynamically adjusts the load distribution and speed ratio of each stage of the compressor through a real-time feedback mechanism. The reward mechanism also includes automatically adjusting the compressor speed and load distribution based on real-time data, equipment health status, and load distribution effect. The real-time data includes speed data, pressure data, and load data.

[0015] In one embodiment of the present invention, the method further includes: fault diagnosis using edge computing nodes to perform anomaly detection and fault mode matching on sensor data in real time, and to generate fault warnings.

[0016] In one embodiment of the present invention, the cloud platform performs real-time analysis of long-term operating data based on big data analysis and transfer learning. The cloud platform automatically identifies and adjusts the energy efficiency scheduling strategy of the compressor through multi-device data clustering analysis.

[0017] In one embodiment of the present invention, the method further includes: based on an edge computing node, performing fault diagnosis and anomaly detection on the real-time data collected by the sensor, discovering potential faults and generating early warning information; The locally processed monitoring data and fault warning information are uploaded to the cloud platform, which uses a big data analysis model to perform cluster analysis on the long-term operating data of the equipment.

[0018] To achieve the above objectives, a second aspect of the present invention provides a multi-stage speed coordination control device for a three-stage magnetic levitation air compressor, comprising: The acquisition module is used to acquire the operating status data of each stage of the compressor. Several sets of sensors are installed in the low-pressure stage, medium-pressure stage and high-pressure stage of the compressor respectively. The operating status data of the compressor is transmitted to the central control unit through the data bus, and the real-time computing platform performs data preprocessing, noise reduction, filtering and real-time analysis. The data extraction module is used to synchronize the data from the several sets of sensors in time and combine them with a deep learning model to extract features from the real-time data. The speed control module is used to input real-time data and optimize the multi-stage speed and load distribution of the compressor in real time. Based on the reward mechanism, it integrates energy efficiency, equipment health status and load balance factors to control the speed ratio and start-stop timing of each stage of the compressor, and automatically achieve energy-saving operation.

[0019] This invention, through the combination of quantum sensors and quantum computing, significantly improves the accuracy of data acquisition and processing, especially in high-frequency dynamic control and monitoring of minute disturbances. The powerful parallel processing capability of quantum computing greatly reduces the latency and errors in traditional computing methods. This invention utilizes deep reinforcement learning combined with multi-task learning to optimize the coordination of compressor speeds at each stage and load distribution. This invention can adaptively adjust and optimize the operating state of each compressor in real time, avoiding the limitations of traditional static control. This invention employs quantum optimization algorithms to solve the combined optimization problem of multi-stage speed coordination, quickly finding the global optimal solution and improving the invention's responsiveness to load fluctuations. This invention introduces model-based reinforcement learning, enabling the invention not only to adapt to current operating conditions but also to perform long-term optimization, predicting and avoiding potential problems in advance, ensuring long-term efficient operation. This invention combines deep learning algorithms to accurately predict the start-up and shutdown timing of the compressor, intelligently controlling compressor start-up and shutdown, and providing early warnings of faults, reducing energy waste and damage to the invention. Through adaptive control and the self-learning capability of deep reinforcement learning, this invention can continuously optimize during long-term operation, avoiding energy efficiency degradation due to load fluctuations or changes in external conditions, ensuring that the compressor always maintains its optimal operating state.

[0020] To achieve the above objectives, a third aspect of this application provides a computer device comprising a processor and a memory; wherein the processor runs a program corresponding to the executable program code stored in the memory, for implementing a multi-stage speed coordination control method for a three-stage magnetic levitation air compressor as described in the first aspect embodiment.

[0021] To achieve the above objectives, the fourth aspect of this application proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a multi-stage speed coordination control method for a three-stage magnetic levitation air compressor as described in the first aspect embodiment.

[0022] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0023] Figure 1 This is a flowchart of a multi-stage speed coordination control method for a three-stage magnetic levitation air compressor according to an embodiment of the present invention; Figure 2 This is a structural diagram of a multi-stage speed coordination control device for a three-stage magnetic levitation air compressor according to an embodiment of the present invention; Figure 3 It is a computer device according to an embodiment of the present invention. Detailed Implementation

[0024] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0025] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0026] The following description, with reference to the accompanying drawings, describes a method and apparatus for multi-stage speed coordination control of a three-stage magnetic levitation air compressor according to an embodiment of the present invention.

[0027] Figure 1 This is a flowchart of a multi-stage speed coordination control method for a three-stage magnetic levitation air compressor according to an embodiment of the present invention, as shown below. Figure 1 As shown, it includes: S1: Obtain the operating status data of each stage of the compressor. Install several sets of sensors in the low-pressure stage, medium-pressure stage and high-pressure stage of the compressor respectively. Based on the operating status data of the compressor, transmit it to the central control unit through the data bus, and perform data preprocessing, noise reduction, filtering and real-time analysis by the real-time computing platform.

