Pump turbine runner analysis method and system

Abstract

The invention discloses a pump turbine runner analysis method and system, and particularly relates to the technical field of runner analysis, and the system comprises a big data acquisition and processing module which is responsible for collecting data related to runner analysis from various sensors and external data sources and carrying out primary processing on the data. Through data cleaning and multi-source fusion, an accurate comprehensive data set is generated, a compression storage optimization module is used for efficient management, a fluid dynamics modeling simulation module is used for establishing a three-dimensional flow channel model, multi-physics field coupling and real-time dynamic simulation are combined, potential problems of the flow channel are predicted, design is improved through an optimization algorithm, and the flow channel model is optimized. Fault prediction is carried out through machine learning, rapid response and maintenance plan generation are carried out through an alarm and a knowledge base, parameters such as pump speed and valves are adjusted, an optimal strategy is made based on a multi-objective optimization algorithm, and system efficiency and reliability are improved by means of artificial intelligence aid decision making and a virtual assistant.

Description

Technical Field

[0001] The present invention relates to the technical field of flow channel analysis, and in particular to a method and system for analyzing the flow channel of a water pump turbine. Background Art

[0002] Pump turbines are important equipment in water conservancy projects, power stations and industrial fields, and their performance directly affects the efficiency of water resource utilization and energy conversion efficiency. As a core component, the flow channel design of the pump turbine determines the flow characteristics of the fluid and the overall performance of the equipment. Efficient flow channel design can not only reduce energy loss, but also reduce vibration and noise during equipment operation, thereby extending the service life of the equipment. However, the flow environment inside the flow channel is extremely complex and is affected by multiple factors such as pressure, temperature, and flow rate. Traditional design methods are difficult to fully consider these complex multi-physical field factors. With the diversification and complexity of modern engineering needs, how to optimize the flow channel design through precise simulation technology and data analysis to improve the efficiency and reliability of pump turbines has become a key issue that the industry needs to solve urgently.

[0003] The existing pump-turbine flow channel design and analysis methods have the following major defects in practical applications: the complex motion characteristics of the fluid in the flow channel, including turbulence, flow separation and cavitation, are often difficult to accurately simulate through traditional calculation methods, resulting in insufficient accuracy and efficiency of design optimization; data acquisition and processing technology is not perfect, the sensor layout is limited, and the collected data may have large errors due to the lack of multi-source fusion and real-time calibration; fluid dynamics simulation is often a single physical field analysis, lacking comprehensive consideration of multi-physical field coupling effects (such as flow, heat conduction, stress, etc.), and it is difficult to fully reflect the actual operating state in the flow channel; most of the existing optimization methods rely on the traditional trial and error mode, with a long design cycle and high cost, and cannot meet the efficient design requirements of complex engineering systems; the existing flow channel optimization algorithm lacks intelligence, and the optimization process relies more on engineering experience, making it difficult to quickly respond to potential problems in the flow channel operation. At the same time, there are limited preventive measures for key problems such as cavitation and flow separation, and it is difficult to significantly improve the operating efficiency and reliability of the equipment. Therefore, we provide a pump-turbine flow channel analysis method and system. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a pump turbine flow passage analysis method and system.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A pump-turbine flow passage analysis method and system, comprising a big data acquisition and processing module for collecting data related to flow passage analysis from various sensors and external data sources and performing preliminary processing on the data, a fluid dynamics modeling and simulation module for establishing an accurate model of the pump-turbine flow passage based on the collected data and performing dynamic simulation, an intelligent monitoring fault diagnosis module for using artificial intelligence and machine learning technology to monitor the system operation status in real time and automatically identify and diagnose potential faults, and an optimization control feedback adjustment module for dynamically adjusting the pump-turbine operation parameters according to the analysis and monitoring results;

[0007] The big data acquisition and processing module includes a sensor network management module for managing and coordinating sensors installed at key positions of pump turbines, a data synchronization timing correction module for processing data timestamps from different sensors to ensure data time synchronization, a data cleaning anomaly detection module for automatically identifying and filtering out noise and abnormal data generated during the acquisition process, a multi-source data fusion module for fusing data from different sources to form a comprehensive data set, and a data compression storage optimization module for using advanced data compression algorithms and storage optimization technologies to reduce storage space usage while ensuring fast data reading and efficient management;

[0008] The fluid dynamics modeling and simulation module includes a three-dimensional flow channel geometry modeling module for constructing a detailed three-dimensional geometry model of the flow channel using CAD and reverse engineering technology, a multi-physics field coupling simulation module for integrating coupling simulation of multiple physical fields such as fluid mechanics, structural mechanics and thermodynamics to comprehensively analyze complex phenomena in the flow channel, a real-time simulation prediction module for performing dynamic simulation based on real-time data to predict possible problems in the flow channel, a simulation result optimization module for adjusting flow channel design parameters through optimization algorithms, and a virtual reality simulation display module for displaying simulation results in the form of virtual reality to help engineers intuitively understand flow behavior and support design decisions.

