Water pump control method and system based on end cloud intelligent pump room

Abstract

The invention discloses a water pump control method and system based on an end-side cloud smart pump room, and the method comprises the steps: tail end data collection: collecting water pump operation parameters, medium state parameters and environmental parameters in real time through multiple types of sensors disposed on a smart pump room system pipeline and equipment; the water pump operation parameters comprise water pump rotating speed, inlet and outlet pressure, motor current and torque; the medium state parameters comprise temperature, humidity, flow and pressure of a medium in the pipeline; the environment parameters comprise temperature, humidity and air conditioner load associated data in the pump room. According to the method, traditional control bottlenecks are broken through through an end-side cloud collaborative architecture, the control precision and energy efficiency of the water pump are remarkably improved, the problems of pressure and flow fluctuation of traditional constant value control are solved through the mode, and the energy utilization rate of a pump room can be increased by 15%-20% through actual measurement.

Description

Technical Field

[0001] This invention relates to the field of smart pumping station technology, and in particular to a pump control method and system for smart pumping stations based on edge cloud. Background Technology

[0002] In scenarios such as large university computer rooms, industrial production workshops, and commercial buildings, smart pump rooms, as a core component of centralized air conditioning control systems, bear the important responsibility of ensuring the stable operation of critical equipment such as refrigeration units, cooling towers, and fans. Their operating status directly affects the energy efficiency and environmental control accuracy of the entire air conditioning system. As the power core within the smart pump room, the control accuracy and coordination efficiency of the water pump are key factors determining whether the system can achieve constant pressure, temperature, humidity, and flow control targets.

[0003] Currently, traditional pump control systems in pump stations generally suffer from the following technical defects: First, the control architecture is relatively independent, often using a single PLC for local logic control, lacking deep collaboration with end-point sensor networks and cloud management platforms, resulting in a broken "end-edge-cloud" data link and the inability to achieve real-time interaction and intelligent decision-making across the entire link; Second, data processing capabilities are weak, as the massive heterogeneous data collected by end-point pressure sensors, temperature and humidity sensors, and flow meters needs to be uploaded to the cloud for processing, which not only increases network transmission pressure but also causes significant latency issues, making it difficult to meet the real-time requirements of pump control; Third, control strategies are rigid, mostly based on preset parameters for setpoint control, unable to adapt to dynamic environmental changes through big data models, leading to problems such as sudden pressure changes and unstable flow under certain operating conditions, affecting the air conditioning system's control effect and causing energy waste. Therefore, this application proposes a pump control method and system for smart pump stations based on end-edge-cloud. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a water pump control method and system based on edge-cloud smart pumping stations.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: The water pump control method based on edge-cloud smart pump room includes the following steps: S1. End-point data acquisition steps: Real-time acquisition of pump operating parameters, medium status parameters, and environmental parameters is achieved through various types of sensors deployed on the pipelines and equipment of the smart pump room system; the pump operating parameters include pump speed, inlet and outlet pressure, motor current and torque; the medium status parameters include the temperature, humidity, flow rate and pressure of the medium in the pipeline; the environmental parameters include temperature, humidity and air conditioning load related data in the pump room; S2. Data Preprocessing and Edge Transmission Steps: The edge control unit cleans, filters, standardizes, and compresses the collected multi-source heterogeneous data to remove abnormal data. Based on ModbusTCP and OPCUA protocols, a real-time communication link is established between the edge control unit and the end sensors and water pump controller. The preprocessed data is divided into real-time control data and non-real-time analysis data. The real-time control data is transmitted to the edge computing module through a millisecond-level transmission channel, while the non-real-time analysis data is uploaded to the cloud platform through a standard API interface. S3. Edge Real-Time Control Steps: The edge computing module calls the pre-loaded basic control model and calculates the target operating parameters of the water pump by combining real-time control data; the basic control model is a fusion model built based on PID algorithm and PLCopen motion control standard, which supports 32 real-time multi-task parallel processing and has a scan cycle as low as 250μs; the edge computing module sends the target operating parameters to the water pump controller through the EtherCAT / Profinet master station interface to control the water pump speed, start and stop status and valve opening, so as to achieve preliminary constant regulation of medium pressure and flow rate; S4. Cloud Model Optimization Steps: The cloud platform receives non-real-time analysis data, combines it with historical operating data to construct a big data training set, and uses cloud AI computing power to iteratively optimize the basic control model to generate an adaptive control model. The iterative optimization process includes feature extraction, operating condition clustering, and parameter tuning, so that the model can match the water pump control requirements under different air conditioning loads and ambient temperatures. The optimized adaptive control model is deployed to the edge computing module and updated through containerization. S5. Closed-loop dynamic adjustment steps: The edge computing module obtains the medium state parameters after the water pump is controlled through the real-time feedback channel and compares them with the preset constant control threshold. If the deviation value exceeds the allowable range, the adaptive control model is called to recalculate the target operating parameters and execute the control operation, forming a closed-loop adjustment mechanism of "acquisition-analysis-control-feedback".