[0028] In one embodiment of the present invention, operating status data of compressors at each stage are collected to provide real-time feedback for subsequent control decisions and to provide basic data for load distribution and speed coordination optimization.

[0029] In one embodiment of the invention, a set of multifunctional sensors is installed at each compressor stage (low-pressure stage, medium-pressure stage, and high-pressure stage). Specifically, high-precision pressure sensors, temperature sensors, vibration sensors, speed sensors, and power sensors are selected. These sensors monitor the core operating parameters of each compressor stage, ensuring stable transmission and analysis of real-time data streams. A dedicated magnetic levitation bearing sensor is installed at the compressor's magnetic levitation bearing location. This sensor can accurately monitor the bearing's operating status, capture minute changes between the rotor and the bearing, and provide early warnings of any potential damage caused by improper speed coordination or friction. High-precision pressure sensors and flow sensors are installed at the airflow inlet and outlet locations of each compressor stage. These sensors continuously monitor gas pressure and flow, providing real-time feedback to help the system assess the load and operating efficiency of each compressor stage.

[0030] In one embodiment of the present invention, the low-pressure stage involves sensors installed at the compressor inlet, outlet, and rotor bearing to collect data on low-pressure gas flow, pressure fluctuations, and rotor operating status. The medium-pressure and high-pressure stages also involve corresponding sensors installed at the inlet / outlet, bearings, and motor to collect parameters such as gas state, rotational speed, vibration data, and power consumption.

[0031] In one embodiment of the invention, all data collected by the sensors is transmitted to the system's central control unit via a high-speed data bus, such as a CAN bus or Ethernet. Data transmission employs a real-time communication protocol to ensure stable and rapid data transmission to the control system under high dynamic load conditions.

[0032] In one embodiment of the invention, immediately after data acquisition, all information is transmitted to a high-performance computing platform built on an embedded real-time operating system. This computing platform is responsible for real-time analysis of the transmitted data and providing real-time feedback. The control platform is equipped with a multi-core processor and a GPU acceleration module to accelerate data processing.

[0033] In one embodiment of the present invention, the raw sensor data is filtered and denoised to remove high-frequency noise and interference while retaining the effective signal. This process is accomplished using techniques such as Kalman filtering or wavelet transform to ensure data accuracy. Principal component analysis is used to reduce the dimensionality of multidimensional data, thereby reducing data redundancy, improving data processing efficiency and computational accuracy, and ensuring that the system can still respond in real time under complex operating conditions.

[0034] In one embodiment of the present invention, a Kalman filter algorithm or a Bayesian filter algorithm is used to achieve multi-sensor data fusion, thereby improving the accuracy and reliability of the data. Through multi-source information fusion, the system can estimate the operating status of compressors at each stage in real time and identify potential anomalies.

[0035] In one embodiment of the present invention, a real-time Fourier transform (FFT) is performed on the vibration and pressure signals collected by the sensors to convert the time-domain signals into frequency-domain signals. This technique can help the system identify possible resonant frequencies or abnormal fluctuations in the compressor, and predict potential faults or overload problems in advance. By analyzing the frequency domain of the vibration data, potential harmonic interference or mechanical imbalances in the system can be identified.

[0036] In one embodiment of the present invention, during the analysis process, by combining frequency domain analysis algorithms such as FFT, the system can detect possible signs of failure (such as mechanical resonance, excessive vibration, abnormal temperature, etc.) in real time. If an abnormality is detected, the system will generate alarm information and provide appropriate response measures through the control platform, such as deceleration, load adjustment, or shutdown, to avoid further damage to the equipment.

[0037] In one embodiment of the present invention, by monitoring the gas flow rate, pressure, and power consumption data of each stage of the compressor in real time, the control system can dynamically evaluate the effect of the current load distribution. Once it is detected that the compressor at a certain stage is overloaded or its speed is not coordinated, the system will initiate corresponding speed adjustment or load distribution optimization strategies through feedback information, providing a basis for subsequent control decisions.

[0038] In one embodiment of the invention, precise sensor deployment and efficient data processing technology provide real-time monitoring and feedback data for the entire system. Through frequency domain analysis, data fusion, and other methods, the system can quickly identify problems and provide feedback. This real-time data forms the basis for all subsequent control decisions, directly affecting the execution of subsequent steps such as load distribution, speed adjustment, fault warning, and long-term optimization.

[0039] S2, based on a multimodal data fusion algorithm, synchronizes the data from the several groups of sensors in time, and extracts features from the real-time data by combining a deep learning model.

[0040] In one embodiment of the present invention, deep reinforcement learning is used to adaptively adjust the speed and load distribution at each level, thereby optimizing the working state of the compressor in real time and improving energy efficiency and system stability.

[0041] Deep Reinforcement Learning Model Design: In one embodiment of the present invention, a deep Q-network is used as the core reinforcement learning model. The deep Q-network utilizes a deep neural network to approximate the Q-value function, enabling the system to make optimal decisions through learning feedback.