[0009] The present invention is further configured as follows: the intelligent monitoring fault diagnosis module includes a real-time monitoring dashboard module for providing an intuitive user interface, displaying key operating parameters and flow channel status in real time, and supporting customized monitoring views; a machine learning driven fault prediction module for predicting possible faults by training machine learning models, analyzing historical and real-time data, and taking preventive measures in advance; an intelligent alarm notification module for automatically triggering alarms based on preset thresholds and model predictions, and sending alarm information to relevant personnel through multiple channels; a fault diagnosis knowledge base module for building and maintaining a knowledge base containing various fault modes and their diagnostic information, supporting rapid fault identification and processing; and an adaptive maintenance scheduling module for intelligently generating maintenance plans and schedules based on fault predictions and system operating status;

[0010] The optimization control feedback adjustment module includes an adaptive control algorithm module for adopting an adaptive control strategy to adjust parameters such as pump speed and valve opening according to real-time data to optimize the flow state in the flow channel, a multi-objective optimization decision module for comprehensively considering multiple objectives such as efficiency, energy consumption, stability, etc., and formulating the best operation strategy through a multi-objective optimization algorithm, a feedback loop management module for establishing a closed-loop feedback mechanism to continuously adjust the control strategy according to the system operation status and optimization results, an artificial intelligence assisted decision-making module for utilizing AI technologies such as deep learning and reinforcement learning to assist in decision-making, and a virtual assistant user interaction module for providing an intelligent virtual assistant that supports voice and text interaction to help operators understand the system status, adjust parameters and obtain suggestions.

[0011] The present invention is further configured as follows: the sensor network management module is responsible for coordinating the data collection of each sensor and verifying the integrity and accuracy of the data, and then passing the data to the data synchronization timing correction module for timestamp correction to ensure the time synchronization of the data; after the data synchronization timing correction module performs timestamp correction on the sensor data, the corrected data is passed to the data cleaning anomaly detection module, which identifies and filters the noise and abnormal data generated during the collection process; after the data cleaning anomaly detection module performs cleaning and anomaly detection on the data, the processed data is passed to the multi-source data fusion module, which fuses the data from different sources to form a comprehensive data set; the comprehensive data set generated by the multi-source data fusion module is passed to the data compression storage optimization module, which reduces the storage space occupancy through data compression algorithms and storage optimization technology, while ensuring fast reading and efficient management of data.

[0012] The present invention is further configured as follows: the three-dimensional flow channel geometry modeling module utilizes CAD and reverse engineering technology to construct a detailed three-dimensional geometry model of the flow channel, and then passes the constructed geometry model to the multi-physics coupling simulation module, which performs coupling simulation on multiple physical fields such as fluid, structure and thermodynamics in the flow channel; after the multi-physics coupling simulation module performs coupling simulation on multiple physical fields in the flow channel, the simulation data is passed to the real-time simulation prediction module, which performs dynamic simulation based on real-time data to predict possible problems in the flow channel; after completing dynamic simulation and problem prediction, the real-time simulation prediction module passes the simulation results to the simulation result optimization module, which adjusts the flow channel design parameters through an optimization algorithm; the optimized simulation results generated by the simulation result optimization module are passed to the virtual reality simulation display module, which displays the simulation results in the form of virtual reality, helping engineers to intuitively understand the flow behavior and support design decisions.

[0013] The present invention is further configured as follows: the real-time monitoring dashboard module is responsible for providing an intuitive user interface, displaying key operating parameters and flow channel status in real time, and passing these data to a machine learning driven fault prediction module, which predicts possible faults by analyzing historical and real-time data; after analyzing historical and real-time data, the machine learning driven fault prediction module predicts possible faults and passes the prediction results to an intelligent alarm notification module, which automatically triggers an alarm according to a preset threshold and prediction results, and sends alarm information to relevant personnel through multiple channels; after triggering the alarm, the intelligent alarm notification module passes the alarm information and fault data to a fault diagnosis knowledge base module, which maintains a knowledge base containing various fault modes and their diagnostic information, and supports rapid fault identification and processing; the fault modes and diagnostic information provided by the fault diagnosis knowledge base module are passed to an adaptive maintenance scheduling module, which intelligently generates maintenance plans and schedules based on fault predictions and system operating status.

[0014] The present invention is further configured as follows: the adaptive control algorithm module adjusts parameters such as pump speed and valve opening according to real-time data to optimize the flow state in the flow channel, and passes the adjustment results to the multi-objective optimization decision module, which comprehensively considers multiple objectives such as efficiency, energy consumption and stability to formulate the best operation strategy; the best operation strategy generated by the multi-objective optimization decision module is passed to the feedback loop management module, which establishes a closed-loop feedback mechanism to continuously adjust the control strategy according to the system operation state and optimization results; the system state and optimization results provided by the feedback loop management module are passed to the artificial intelligence auxiliary decision module, which uses deep learning and reinforcement learning techniques to assist decision-making and optimize the operation strategy; the auxiliary decision suggestions generated by the artificial intelligence auxiliary decision module are passed to the virtual assistant user interaction module, which helps the operator understand the system state, adjust parameters and obtain optimization suggestions through voice or text interaction to support efficient operation and management of the system.

[0015] The present invention is further configured to include the following steps:

[0016] S1, flow field measurement data analysis;

[0017] S2, multi-objective optimization analysis of flow channel numerical simulation;

[0018] S3. Optimization of multiple development control strategies for intelligent monitoring systems.

[0019] The present invention is further configured as follows: in the step S1, flow field measurement data analysis:

[0020] S1.1. Install a variety of high-precision sensors at key locations of the pump turbine (impeller inlet and outlet, guide vane area), including pressure sensors, flow rate sensors, temperature sensors and vibration sensors, to form a sensor network. The sensor layout needs to be optimized according to the flow channel structure characteristics and working conditions to ensure that all key monitoring points are covered. At the same time, materials with strong anti-interference performance, ceramic pressure sensors and vibration sensors with nickel-based alloy shells are selected;

[0021] S1.2. Input the collected sensor signals into the data acquisition terminal and perform time stamp correction through time synchronization technology (IEEE 1588 Precision Time Protocol) to ensure that all data have a consistent time reference. After synchronization, the signal is pre-processed using a Kalman filter to remove abnormal jitter and noise to ensure high signal accuracy.