[0006] As a further aspect of the present invention, in the terminal data acquisition step, multiple types of sensors are hot-swappable and support dynamic configuration of data acquisition frequency; wherein, the acquisition frequency of pressure sensor and flow sensor is not less than 100Hz, the acquisition frequency of temperature and humidity sensor is 10-50Hz, and the acquisition frequency of water pump operating parameters is synchronized with water pump control cycle.

[0007] As a further aspect of the present invention, in the edge real-time control step, the edge computing module has a built-in RT-Linux operating system and adopts a multi-core CPU and NPU collaborative architecture; wherein, the CPU core is used to execute real-time control tasks, and the NPU core is used to call the adaptive control model to perform inference calculations, so as to realize the millisecond-level issuance of control strategies.

[0008] As a further aspect of the present invention, in the cloud model optimization step, the historical operating data includes at least 3 months of water pump operating parameters, medium state parameters, and corresponding air conditioning load data; the big data training set is stored in XML standard format, supports data interaction with MES and ERP systems, and realizes cross-platform data sharing.

[0009] A water pump control system based on edge-cloud smart pump station includes: S1. End-point sensing layer: includes pressure sensor, temperature and humidity sensor, flow meter and motor status sensor. All sensors are equipped with standard industrial communication interfaces for collecting multi-source data on pump operation and surrounding environment. S2. Edge Control Layer: This layer includes an edge control unit, an edge computing module, and a water pump controller. The edge control unit integrates an EtherCAT / Profinet master station, supporting the parsing of over a hundred industrial control protocols. The edge computing module is equipped with the MatriXControl programming platform and the MatriXFactory data engine, supporting IEC61131-3 standard programming language and C / C++ algorithm development. The water pump controller is connected to the edge computing module via a real-time bus, receiving control commands and driving the water pump to operate. S3. Cloud Management Layer: Includes a cloud server, an AI model training module, and a data storage module; the cloud server provides a web-based management interface, supporting device status monitoring and control parameter configuration; the AI ​​model training module is built based on 6 TOPS-level NPU computing power and is used to achieve iterative optimization of the control model; the data storage module uses a distributed database to store historical running data and optimized model files; S4. Communication Transmission Layer: Includes industrial Ethernet, 4G / WIFI wireless communication module and VPN encrypted channel; the industrial Ethernet is used for real-time data transmission between the edge control layer and the end perception layer, the wireless communication module is used for cloud uploading of non-real-time data, and the VPN encrypted channel ensures the security of cross-network data transmission.

[0010] As a further aspect of the present invention, the edge computing module of the edge control layer supports containerized deployment and can simultaneously load the basic control model, the adaptive control model, and the predictive maintenance model; wherein, the predictive maintenance model achieves early warning of water pump failure by analyzing abnormal fluctuations in the current and torque of the water pump motor.

[0011] The end-sensing layer also includes a visualization HMI configuration unit, which is connected to the edge computing module to display the pump's operating status, medium parameters, and control strategy execution in real time, and supports the input of manual intervention control commands.

[0012] As a further aspect of the present invention, the cloud management layer also includes a permission management module, which assigns different operation permissions based on roles to achieve hierarchical control over operations such as control parameter modification, model update, and fault handling.

[0013] As a further embodiment of the present invention, the edge control unit adopts the MX6300Pro edge intelligent controller, which is equipped with an 8-core ARM processor and 6 TOPSNPU, supports 32-axis synchronous control, and can connect up to 128 slave devices to meet the multi-pump collaborative control requirements of large-scale intelligent pumping stations.

[0014] As a further aspect of the present invention, the system also includes a redundant backup unit, which is deployed in parallel with the edge control module. When the main control module fails, the redundant unit can take over the control task within 50ms, ensuring the continuity of system operation.

[0015] The beneficial effects of this invention are as follows: This invention breaks through traditional control bottlenecks with an "edge-cloud" collaborative architecture, significantly improving the control accuracy and energy efficiency of water pumps. The end-point sensing layer utilizes high-precision sensors with 100Hz high-frequency acquisition and a hot-swappable design to ensure comprehensive and real-time data. The edge layer's MX6300Pro controller achieves millisecond-level command response with a 250μs scan cycle. Cloud-based AI computing power generates adaptive models through operating condition clustering, dynamically matching load and environmental changes. This model solves the pressure and flow fluctuation problems of traditional setpoint control, and actual measurements show it can improve pump station energy utilization by 15%-20%.