[0042] In one embodiment of the invention, data such as speed, load, pressure, temperature, and vibration of each compressor level are included. This data is transmitted to the model in real time via sensors to form a description of the current system state. Key actions include: adjusting the speed of each level, start-up or shutdown control, adjusting gas flow rate, and load distribution. The action space design must meet actual control requirements, such as precise adjustment of load balancing and speed coordination. During training, the system continuously optimizes its decision-making strategy through interaction with the environment (i.e., the model takes actions based on the state, and the environment provides feedback). The training objective is to maximize cumulative rewards, i.e., maximizing energy efficiency and system stability. Energy efficiency reward: A positive reward is given if the adjusted compressor operation reduces power consumption or improves energy efficiency. Load balancing reward: A positive reward is given if the system successfully avoids overload or inefficient operation of a certain compressor level. System stability reward: A reward is also given if the system avoids frequent start-ups and shutdowns or oscillations.

[0043] Multi-stage speed coordination and optimization: In one embodiment of the invention, a multi-task learning method is employed, allowing a deep neural network to learn multiple tasks simultaneously. Specific tasks include load allocation, speed optimization, and energy efficiency improvement, thereby achieving speed coordination and load balance. This ensures uniform load distribution across compressor stages, preventing any single compressor stage from bearing excessive load. Under varying load conditions, the speed of each compressor stage is dynamically optimized, ensuring optimal efficiency during collaborative operation. By adjusting speed and load allocation, energy consumption is minimized, improving the overall system energy efficiency. The system continuously collects real-time data through sensors and inputs this data into a deep reinforcement learning model. Based on the model's predictions and decisions, the control system immediately adjusts the speed and load allocation of each compressor stage. When a compressor stage is overloaded, the model adjusts to reduce the load on that stage and transfer some of the load to other compressor stages. Furthermore, the system adjusts and relearns based on real-time feedback, allowing the model to continuously adapt in dynamic environments. Through each control adjustment, the system can optimize energy efficiency in real time, avoiding unnecessary energy loss. Precise speed and load adjustments prevent energy inefficiency caused by overload in any compressor stage.

[0044] Adaptive Load Allocation Algorithm: In one embodiment of this invention, the system combines deep reinforcement learning and fuzzy control algorithms. Through real-time sensor data analysis, it flexibly adjusts the load allocation of each compressor stage to ensure that each compressor operates within its optimal load range. When load fluctuations are significant, the fuzzy control algorithm provides more flexible adjustment capabilities, regulating the load of each compressor through fuzzy rules to prevent overload. By analyzing historical data and combining it with long short-term memory networks in deep learning, the system predicts load fluctuations over a future period and performs corresponding load allocation and speed adjustments in advance. This proactive control effectively avoids uneven load distribution and energy efficiency losses caused by large load fluctuations. When load predictions change, the system can adjust the operating status of each compressor in advance to prevent excessive load concentration in a particular stage and system instability. Based on real-time load and gas demand, the system can dynamically adjust the operating intensity of each compressor stage to ensure reasonable system load allocation and coordinated operation of each compressor stage. When a compressor stage experiences high load, other compressor stages automatically share the load, thereby preventing individual compressors from operating under overload conditions.

[0045] In one embodiment of the present invention, a combination of deep reinforcement learning and multi-task learning is employed, enabling the system to not only optimize the speed of each stage of the compressor in real time, but also to simultaneously address multiple tasks such as load allocation, speed optimization, and energy efficiency improvement within the same model. This avoids the limitation of traditional methods that can only optimize for a single task.

[0046] In one embodiment of the present invention, the system has a real-time feedback mechanism that can dynamically adjust the speed and load distribution of each stage of the compressor based on the data collected by the sensor, so as to ensure that the system can maintain the best operating state under load changes and external disturbances.

[0047] In one embodiment of the invention, fuzzy control provides finer-grained adjustments for load regulation, while deep reinforcement learning provides an intelligent optimization mechanism for decision-making. The combination of the two makes the system more flexible and adaptable.

[0048] In one embodiment of the present invention, by combining the LSTM network in deep learning, the system can perform optimization control in advance based on historical data and predicted load fluctuations, reduce the impact of sudden load fluctuations, and ensure more stable system operation.

[0049] In one embodiment of the present invention, by precisely coordinating the rotational speed and load, the system can maximize the system's energy efficiency and avoid energy waste while ensuring the stable operation of compressors at all levels.

[0050] S3. Construct a reinforcement learning model, input real-time data, and optimize the multi-stage speed and load distribution of the compressor in real time. Based on the reward mechanism, the model integrates energy efficiency, equipment health status and load balance factors, and controls the speed ratio and start-stop timing of each stage of the compressor to automatically achieve energy-saving operation.