[0022] S1.3. Through the data fusion algorithm (multi-sensor data fusion method based on Dempster-Shafer theory), the data from different types of sensors are comprehensively processed to form a unified data set containing parameters such as pressure, flow rate, temperature and vibration. During the fusion process, the data is verified to eliminate abnormal deviations;

[0023] S1.4. Use sparse matrix compression technology and efficient storage framework (HDF5 file format) to reduce data storage requirements to 10% to 20% of the original volume without losing key information. The storage optimization process is equipped with a high-speed data access interface.

[0024] The present invention is further configured as follows: in the step S2, multi-objective optimization analysis of flow channel numerical simulation:

[0025] S2.1. Use high-precision laser scanners to collect data on the inner surface of the pump-turbine flow channel, and build a detailed three-dimensional geometric model through reverse engineering software (Geomagic Design X). Perform unstructured meshing in CFD simulation software (ANSYS Fluent), focusing on encryption processing of the flow channel bends and blade surfaces to capture micro-flow phenomena;

[0026] S2.2. In the simulation platform (COMSOL Multiphysics), fluid mechanics, thermodynamics and structural mechanics are coupled for simulation to analyze the flow, heat exchange and structural stress distribution in the channel. Water and anti-corrosion fluid are used as the main working fluids, and the pipe wall material is wear-resistant stainless steel (304L). In the simulation, boundary parameters under actual operating conditions are applied;

[0027] S2.3. Based on the above coupling analysis results, a dynamic model is established in combination with real-time collected data to simulate possible emergencies (flow separation, cavitation), and the flow field changes within the next 5 to 10 seconds are dynamically predicted through the LSTM deep learning network to discover areas that may cause faults in advance;

[0028] S2.4. Use genetic algorithm (GA) to optimize the flow channel geometry design parameters, focusing on adjusting the blade installation angle, flow channel contraction ratio and guide vane position, with the goal of reducing the pressure loss by 15% to 20%. The optimization results are exported to the CAD model for revision and verification;

[0029] S2.5. Virtual reality technology (Unity3D) is used to present the simulation results as a dynamic three-dimensional scene. Engineers can observe the fluid flow characteristics, pressure distribution and eddy area through a head-mounted display, and interact with the model to adjust design decisions.

[0030] The present invention is further configured as follows: in the step S3, in the optimization of the multi-development control strategy of the intelligent monitoring system:

[0031] S3.1. Set up a multi-screen monitoring system in the central monitoring room, and use a custom dashboard to display key operating parameters such as pressure, temperature, and vibration in real time. The dashboard is connected to the sensor network based on the industrial protocol (MODBUS) and sets alarm thresholds;

[0032] S3.2. Build a machine learning model (Random Forest) based on historical data and real-time data to predict the probability of fault occurrence. The model training data comes from long-term operation monitoring results and a known fault case library. The fault prediction accuracy can reach more than 95%;

[0033] S3.3. Intelligent alarms are triggered by the prediction results of the comprehensive evaluation model and the set alarm thresholds. Alarm information is sent to engineers and maintenance teams via SMS, email and IoT devices to ensure rapid response to problems.

[0034] S3.4. Based on the predictive maintenance strategy, combined with the current operating status and historical maintenance records, an optimized maintenance plan is generated (regular replacement of high-wear parts), and the maintenance cycle is dynamically adjusted through the scheduling algorithm to extend the overall life of the equipment and reduce downtime;

[0035] S3.5. Use deep reinforcement learning technology (based on DQN strategy optimization) to adjust operating parameters (pump speed, valve opening) in real time to achieve the best balance between efficiency and stability. The feedback control mechanism ensures that the system can self-adjust under dynamic conditions and support long-term efficient operation.

[0036] The beneficial effects of the present invention are:

[0037] 1. The present invention generates an accurate comprehensive data set through data cleaning and multi-source fusion, and manages it efficiently with a compression storage optimization module. It uses a fluid dynamics modeling and simulation module to establish a three-dimensional flow channel model, combines multi-physical field coupling and real-time dynamic simulation, predicts potential flow channel problems and improves the design through optimization algorithms, uses machine learning to predict faults, quickly responds and generates maintenance plans through alarms and knowledge bases, adjusts parameters such as pump speed and valves, formulates the best strategy based on a multi-objective optimization algorithm, and uses artificial intelligence to assist decision-making and virtual assistants to improve system efficiency and reliability.

[0038] 2. The present invention arranges pressure, flow rate, temperature and vibration sensors at key positions, and adopts data fusion algorithm and Kalman filtering technology to ensure the high accuracy and consistency of collected data; combines CFD simulation with multi-physical field coupling analysis, dynamically simulates the flow, heat exchange and stress distribution in the flow channel, and predicts potential fault areas through LSTM deep learning model, effectively preventing problems such as flow separation and cavitation, optimizes the flow channel geometric parameters through genetic algorithm, reduces pressure loss by 15% to 20%, and uses virtual reality technology to intuitively present the optimization results, thereby improving the efficiency of design decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the system modules in the present invention.