[0016] This invention: The system leverages multiple redundancies and intelligent operation and maintenance to improve reliability and reduce costs. On the hardware side, edge-layer redundant units can take over faulty modules within 50ms, and encrypted communication ensures continuous control loop. On the software side, a predictive maintenance model monitors motor parameters to provide early warnings of faults, reducing unplanned downtime by more than 80%. End-point HMIs and cloud-based web interfaces enable end-to-end visual monitoring, and with access control, operational efficiency is improved by 30%, effectively reducing labor costs. Attached Figure Description

[0017] Figure 1 This is a flowchart of the water pump control method for a smart pumping station based on edge cloud proposed in this invention; Figure 2 This is a flowchart of the water pump control system architecture for an edge-cloud-based smart pumping station proposed in this invention. Figure 3 This is a flowchart of the closed-loop dynamic adjustment mechanism proposed in this invention. Detailed Implementation

[0018] 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.

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] Reference Figure 1-3 The water pump control method based on edge-cloud smart pump room includes the following steps: S1. End-point data acquisition steps: Real-time acquisition of pump operating parameters, medium status parameters, and environmental parameters is achieved through various types of sensors deployed on the pipelines and equipment of the smart pump room system. Pump operating parameters include pump speed, inlet and outlet pressure, motor current and torque; medium status parameters include temperature, humidity, flow rate and pressure of the medium in the pipeline; environmental parameters include temperature, humidity and air conditioning load related data in the pump room. S2. Data Preprocessing and Edge Transmission Steps: The edge control unit cleans, filters, standardizes, and compresses the collected multi-source heterogeneous data to remove abnormal data. Based on ModbusTCP and OPCUA protocols, a real-time communication link is established between the edge control unit and the end sensors and water pump controller. The preprocessed data is divided into real-time control data and non-real-time analysis data. The real-time control data is transmitted to the edge computing module through a millisecond-level transmission channel, while the non-real-time analysis data is uploaded to the cloud platform through a standard API interface. S3. Edge Real-Time Control Steps: The edge computing module calls the pre-loaded basic control model and calculates the target operating parameters of the water pump by combining real-time control data. The basic control model is a fusion model built on PID algorithm and PLCopen motion control standard, which supports 32 real-time multi-task parallel processing with a scan cycle as low as 250μs. The edge computing module sends the target operating parameters to the water pump controller through the EtherCAT / Profinet master station interface to control the water pump speed, start / stop status and valve opening, so as to achieve preliminary constant regulation of medium pressure and flow. S4. Cloud Model Optimization Steps: The cloud platform receives non-real-time analysis data, combines it with historical operating data to build a big data training set, and uses cloud AI computing power to iteratively optimize the basic control model to generate an adaptive control model. The iterative optimization process includes feature extraction, operating condition clustering, and parameter tuning, so that the model can match the water pump control requirements under different air conditioning loads and ambient temperatures. The optimized adaptive control model is deployed to the edge computing module and updated through containerization. S5. Closed-loop dynamic adjustment steps: The edge computing module obtains the medium state parameters after the water pump is controlled through the real-time feedback channel and compares them with the preset constant control threshold. If the deviation value exceeds the allowable range, the adaptive control model is called to recalculate the target operating parameters and execute the control operation, forming a closed-loop adjustment mechanism of "acquisition-analysis-control-feedback".

[0021] In this embodiment, during the terminal data acquisition step, multiple types of sensors are hot-swappable and support dynamic configuration of data acquisition frequency; among them, the acquisition frequency of pressure sensor and flow sensor is not less than 100Hz, the acquisition frequency of temperature and humidity sensor is 10-50Hz, and the acquisition frequency of water pump operating parameters is synchronized with the water pump control cycle.

[0022] In this embodiment, in the edge real-time control step, the edge computing module has a built-in RT-Linux operating system and adopts a multi-core CPU and NPU collaborative architecture. The CPU core is used to execute real-time control tasks, and the NPU core is used to call the adaptive control model to perform inference calculations, so as to realize the millisecond-level issuance of control strategies.

[0023] In this embodiment, the historical operating data in the cloud model optimization step includes at least 3 months of pump operating parameters, medium status parameters and corresponding air conditioning load data; the big data training set is stored in XML standard format, which supports data interaction with MES and ERP systems, realizes cross-platform data sharing, and the adoption of XML standard format breaks down the data barriers between different systems, laying the foundation for the integration of smart pump rooms with other enterprise management systems.