[0051] In one embodiment of the present invention, a quantum optimization algorithm is used to globally optimize the rotational speed and load distribution at each level, thereby improving the energy efficiency of the system and reducing the computational burden, thus achieving fast and accurate system adjustment, especially when the dynamic load changes.

[0052] Application of the quantum approximation optimization algorithm: In compressor systems, the quantum approximation optimization algorithm is used to optimize the speed and load distribution of multiple compressor stages, ensuring that each compressor stage works collaboratively and maximizing the overall energy efficiency of the system. The optimization strategy calculated by the quantum algorithm ensures that the system dynamically adjusts the speed and load distribution under changing load conditions, avoiding overloading or energy efficiency degradation of any compressor stage.

[0053] In one embodiment of the present invention, an objective function is designed to measure the energy efficiency of the system. This function may include factors such as energy consumption, compressor stability, and load balancing. The optimization objective of the quantum approximation optimization algorithm is to minimize this objective function. Quantum bits are used to represent the speed and load adjustment parameters of each stage of the compressor. Quantum gates are applied to perform quantum operations to search the optimal solution space. After each quantum operation, the result of the quantum state is measured and converted into classical data, which is then used for subsequent optimization through classical computation. Finally, the solution output by the quantum approximation optimization algorithm is the optimal speed and load adjustment strategy. Because quantum computing can explore multiple solution spaces in parallel under quantum superposition states, the quantum approximation optimization algorithm can evaluate multiple possible speed and load adjustment schemes in a very short time. Traditional classical optimization algorithms cannot achieve such efficient parallel computation, which allows the quantum approximation optimization algorithm to significantly reduce response latency and improve control efficiency when dynamically adjusting the compressor system.

[0054] Global Control Optimization: In multi-stage compressor systems, the speed and load distribution of each compressor stage need to be coordinated. The quantum approximation optimization algorithm not only optimizes and adjusts each compressor stage but also performs optimal control globally, ensuring coordinated operation between compressor stages. The system uses quantum computing to quickly calculate the optimal speed adjustment and load distribution strategy under current load fluctuations. By addressing multi-dimensional decision variables and considering the mutual influence between compressor stages, quantum optimization effectively solves problems such as load imbalance and speed misalignment. The system may calculate how to evenly distribute the load across compressor stages or how to adjust the speed to maximize energy efficiency while ensuring stable system operation when the load increases.

[0055] In one embodiment of the present invention, the quantum approximation optimization algorithm can simultaneously handle load distribution problems at multiple compressor levels, preventing overload of any single compressor. When load fluctuations are significant, the system can optimize global load distribution using the quantum approximation optimization algorithm, ensuring balanced load across compressors and preventing individual compressors from experiencing low energy efficiency due to excessive load. Based on current load demands, the system uses the quantum optimization algorithm to quickly calculate the optimal speed of each compressor stage, achieving optimal energy efficiency and load balance.

[0056] Real-time Quantum Computing and Decision Making: In traditional compressor systems, load changes necessitate real-time calculations and adjustments, often hampered by computational latency. Quantum computing, however, enables the quantum approximation optimization algorithm to deliver optimal decisions in an extremely short time, achieving near real-time control and adjustment. Whenever the system detects a change in load or speed, real-time data is transmitted to the quantum computing module, where the quantum approximation optimization algorithm immediately performs calculations, rapidly optimizing the system's speed and load allocation strategy. The parallelism and ultra-high computational speed of quantum computing ensure the system's rapid response to every load fluctuation.

[0057] Reduced response latency: In one embodiment of the present invention, the quantum approximation optimization algorithm can leverage the advantages of quantum parallel computing to process a large solution space in a short time, greatly reducing the response latency in traditional classical optimization methods. In this way, the system can provide more timely decisions under rapidly fluctuating loads, avoiding the delayed response problem in traditional methods.

[0058] In one embodiment of the present invention, when the compressor load fluctuates greatly, the system can use real-time quantum optimization calculations to quickly adjust the speed of each stage of the compressor, ensuring that the system always operates in the optimal state.

[0059] In one embodiment of the present invention, the quantum approximation optimization algorithm in quantum computing enables the compressor system to quickly find the global optimum in a high-dimensional optimization space, avoiding the limitation of traditional algorithms that are prone to getting trapped in local optima. The parallelism of quantum computing provides unprecedented computational advantages for solving multi-stage speed coordination and load distribution problems.

[0060] In one embodiment of the present invention, quantum computing can quickly calculate the optimal solution when the load changes in real time, which greatly reduces the response delay in traditional control methods, improves the real-time adjustment capability of the system, and ensures that the system can continue to operate efficiently in a rapidly changing working environment.

[0061] In one embodiment of the present invention, quantum optimization can not only adjust individual compressor levels, but also achieve global load and speed coordination, ensuring that compressors at all levels work together and avoiding efficiency losses caused by local control.