[0040] Figure 2 Schematic diagram of the method flow in the present invention. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0042] Example 1

[0043] like Figure 1 As shown, a pump-turbine flow passage analysis system includes a big data acquisition and processing module for collecting data related to flow passage analysis from various sensors and external data sources and performing preliminary processing on the data, a fluid dynamics modeling and simulation module for establishing an accurate model of the pump-turbine flow passage based on the collected data and performing dynamic simulation, an intelligent monitoring fault diagnosis module for using artificial intelligence and machine learning technology to monitor the system operation status in real time and automatically identify and diagnose potential faults, and an optimization control feedback adjustment module for dynamically adjusting the pump-turbine operation parameters according to the analysis and monitoring results;

[0044] The big data acquisition and processing module includes a sensor network management module for managing and coordinating sensors installed at key positions of pump turbines, a data synchronization timing correction module for processing data timestamps from different sensors to ensure data time synchronization, a data cleaning anomaly detection module for automatically identifying and filtering out noise and abnormal data generated during the acquisition process, a multi-source data fusion module for fusing data from different sources to form a comprehensive data set, and a data compression storage optimization module for using advanced data compression algorithms and storage optimization technologies to reduce storage space usage while ensuring fast data reading and efficient management;

[0045] The fluid dynamics modeling and simulation module includes a three-dimensional flow channel geometry modeling module for constructing a detailed three-dimensional geometry model of the flow channel using CAD and reverse engineering technology, a multi-physics field coupling simulation module for integrating coupling simulation of multiple physical fields such as fluid mechanics, structural mechanics and thermodynamics to comprehensively analyze complex phenomena in the flow channel, a real-time simulation prediction module for performing dynamic simulation based on real-time data to predict possible problems in the flow channel, a simulation result optimization module for adjusting flow channel design parameters through optimization algorithms, and a virtual reality simulation display module for displaying simulation results in a virtual reality form to help engineers intuitively understand flow behavior and support design decisions;

[0046] The intelligent monitoring fault diagnosis module includes a real-time monitoring dashboard module for providing an intuitive user interface, displaying key operating parameters and flow channel status in real time, and supporting customized monitoring views; a machine learning driven fault prediction module for predicting possible faults by training machine learning models, analyzing historical and real-time data, and taking preventive measures in advance; an intelligent alarm notification module for automatically triggering alarms based on preset thresholds and model predictions, and sending alarm information to relevant personnel through multiple channels; a fault diagnosis knowledge base module for building and maintaining a knowledge base containing various fault modes and their diagnostic information, supporting rapid fault identification and processing; and an adaptive maintenance scheduling module for intelligently generating maintenance plans and schedules based on fault predictions and system operating status;

[0047] The optimization control feedback adjustment module includes an adaptive control algorithm module for adopting an adaptive control strategy to adjust parameters such as pump speed and valve opening according to real-time data to optimize the flow state in the flow channel, a multi-objective optimization decision module for comprehensively considering multiple objectives such as efficiency, energy consumption, stability, etc., and formulating the best operation strategy through a multi-objective optimization algorithm, a feedback loop management module for establishing a closed-loop feedback mechanism to continuously adjust the control strategy according to the system operation state and optimization results, an artificial intelligence auxiliary decision module for assisting decision making by using AI technologies such as deep learning and reinforcement learning, and a virtual assistant user interaction module for providing an intelligent virtual assistant that supports voice and text interaction to help operators understand the system state, adjust parameters and obtain suggestions;

[0048] The sensor network management module is responsible for coordinating the data collection of each sensor and verifying the integrity and accuracy of the data, and then passing the data to the data synchronization timing correction module for timestamp correction to ensure the time synchronization of the data; after the data synchronization timing correction module performs timestamp correction on the sensor data, the corrected data is passed to the data cleaning anomaly detection module, which identifies and filters the noise and abnormal data generated during the collection process; after the data cleaning anomaly detection module cleans and detects anomalies on the data, the processed data is passed to the multi-source data fusion module, which fuses the data from different sources to form a comprehensive data set; the comprehensive data set generated by the multi-source data fusion module is passed to the data compression storage optimization module, which reduces the storage space occupancy through data compression algorithms and storage optimization technology, while ensuring fast reading and efficient management of data;

[0049] The three-dimensional flow channel geometry modeling module uses CAD and reverse engineering technology to construct a detailed three-dimensional geometry model of the flow channel, and then passes the constructed geometry model to the multi-physics coupling simulation module, which performs coupling simulation on multiple physical fields such as fluid, structure and thermodynamics in the flow channel; after the multi-physics coupling simulation module performs coupling simulation on multiple physical fields in the flow channel, it passes the simulation data to the real-time simulation prediction module, which performs dynamic simulation based on real-time data to predict possible problems in the flow channel; after completing dynamic simulation and problem prediction, the real-time simulation prediction module passes the simulation results to the simulation result optimization module, which adjusts the flow channel design parameters through the optimization algorithm; the optimized simulation results generated by the simulation result optimization module are passed to the virtual reality simulation display module, which displays the simulation results in the form of virtual reality to help engineers intuitively understand flow behavior and support design decisions;

[0050] The real-time monitoring dashboard module is responsible for providing an intuitive user interface, displaying key operating parameters and flow channel status in real time, and passing these data to the machine learning driven fault prediction module, which predicts possible faults by analyzing historical and real-time data; after analyzing historical and real-time data, the machine learning driven fault prediction module predicts possible faults and passes the prediction results to the intelligent alarm notification module, which automatically triggers alarms according to preset thresholds and prediction results, and sends alarm information to relevant personnel through multiple channels; after triggering the alarm, the intelligent alarm notification module passes the alarm information and fault data to the fault diagnosis knowledge base module, which maintains a knowledge base containing various fault modes and their diagnostic information to support rapid fault identification and processing; the fault modes and diagnostic information provided by the fault diagnosis knowledge base module are passed to the adaptive maintenance scheduling module, which intelligently generates maintenance plans and schedules based on fault predictions and system operating status;

[0051] The adaptive control algorithm module adjusts parameters such as pump speed and valve opening according to real-time data to optimize the flow state in the flow channel, and passes the adjustment results to the multi-objective optimization decision module, which comprehensively considers multiple objectives such as efficiency, energy consumption and stability to formulate the best operation strategy; the best operation strategy generated by the multi-objective optimization decision module is passed to the feedback loop management module, which establishes a closed-loop feedback mechanism to continuously adjust the control strategy according to the system operation status and optimization results; the system status and optimization results provided by the feedback loop management module are passed to the artificial intelligence auxiliary decision module, which uses deep learning and reinforcement learning technologies to assist decision-making and optimize the operation strategy; the auxiliary decision suggestions generated by the artificial intelligence auxiliary decision module are passed to the virtual assistant user interaction module, which helps operators understand the system status, adjust parameters and obtain optimization suggestions through voice or text interaction to support efficient operation and management of the system.

[0052] In the above embodiment, the sensor network management module coordinates the data collected by each sensor, generates an accurate comprehensive data set through data cleaning and multi-source fusion, and manages it efficiently with the compression storage optimization module; then, the three-dimensional flow channel model is established by using the fluid dynamics modeling and simulation module, combined with multi-physics field coupling and real-time dynamic simulation, to predict potential problems in the flow channel and improve the design through the optimization algorithm; the intelligent monitoring module analyzes the operating status in real time, predicts faults through machine learning, and quickly responds and generates maintenance plans through alarms and knowledge bases; finally, the optimization control module dynamically adjusts parameters such as pump speed and valves, formulates the best strategy based on the multi-objective optimization algorithm, and improves the system efficiency and reliability with the help of artificial intelligence-assisted decision-making and virtual assistants. Through the collaborative work of various modules, the system realizes dynamic optimization of flow channel performance, fault prevention and intelligent maintenance, and significantly improves operating efficiency and stability.

[0053] Among them, this study is carried out on a real pump turbine model, the runner high pressure side diameter Dhi is 4.16m, the speed nd is 430r / min, and the specific speed ns is calculated according to the following formula:

[0054] ns=Cns

[0055] In the formula, Cns is the correction coefficient, which is 3.65, Qd is the design flow rate of the pump, and Hd is the design head of the pump. To facilitate comparative analysis, the dimensionless flow coefficient Cφ and the head coefficient Cψ are defined as follows:

[0056]

[0057]

[0058] Where H is the head, Q is the flow rate, g is the gravitational acceleration of 9.81m / s2, and Ω is the angular velocity of the impeller. In this example, the design flow coefficient is 0.043 and the specific speed ns is 115.0.

[0059] Example 2

[0060] like Figure 1-2 As shown, a method for analyzing a flow path of a pump turbine comprises the following steps:

[0061] S1, flow field measurement data analysis;

[0062] S2, multi-objective optimization analysis of flow channel numerical simulation;

[0063] S3, optimization of multi-development control strategies for intelligent monitoring systems;

[0064] In the step S1, flow field measurement data analysis:

[0065] S1.1. Install a variety of high-precision sensors at key locations of the pump turbine (impeller inlet and outlet, guide vane area), including pressure sensors, flow rate sensors, temperature sensors and vibration sensors, to form a sensor network. The sensor layout needs to be optimized according to the flow channel structure characteristics and working conditions to ensure that all key monitoring points are covered. At the same time, materials with strong anti-interference performance, ceramic pressure sensors and vibration sensors with nickel-based alloy shells are selected;

[0066] S1.2. Input the collected sensor signals into the data acquisition terminal and perform time stamp correction through time synchronization technology (IEEE 1588 Precision Time Protocol) to ensure that all data have a consistent time reference. After synchronization, the signal is pre-processed using a Kalman filter to remove abnormal jitter and noise to ensure high signal accuracy.

[0067] S1.3. Through the data fusion algorithm (multi-sensor data fusion method based on Dempster-Shafer theory), the data from different types of sensors are comprehensively processed to form a unified data set containing parameters such as pressure, flow rate, temperature and vibration. During the fusion process, the data is verified to eliminate abnormal deviations;

[0068] S1.4, using sparse matrix compression technology and efficient storage framework (HDF5 file format), reducing data storage requirements to 10% to 20% of the original volume without losing key information, and the storage optimization process is equipped with a high-speed data access interface;

[0069] In the step S2, multi-objective optimization analysis of flow channel numerical simulation:

[0070] S2.1. Use high-precision laser scanners to collect data on the inner surface of the pump-turbine flow channel, and build a detailed three-dimensional geometric model through reverse engineering software (Geomagic Design X). Perform unstructured meshing in CFD simulation software (ANSYS Fluent), focusing on encryption processing of the flow channel bends and blade surfaces to capture micro-flow phenomena;

[0071] S2.2. In the simulation platform (COMSOL Multiphysics), fluid mechanics, thermodynamics and structural mechanics are coupled for simulation to analyze the flow, heat exchange and structural stress distribution in the channel. Water and anti-corrosion fluid are used as the main working fluids, and the pipe wall material is wear-resistant stainless steel (304L). In the simulation, boundary parameters under actual operating conditions are applied;

[0072] S2.3. Based on the above coupling analysis results, a dynamic model is established in combination with real-time collected data to simulate possible emergencies (flow separation, cavitation), and the flow field changes within the next 5 to 10 seconds are dynamically predicted through the LSTM deep learning network to discover areas that may cause faults in advance;

[0073] S2.4. Use genetic algorithm (GA) to optimize the flow channel geometry design parameters, focusing on adjusting the blade installation angle, flow channel contraction ratio and guide vane position, with the goal of reducing the pressure loss by 15% to 20%. The optimization results are exported to the CAD model for revision and verification;