[0062] In one embodiment of the present invention, an intelligent start-stop control and overload warning system is used to optimize start-stop timing, reduce energy waste, and achieve fault warning and protection to avoid equipment damage.

[0063] Intelligent start-stop control: Utilizes a Long Short-Term Memory (LSTM) network for start-stop decision prediction. Data input: The LSTM model's input includes multi-dimensional data from various compressor levels, such as: Load data: real-time load changes and predicted load values; Speed ​​data: current speeds of each compressor stage; Vibration data: compressor vibration signals, potentially reflecting the equipment's health status; Temperature and pressure data: operating temperature and pressure data to help determine if the equipment is under overload.

[0064] In one embodiment of the present invention, model training involves training an LSTM model using historical operating data. The model learns from historical load changes, speed adjustments, and start-stop history to predict future start-stop timings. The training objective is to minimize unnecessary start-stop cycles, thereby reducing energy waste and equipment wear.

[0065] In one embodiment of the present invention, the LSTM network can predict future load fluctuations. Based on the start-stop decision output by the model, the system will start the compressor during peak load periods and shut it down in time when the load decreases, thus avoiding unnecessary energy consumption.

[0066] In one embodiment of the invention, based on the prediction results of the LSTM model, the system uses an intelligent algorithm to determine the start-up and shutdown timing for each compressor level. This avoids equipment damage and energy waste caused by excessively frequent start-ups and shutdowns. The system can also optimize start-up and shutdown decisions through dynamic threshold adjustments. For example, the start-up and shutdown frequency may increase when the system load is low, but decrease when the load increases to avoid frequent switching.

[0067] In one embodiment of the invention, the system dynamically analyzes the matching degree between load and rotational speed to reduce unnecessary start-stop operations under low loads, thus avoiding energy waste. The system automatically determines whether to start or stop equipment based on actual needs, avoiding energy waste and system instability caused by frequent start-stop operations.

[0068] In one embodiment of the invention, reinforcement learning or genetic algorithms are used to optimize the start-stop strategy, making it more refined. Reinforcement learning can optimize start-stop decisions by exploring and utilizing equilibrium, further reducing equipment wear and energy consumption. Real-time overload prediction and protection: Overload prediction using a convolutional neural network (CNN): In one embodiment of the invention, the input data of the CNN model includes real-time vibration signals, temperature data, and pressure data from each stage of the compressor. These signals are continuously collected by sensors, and the CNN extracts features from the signals to determine whether there is an overload or fault risk.

[0069] In one embodiment of the invention, a CNN model is trained using historical data to learn signal characteristics under normal and overload conditions. The training set includes data from different states such as normal operation, high load, and equipment wear and tear.

[0070] In one embodiment of the present invention, CNN can identify potential fault risks, such as overload, overheating or abnormal vibration, in real time by learning the characteristics of different fault modes.

[0071] Real-time monitoring and early warning: By continuously monitoring the compressor's operating status, the CNN can instantly identify abnormal patterns. If the system detects abnormal signals such as vibration, pressure, or temperature, the CNN model will immediately issue an early warning.

[0072] In one embodiment of the present invention, when the CNN identifies that the compressor may be overloaded, the system will make adjustments in advance, such as adjusting the speed or switching the working mode, to prevent the equipment from being damaged due to overload.

[0073] In one embodiment of the invention, the monitoring system, in conjunction with real-time data feedback from the control system, is able to make necessary adjustments to the compressor, such as slowing down, stopping, or switching to a standby compressor level.

[0074] Fault protection and automatic adjustment: When the CNN detects a risk of overload or failure, the system can automatically take protective measures. In one embodiment of the present invention, the system reduces the compressor speed based on the warning signal to prevent the equipment from continuing to bear excessive load.

[0075] In one embodiment of the present invention, if a compressor level is overloaded, the system will automatically switch the working mode and enable other idle compressor levels to share the load, so as to avoid the single compressor from shutting down or being damaged due to overload.

[0076] In one embodiment of the present invention, the system automatically adjusts the compressor's operating strategy based on real-time monitored anomalies to ensure that the equipment does not work under overload and to reduce equipment wear and malfunctions.

[0077] Adaptive Start-Stop Strategy: Dynamic Start-Stop Strategy Adjustment: The system dynamically adjusts its start-stop strategy by combining load forecasting and real-time monitoring data. Through reinforcement learning or adaptive control algorithms, the system can automatically optimize start-stop control decisions based on current workload forecasts.

[0078] In one embodiment of the present invention, based on load change trends and the system's historical operating patterns, the system can predict future load fluctuations and adjust its start-up and shutdown strategies in advance. For example, when load fluctuations are not significant, the system may delay start-up and shutdown to avoid energy loss caused by frequent starts.

[0079] Reduce wear and tear caused by frequent start-stop: Through intelligent start-stop control and overload prediction, the system can effectively reduce unnecessary start-stop times and avoid equipment wear and energy loss caused by frequent start-stop.

[0080] In one embodiment of the present invention, the system will reasonably adjust the start-stop frequency according to the frequency and amplitude of load fluctuations to ensure the efficient operation of the equipment and avoid damage to the equipment caused by excessive start-stop.

[0081] Refined start-stop control: By combining real-time load monitoring and predictive models, the system can perform refined control, delaying start-stop when load fluctuations are small and starting-stopping promptly when load fluctuations are large, thereby achieving optimal energy utilization and equipment protection.

[0082] In one embodiment of the present invention, by using an LSTM network and a CNN model, the system can intelligently and in real-time predict start-up and shutdown timing, optimize start-up and shutdown control, and identify potential overload risks in advance. Compared with traditional start-up and shutdown control methods, this method can significantly improve the system's operating efficiency and prevent equipment failures.

[0083] In one embodiment of the present invention, the system can adaptively adjust the start-stop strategy based on load prediction and actual demand. By combining deep learning and reinforcement learning, the system can not only reduce energy waste from frequent start-stop operations, but also effectively avoid equipment damage caused by overload or unreasonable start-stop operations.

[0084] In one embodiment of the present invention, real-time monitoring using a CNN model enables timely identification of overloads or other potential faults, and allows for automatic adjustment measures. This gives the system enhanced fault warning and self-repair capabilities.

[0085] In one embodiment of the present invention, the system’s intelligent start-stop control and overload warning mechanism can significantly improve the energy efficiency of the equipment, while reducing unnecessary equipment wear and tear and extending the service life of the compressor.

[0086] In one embodiment of the present invention, through a self-learning and long-term optimization mechanism, the system can continuously adjust itself according to changes in the operating environment and improve long-term operating efficiency, optimize the overall performance of the system, and reduce long-term operation and maintenance costs.

[0087] Deep Reinforcement Learning Self-Learning Mechanism: Selection and Implementation of Reinforcement Learning Algorithms: Deep Reinforcement Learning (DRL) algorithms such as Deep Q-Network (DQN) and A3C (Actor-Critic) will be applied to the system's self-learning mechanism. These algorithms can gradually optimize control strategies by analyzing a large amount of historical operating data and adapt to various load fluctuations, environmental changes, and unstable operating states in the system.

[0088] In one embodiment of the present invention, the system continuously collects data generated during operation, including: Equipment Start-up / Stop History: Records the number of starts and stops, and the timing of each compressor stage. Load Fluctuation Data: Real-time monitoring of system load changes, especially data during sudden load fluctuations. Speed ​​Adjustment Data: Speed ​​change data for each compressor stage, including the impact of load changes on speed. System Health Data: Includes monitoring data on equipment vibration, temperature, pressure, etc., to help analyze the system's operational health. This data is then input into a deep reinforcement learning model for training. The Q-value function or policy gradient in deep learning is used to evaluate the effectiveness of different control strategies, with the goal of maximizing long-term cumulative rewards, i.e., improving energy efficiency, reducing failures, and optimizing equipment lifespan.

[0089] In one embodiment of the invention, the system continuously analyzes long-term operational data and improves the current control strategy through Q-value updates or strategy iterations. Deep reinforcement learning algorithms can automatically adjust the system's control strategy based on environmental changes. Self-adjustment: By learning from historical data, the system can self-adjust during long-term operation, avoiding the limitations of human intervention and achieving fully automated optimization. Dynamic adjustment: During long-term operation, the system continuously optimizes the control strategy through experience playback and delayed reward update mechanisms to adapt to different load demands, environmental changes, and operating conditions.

[0090] In one embodiment of the invention, feedback data from each run is used to update the model. Reinforcement learning learns the optimal strategy through repeated exploration and utilization, enabling the system to adjust control strategies under unstable conditions and adaptively optimize based on real-time load and equipment health status.

[0091] In one embodiment of the present invention, model-based reinforcement learning enables the system to predict future loads and environmental changes by learning a dynamic model of the environment, thereby making optimization decisions.

[0092] In one embodiment of the present invention, by constructing an environmental model of the system, the system can simulate future changes in workload and equipment health status, and make control decisions in advance based on these changes.

[0093] In one embodiment of the present invention, the system can adjust the operating strategy in advance based on the prediction results, optimize it before the future load increases or the equipment fails, and avoid overload, energy efficiency degradation or equipment damage.

[0094] In one embodiment of the invention, through adaptive optimization, the system can reduce unnecessary energy waste and equipment failures, thereby lowering long-term operation and maintenance costs. For example, when load demand is low, the system can save energy by adjusting the speed or shutting down certain compressor stages, while when load demand increases, the system automatically activates standby compressor stages to meet the demand. Energy-saving operation: Through long-term optimization, the system can optimize energy efficiency based on load fluctuations and historical data, reducing energy waste while ensuring stability.