[0074] S2.5. Use virtual reality technology (Unity3D) to present simulation results as dynamic three-dimensional scenes. Engineers can observe fluid flow characteristics, pressure distribution and eddy areas through head-mounted displays, and interact with the model to adjust design decisions;

[0075] In the step S3, in the optimization of the multi-development control strategy of the intelligent monitoring system:

[0076] S3.1. Set up a multi-screen monitoring system in the central monitoring room, and use a custom dashboard to display key operating parameters such as pressure, temperature, and vibration in real time. The dashboard is connected to the sensor network based on the industrial protocol (MODBUS) and sets alarm thresholds;

[0077] S3.2. Build a machine learning model (Random Forest) based on historical data and real-time data to predict the probability of fault occurrence. The model training data comes from long-term operation monitoring results and a known fault case library. The fault prediction accuracy can reach more than 95%;

[0078] S3.3. Intelligent alarms are triggered by the prediction results of the comprehensive evaluation model and the set alarm thresholds. Alarm information is sent to engineers and maintenance teams via SMS, email and IoT devices to ensure rapid response to problems.

[0079] S3.4. Based on the predictive maintenance strategy, combined with the current operating status and historical maintenance records, an optimized maintenance plan is generated (regular replacement of high-wear parts), and the maintenance cycle is dynamically adjusted through the scheduling algorithm to extend the overall life of the equipment and reduce downtime;

[0080] S3.5. Use deep reinforcement learning technology (based on DQN strategy optimization) to adjust operating parameters (pump speed, valve opening) in real time to achieve the best balance between efficiency and stability. The feedback control mechanism ensures that the system can self-adjust under dynamic conditions and support long-term efficient operation.

[0081] In the above embodiments, by arranging pressure, flow rate, temperature and vibration sensors at key positions, and adopting data fusion algorithm and Kalman filtering technology, the high accuracy and consistency of the collected data are ensured; by combining CFD simulation and multi-physical field coupling analysis, the flow, heat exchange and stress distribution in the flow channel are dynamically simulated, and the potential fault area is predicted by the LSTM deep learning model, effectively preventing problems such as flow separation and cavitation; at the same time, the flow channel geometric parameters are optimized by genetic algorithm to reduce the pressure loss by 15% to 20%, and the optimization results are intuitively presented by virtual reality technology to improve the efficiency of design decision-making. During operation, the intelligent monitoring system predicts the probability of failure based on the machine learning model, and adjusts the operating parameters in real time through deep reinforcement learning to achieve the best balance between efficiency and stability; combined with predictive maintenance strategies, the maintenance plan is optimized, the service life of the equipment is significantly extended and the downtime is reduced. Experimental verification shows that the system has excellent reliability and efficiency in a high-pressure circulating water environment and is suitable for long-term operation under complex working conditions.

[0082] The laboratory uses a high-pressure circulating water system to simulate actual operating conditions, with a temperature range of 5°C-80°C, a flow range of 0.5-10 m³ / s, and a pressure range of 1-20 MPa. A vibration isolation platform is installed in the laboratory to reduce external interference.

[0083] The channel wall material is 304L stainless steel, the sensor material is ceramic and nickel-based alloy, and the fluid in the simulation is water (with a small amount of preservative);

[0084] The sensor sampling rate was set to 500 Hz, the data storage rate was 1 GB per hour, and the total experimental duration was 100 hours.

[0085] Working principle: When the present invention is in use, pressure, flow rate, temperature and vibration sensors are arranged at key positions of the water pump (impeller inlet and outlet and guide vane area) to collect real-time operation data in the flow channel. Through data cleaning and anomaly detection technology, noise and abnormal signals are removed, and after generating a comprehensive data set using multi-source data fusion technology, efficient storage optimization technology is used to manage these data to ensure that the analysis results can be quickly called while reducing storage space requirements. This data processing method improves the accuracy and consistency of flow channel operation data, providing a reliable foundation for subsequent simulation and optimization.

[0086] During the fluid dynamics modeling and simulation stage, the system constructs a detailed three-dimensional geometric model of the flow channel through reverse engineering and laser scanning technology, and simulates complex phenomena such as fluid flow, heat transfer, and structural stress distribution in CFD simulation and multi-physics field coupling analysis. Combined with the dynamic simulation prediction function, it can detect possible problems in the flow channel (flow separation or cavitation) in advance, and optimize the flow channel geometry design through genetic algorithms to reduce pressure loss by 15% to 20%. The optimized design is visualized through virtual reality technology, so that engineers can intuitively understand the fluid flow characteristics and adjust design decisions based on real-time feedback. This function effectively improves the efficiency and accuracy of flow channel design and reduces potential operational risks.

[0087] In actual operation, the intelligent monitoring system analyzes the operating status in real time, predicts the probability of potential faults through machine learning, and quickly identifies problems in combination with the fault diagnosis knowledge base to generate preventive maintenance plans, thereby reducing downtime and extending equipment life. At the same time, the system uses deep reinforcement learning to dynamically adjust parameters such as pump speed and valve opening to ensure the best balance between efficiency and stability. Through the closed-loop feedback mechanism and virtual assistant, operators can quickly grasp the operating status of the equipment and obtain optimization suggestions. This intelligent monitoring and control strategy not only improves operating efficiency and system reliability, but also adapts to the needs of long-term stable operation under complex working conditions, and is particularly suitable for harsh conditions such as high-pressure circulating water environments.