[0095] In one embodiment of the present invention, based on prediction and optimization of long-term operating data, the system can identify potential faults in advance and adjust operating strategies, thereby avoiding equipment damage and excessive wear and tear, and extending the service life of the equipment.

[0096] In one embodiment of the present invention, the system utilizes reinforcement learning strategies and model predictions to automatically optimize the speed, load allocation, and start-stop decisions of each compressor based on predictions of future load, environmental changes, and system health. If an increase in future load is predicted, the system can make adjustments in advance by adjusting the speed, increasing the number of compressor levels in operation, etc., to ensure that there is no energy efficiency degradation or equipment overload problem when the load increases.

[0097] In one embodiment of the present invention, the system designs different control strategies according to different operating stages (such as start-stop stage, load adjustment stage, full-load operation stage, etc.). Start-stop stage: During the start-stop stage, the system optimizes start-stop timing to reduce unnecessary starts and stops, ensuring energy savings when load demand is low and rapid response when load demand increases. Load adjustment stage: During the load adjustment stage, the system can dynamically adjust the load and speed of each compressor stage to ensure the system is in optimal operating condition. Full-load operation stage: When the load is high, the system will activate more compressor stages and optimize speed and load distribution to ensure that the compressors can still operate efficiently under high load.

[0098] In one embodiment of the present invention, the system analyzes the operating mode of each working stage based on historical data and continuously optimizes the control strategy for each stage to achieve optimal load distribution and speed coordination. Data-driven phased decision-making: When the load is low, the system can choose to shut down certain compressor stages and reduce their speed to save energy; when the load is high, the system will automatically increase the number of operating compressor stages and improve energy efficiency by optimizing speed distribution.

[0099] In one embodiment of the present invention, during multi-stage operation, the system self-adjusts based on real-time feedback to ensure optimal operation at any stage. This self-learning mechanism, through repeated optimization and adjustment, ensures stable energy efficiency and reliability during long-term operation.

[0100] In one embodiment of the present invention, based on deep reinforcement learning and model-based reinforcement learning, the system can continuously adjust and optimize its control strategy through long-term operating data, thereby maximizing system energy efficiency and reducing failures and equipment wear.

[0101] In one embodiment of the present invention, the system can monitor load, equipment status, and environmental changes in real time, and dynamically optimize based on actual feedback and historical data. Through continuous learning and adaptive adjustment, the system can maintain high efficiency and stability during long-term operation.

[0102] In one embodiment of the present invention, the system can not only optimize according to the current working state, but also design refined control strategies according to different working stages (such as start-up, shutdown, load adjustment, full-load operation, etc.) to ensure that each stage can operate efficiently and stably.

[0103] In one embodiment of the present invention, through adaptive long-term optimization, the system can predict future demand based on load changes, adjust the operating strategy in advance to save energy, reduce operation and maintenance costs, and prevent equipment failure.

[0104] To achieve the above embodiments, such as Figure 2 As shown, this embodiment also provides a multi-stage speed coordination control device 10 for a three-stage magnetic levitation air compressor, comprising: The acquisition module 100 is used to acquire the operating status data of each stage of the compressor. Several sets of sensors are installed in the low-pressure stage, medium-pressure stage and high-pressure stage of the compressor respectively. The operating status data of the compressor is transmitted to the central control unit through the data bus, and the real-time computing platform performs data preprocessing, noise reduction, filtering and real-time analysis. The data extraction module 200 is used to synchronize the data from the several sets of sensors in time and combine them with a deep learning model to extract features from the real-time data. The speed control module 300 is used to input real-time data, optimize the multi-stage speed and load distribution of the compressor in real time, and control the speed ratio and start-stop timing of each stage of the compressor based on the reward mechanism, comprehensive energy efficiency, equipment health status and load balance factors, so as to automatically achieve energy-saving operation.

[0105] This invention, through the combination of quantum sensors and quantum computing, significantly improves the accuracy of data acquisition and processing, especially in high-frequency dynamic control and monitoring of minute disturbances. The powerful parallel processing capability of quantum computing greatly reduces the latency and errors in traditional computing methods. This invention utilizes deep reinforcement learning combined with multi-task learning to optimize the coordination of compressor speeds at each stage and load distribution. This invention can adaptively adjust and optimize the operating state of each compressor in real time, avoiding the limitations of traditional static control. This invention employs quantum optimization algorithms to solve the combined optimization problem of multi-stage speed coordination, quickly finding the global optimal solution and improving the invention's responsiveness to load fluctuations. This invention introduces model-based reinforcement learning, enabling the invention not only to adapt to current operating conditions but also to perform long-term optimization, predicting and avoiding potential problems in advance, ensuring long-term efficient operation. This invention combines deep learning algorithms to accurately predict the start-up and shutdown timing of the compressor, intelligently controlling compressor start-up and shutdown, and providing early warnings of faults, reducing energy waste and damage to the invention. Through adaptive control and the self-learning capability of deep reinforcement learning, this invention can continuously optimize during long-term operation, avoiding energy efficiency degradation due to load fluctuations or changes in external conditions, ensuring that the compressor always maintains its optimal operating state.

[0106] To implement the methods of the above embodiments, the present invention also provides a computer device, such as... Figure 3 As shown, the computer device 600 includes a memory 601 and a processor 602; wherein, the processor 602 reads the executable program code stored in the memory 601 to run a program corresponding to the executable program code, so as to implement the various steps of the multi-stage speed coordinated control method for a three-stage magnetic levitation air compressor described above.

[0107] To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a multi-stage speed coordination control method for a three-stage magnetic levitation air compressor as described in the foregoing embodiments.

[0108] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above 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 one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0109] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A method for multi-stage speed coordinated control of a three-stage magnetic levitation air compressor, characterized in that, Includes the following steps: The system acquires the operating status data of each stage of the compressor. Several sets of sensors are installed in the low-pressure stage, medium-pressure stage and high-pressure stage of the compressor. The operating status data of the compressor is transmitted to the central control unit through the data bus, and the real-time computing platform performs data preprocessing, noise reduction, filtering and real-time analysis. Based on a multimodal data fusion algorithm, the data from the several sets of sensors are synchronized in time, and feature extraction of real-time data is performed by combining a deep learning model. A reinforcement learning model is constructed, real-time data is input, and the multi-stage speed and load distribution of the compressor are optimized in real time. Based on the reward mechanism, the model integrates energy efficiency, equipment health status and load balance factors, controls the speed ratio and start-stop timing of each stage of the compressor, and automatically achieves energy-saving operation.

2. The method according to claim 1, characterized in that, The plurality of sensors include pressure sensors, temperature sensors, vibration sensors, speed sensors, and power sensors. The sensors are respectively installed in the compressor components of the low-pressure stage, medium-pressure stage, and high-pressure stage. The compressor components include an air inlet, an air outlet, a rotor bearing, and a motor. The sensors monitor the parameters of the compressor, such as gas flow rate, pressure, vibration, and power consumption, in real time.

3. The method according to claim 1, characterized in that, The compressor is equipped with a magnetic levitation bearing sensor at the magnetic levitation bearing location. The magnetic levitation bearing sensor monitors the changes between the rotor and the bearing. The data bus transmission is based on the data bus and uses a real-time communication protocol.

4. The method according to claim 1, characterized in that, The deep learning model combines convolutional neural networks and long short-term memory networks to extract spatial and temporal features, respectively, and constructs a dynamic health assessment model for the operating status of each stage of the compressor.

5. The method according to claim 1, characterized in that, The reinforcement learning model dynamically adjusts the load distribution and speed ratio of each stage of the compressor through a real-time feedback mechanism. The reward mechanism also includes automatically adjusting the compressor speed and load distribution based on real-time data, equipment health status, and load distribution effect. The real-time data includes speed data, pressure data, and load data.

6. The method according to claim 1, characterized in that, The method further includes: fault diagnosis using edge computing nodes to perform real-time anomaly detection and fault mode matching on sensor data, and generate fault warnings.

7. The method according to claim 1, characterized in that, The cloud platform is based on big data analysis and transfer learning to perform real-time analysis of long-term operating data. Through multi-device data clustering analysis, the cloud platform automatically identifies and adjusts the energy efficiency scheduling strategy of the compressor.

8. The method according to claim 1, characterized in that, The method further includes: Based on edge computing nodes, fault diagnosis and anomaly detection are performed on the real-time data collected by the sensors to discover potential faults and generate early warning information. The locally processed monitoring data and fault warning information are uploaded to the cloud platform, which uses a big data analysis model to perform cluster analysis on the long-term operating data of the equipment.

9. A multi-stage speed coordination control device for a three-stage magnetic levitation air compressor, characterized in that, include: The acquisition module is used to acquire the operating status data of each stage of the compressor. Several sets of sensors are installed in the low-pressure stage, medium-pressure stage and high-pressure stage of the compressor respectively. The operating status data of the compressor is transmitted to the central control unit through the data bus, and the real-time computing platform performs data preprocessing, noise reduction, filtering and real-time analysis. The data extraction module is configured to perform time-series synchronization of data from the several sets of sensors based on a multimodal data fusion algorithm, and to extract features from the real-time data by combining a deep learning model. The speed control module is configured to build a reinforcement learning model, input real-time data, and optimize the multi-level speed and load distribution of the compressor in real time. Based on the reward mechanism, it integrates energy efficiency, equipment health status and load balance factors to control the speed ratio and start-stop timing of each stage of the compressor, thereby automatically achieving energy-saving operation.

10. An electronic device, characterized in that, It includes a processor, a memory, and a communication interface. The memory stores a computer program. When the processor executes the computer program, it implements a multi-stage speed coordination control method for a three-stage magnetic levitation air compressor as described in any one of claims 1 to 8.

Images