[0088] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A pump-turbine flow channel analysis system, characterized in that: It includes a big data acquisition and processing module for collecting data related to flow channel analysis from various sensors and external data sources and performing preliminary processing on the data, a fluid dynamics modeling and simulation module for establishing an accurate model of the flow channel of the pump-turbine based on the collected data and performing dynamic simulation, an intelligent monitoring fault diagnosis module for using artificial intelligence and machine learning technology to monitor the system operating status in real time and automatically identify and diagnose potential faults, and an optimization control feedback adjustment module for dynamically adjusting the operating parameters of the pump-turbine according to the analysis and monitoring results; The big data acquisition and processing module includes a sensor network management module for managing and coordinating sensors installed at key positions of pump turbines, a data synchronization timing correction module for processing data timestamps from different sensors to ensure data time synchronization, a data cleaning anomaly detection module for automatically identifying and filtering out noise and abnormal data generated during the acquisition process, a multi-source data fusion module for fusing data from different sources to form a comprehensive data set, and a data compression storage optimization module for using advanced data compression algorithms and storage optimization technologies to reduce storage space usage while ensuring fast data reading and efficient management; The fluid dynamics modeling and simulation module includes a three-dimensional flow channel geometry modeling module for constructing a detailed three-dimensional geometry model of the flow channel using CAD and reverse engineering technology, a multi-physics field coupling simulation module for integrating coupling simulation of multiple physical fields such as fluid mechanics, structural mechanics and thermodynamics to comprehensively analyze complex phenomena in the flow channel, a real-time simulation prediction module for performing dynamic simulation based on real-time data to predict possible problems in the flow channel, a simulation result optimization module for adjusting flow channel design parameters through optimization algorithms, and a virtual reality simulation display module for displaying simulation results in the form of virtual reality to help engineers intuitively understand flow behavior and support design decisions.

2. A pump-turbine flow passage analysis system according to claim 1, characterized in that: The intelligent monitoring fault diagnosis module includes a real-time monitoring dashboard module for providing an intuitive user interface, displaying key operating parameters and flow channel status in real time, and supporting customized monitoring views; a machine learning driven fault prediction module for predicting possible faults by training machine learning models, analyzing historical and real-time data, and taking preventive measures in advance; an intelligent alarm notification module for automatically triggering alarms based on preset thresholds and model predictions, and sending alarm information to relevant personnel through multiple channels; a fault diagnosis knowledge base module for building and maintaining a knowledge base containing various fault modes and their diagnostic information, supporting rapid fault identification and processing; and an adaptive maintenance scheduling module for intelligently generating maintenance plans and schedules based on fault predictions and system operating status; The optimization control feedback adjustment module includes an adaptive control algorithm module for adopting an adaptive control strategy to adjust parameters such as pump speed and valve opening according to real-time data to optimize the flow state in the flow channel, a multi-objective optimization decision module for comprehensively considering multiple objectives such as efficiency, energy consumption, stability, etc., and formulating the best operation strategy through a multi-objective optimization algorithm, a feedback loop management module for establishing a closed-loop feedback mechanism to continuously adjust the control strategy according to the system operation status and optimization results, an artificial intelligence assisted decision-making module for utilizing AI technologies such as deep learning and reinforcement learning to assist in decision-making, and a virtual assistant user interaction module for providing an intelligent virtual assistant that supports voice and text interaction to help operators understand the system status, adjust parameters and obtain suggestions.

3. A pump-turbine flow path analysis system according to claim 1, characterized in that: The sensor network management module is responsible for coordinating the data collection of each sensor and verifying the integrity and accuracy of the data, and then passing the data to the data synchronization timing correction module for timestamp correction to ensure the time synchronization of the data; after the data synchronization timing correction module performs timestamp correction on the sensor data, the corrected data is passed to the data cleaning anomaly detection module, which identifies and filters the noise and abnormal data generated during the collection process; after the data cleaning anomaly detection module cleans and detects anomalies on the data, the processed data is passed to the multi-source data fusion module, which fuses the data from different sources to form a comprehensive data set; the comprehensive data set generated by the multi-source data fusion module is passed to the data compression storage optimization module, which reduces the storage space occupancy through data compression algorithms and storage optimization technology, while ensuring fast reading and efficient management of data.

4. A pump-turbine flow passage analysis system according to claim 1, characterized in that: The three-dimensional flow channel geometry modeling module uses CAD and reverse engineering technology to construct a detailed three-dimensional geometry model of the flow channel, and then transmits the constructed geometry model to the multi-physics coupling simulation module, which performs coupling simulation on multiple physical fields such as fluid, structure and thermodynamics in the flow channel; After the multi-physics coupling simulation module performs coupling simulation on the multi-physics fields in the flow channel, the simulation data is transmitted to the real-time simulation prediction module, which performs dynamic simulation based on the real-time data to predict problems that may occur in the flow channel; After completing dynamic simulation and problem prediction, the real-time simulation prediction module passes the simulation results to the simulation result optimization module, which adjusts the flow channel design parameters through the optimization algorithm; the optimized simulation results generated by the simulation result optimization module are passed to the virtual reality simulation display module, which displays the simulation results in the form of virtual reality, helping engineers to intuitively understand the flow behavior and support design decisions.

5. A pump-turbine flow passage analysis system according to claim 2, characterized in that: The real-time monitoring dashboard module is responsible for providing an intuitive user interface, displaying key operating parameters and flow channel status in real time, and passing these data to the machine learning driven fault prediction module, which predicts possible faults by analyzing historical and real-time data; after analyzing historical and real-time data, the machine learning driven fault prediction module predicts possible faults and passes the prediction results to the intelligent alarm notification module, which automatically triggers alarms according to preset thresholds and prediction results, and sends alarm information to relevant personnel through multiple channels; after triggering the alarm, the intelligent alarm notification module passes the alarm information and fault data to the fault diagnosis knowledge base module, which maintains a knowledge base containing various fault modes and their diagnostic information to support rapid fault identification and processing; the fault modes and diagnostic information provided by the fault diagnosis knowledge base module are passed to the adaptive maintenance scheduling module, which intelligently generates maintenance plans and schedules based on fault predictions and system operating status.

6. A pump-turbine flow passage analysis system according to claim 2, characterized in that: The adaptive control algorithm module adjusts parameters such as pump speed and valve opening according to real-time data to optimize the flow state in the flow channel, and passes the adjustment results to the multi-objective optimization decision module, which comprehensively considers multiple objectives such as efficiency, energy consumption and stability to formulate the best operation strategy; the best operation strategy generated by the multi-objective optimization decision module is passed to the feedback loop management module, which establishes a closed-loop feedback mechanism to continuously adjust the control strategy according to the system operation status and optimization results; the system status and optimization results provided by the feedback loop management module are passed to the artificial intelligence auxiliary decision module, which uses deep learning and reinforcement learning technologies to assist decision-making and optimize the operation strategy; the auxiliary decision suggestions generated by the artificial intelligence auxiliary decision module are passed to the virtual assistant user interaction module, which helps operators understand the system status, adjust parameters and obtain optimization suggestions through voice or text interaction to support efficient operation and management of the system.

7. A pump-turbine flow passage analysis system according to any one of claims 1 to 6, wherein the analysis method is characterized in that: The following steps are involved: S1, flow field measurement data analysis; S2, multi-objective optimization analysis of flow channel numerical simulation; S3. Optimization of multiple development control strategies for intelligent monitoring systems.

8. A pump-turbine flow channel analysis method according to claim 7, characterized in that: In the step S1, flow field measurement data analysis: S1.

1. Install a variety of high-precision sensors at key locations of the pump-turbine, including pressure sensors, flow rate sensors, temperature sensors and vibration sensors, to form a sensor network. The sensor layout needs to be optimized according to the flow channel structure characteristics and working conditions to ensure that all key monitoring points are covered. At the same time, materials with strong anti-interference performance, ceramic pressure sensors and vibration sensors with nickel-based alloy shells are selected; S1.

2. Input the collected sensor signals into the data acquisition terminal and perform time stamp correction through time synchronization technology to ensure that all data have a consistent time reference. After synchronization, the signal is pre-processed using a Kalman filter to remove abnormal jitter and noise to ensure high signal accuracy. S1.

3. Through the data fusion algorithm, the data from different types of sensors are comprehensively processed to form a unified data set containing parameters such as pressure, flow rate, temperature and vibration. During the fusion process, the data is verified to eliminate abnormal deviations; S1.

4. Use sparse matrix compression technology and efficient storage framework to reduce data storage requirements to 10% to 20% of the original volume without losing key information. The storage optimization process is equipped with a high-speed data access interface.

9. A pump-turbine flow channel analysis method according to claim 7, characterized in that: In the step S2, multi-objective optimization analysis of flow channel numerical simulation: S2.

1. Use high-precision laser scanners to collect data on the inner surface of the pump-turbine flow channel, and use reverse engineering software to build a detailed three-dimensional geometric model. Perform unstructured meshing in CFD simulation software, focusing on encryption processing of the flow channel bends and blade surfaces to capture micro-flow phenomena; S2.

2. In the simulation platform, fluid mechanics, thermodynamics and structural mechanics are coupled for simulation to analyze the flow, heat exchange and structural stress distribution in the channel. Water and anti-corrosion fluid are used as the main working fluids, and the pipe wall material is wear-resistant stainless steel. Boundary parameters under actual operating conditions are applied in the simulation. S2.

3. Based on the above coupling analysis results, a dynamic model is established in combination with real-time collected data to simulate possible emergencies. The flow field changes within the next 5 to 10 seconds are dynamically predicted through the LSTM deep learning network to discover areas that may cause faults in advance. S2.

4. Use genetic algorithm to optimize the flow channel geometry design parameters, focusing on adjusting the blade installation angle, flow channel contraction ratio and guide vane position, with the goal of reducing the pressure loss by 15% to 20%. The optimization results are exported to the CAD model for revision and verification; S2.

5. Virtual reality technology is used to present simulation results as dynamic three-dimensional scenes. Engineers can observe fluid flow characteristics, pressure distribution and eddy areas through head-mounted displays, and interact with the model to adjust design decisions.

10. A pump-turbine flow channel analysis method according to claim 7, characterized in that: In the step S3, in the optimization of the multi-development control strategy of the intelligent monitoring system: S3.

1. Set up a multi-screen monitoring system in the central monitoring room, and use a custom dashboard to display key operating parameters such as pressure, temperature, and vibration in real time. The dashboard is connected to the sensor network based on industrial protocols and sets alarm thresholds; S3.

2. Build a machine learning model based on historical data and real-time data to predict the probability of failure. The model training data comes from long-term operation monitoring results and known fault case libraries. The fault prediction accuracy can reach more than 95%; S3.

3. Intelligent alarms are triggered by the prediction results of the comprehensive evaluation model and the set alarm thresholds. Alarm information is sent to engineers and maintenance teams via SMS, email and IoT devices to ensure rapid response to problems. S3.

4. Based on the predictive maintenance strategy, combined with the current operating status and historical maintenance records, an optimized maintenance plan is generated, and the maintenance cycle is dynamically adjusted through the scheduling algorithm to extend the overall life of the equipment and reduce downtime; S3.

5. Use deep reinforcement learning technology to adjust operating parameters in real time to achieve the best balance between efficiency and stability. The feedback control mechanism ensures that the system can self-adjust under dynamic conditions to support long-term efficient operation.

Images