[0024] A water pump control system based on edge-cloud smart pump station includes: S1. End-point sensing layer: includes pressure sensor, temperature and humidity sensor, flow meter and motor status sensor. All sensors are equipped with standard industrial communication interfaces to collect multi-source data on pump operation and the surrounding environment. S2. Edge Control Layer: This layer includes an edge control unit, an edge computing module, and a water pump controller. The edge control unit integrates an EtherCAT / Profinet master station, supporting the parsing of over a hundred industrial control protocols. The edge computing module is equipped with the MatriXControl programming platform and the MatriXFactory data engine, supporting IEC61131-3 standard programming language and C / C++ algorithm development. The water pump controller is connected to the edge computing module via a real-time bus, receiving control commands and driving the water pump. S3. Cloud Management Layer: Includes cloud server, AI model training module and data storage module; cloud server provides web management interface, supports device status monitoring and control parameter configuration; AI model training module is built based on 6TOPS NPU computing power, used to realize iterative optimization of control model; data storage module adopts distributed database to store historical running data and optimized model files; S4. Communication Transmission Layer: Includes industrial Ethernet, 4G / WIFI wireless communication module and VPN encrypted channel; industrial Ethernet is used for real-time data transmission between edge control layer and end perception layer, wireless communication module is used for cloud upload of non-real-time data, and VPN encrypted channel ensures the security of cross-network data transmission.

[0025] In this embodiment, the edge computing module of the edge control layer supports containerized deployment and can simultaneously load the basic control model, the adaptive control model, and the predictive maintenance model. The predictive maintenance model analyzes abnormal fluctuations in the pump motor current and torque to provide early warnings of pump failures. The introduction of predictive maintenance transforms the traditional "post-event maintenance" into "pre-event warning," greatly reducing the impact of equipment failures on production and extending the service life of the equipment.

[0026] In this embodiment, the end-sensing layer also includes a visualization HMI configuration unit. The HMI configuration unit is connected to the edge computing module and is used to display the pump operating status, medium parameters and control strategy execution status in real time, and supports the input of manual intervention control commands.

[0027] In this embodiment, the cloud management layer also includes a permission management module. The permission management module assigns different operation permissions based on roles to achieve hierarchical control over operations such as control parameter modification, model update, and fault handling.

[0028] In this embodiment, the edge control unit adopts the MX6300Pro edge intelligent controller. The controller is equipped with an 8-core ARM processor and 6 TOPS NPUs, supporting 32-axis synchronous control and connecting up to 128 slave devices. This meets the multi-pump collaborative control requirements of large-scale intelligent pumping stations. The powerful hardware configuration provides ample computing and connectivity capabilities for multi-pump collaborative control, ensuring the coordination and synchronization of operation among various devices within the large pumping station. The controller's high scalability also facilitates future equipment expansion within the pumping station.

[0029] In this embodiment, the system also includes a redundant backup unit, which is deployed in parallel with the edge control module. When the main control module fails, the redundant unit can take over the control tasks within 50ms, ensuring the continuity of system operation. The millisecond-level fault switching speed minimizes the impact of system failures on pump operation, meeting the needs of industrial scenarios with high requirements for operational continuity. Redundancy backup design is a key means to improve system reliability, providing dual protection for the stable operation of the smart pump station.

[0030] Working Principle: This system relies on a four-layer architecture of "end-point sensing - edge control - cloud management - communication transmission" to achieve precise and efficient control of the water pump through end-edge-cloud collaboration. The end-point sensing layer uses hot-swappable sensors to collect multi-dimensional data in real time. Pressure and flow sensors capture medium parameters at a frequency of at least 100Hz, while temperature and humidity sensors and motor status sensors simultaneously collect environmental and equipment operation data to ensure comprehensive information. The data is preprocessed by the MX6300Pro controller in the edge control layer, cleaned, filtered, and then classified before transmission: real-time control data is directly transmitted to the edge computing module via a millisecond-level channel, while non-real-time data is encrypted and uploaded to the cloud. The edge computing module, supported by the RT-Linux system, utilizes a multi-core CPU and NPU working in tandem to quickly calculate control parameters using a PID fusion model. These parameters are then sent to the water pump controller within 250μs via the EtherCAT bus, achieving initial pressure and flow stability. The cloud management layer utilizes a 6TOPS-level NPU computing power, combined with over 3 months of historical data to train a model. Through operating condition clustering and parameter optimization, an adaptive model is generated and then updated to the edge in a containerized manner. The system constructs a closed-loop mechanism, with edge modules comparing control parameters and thresholds in real time. When deviations exceed limits, the optimization model is invoked for dynamic adjustment. The communication transmission layer ensures real-time communication through industrial Ethernet, and the VPN encrypted channel ensures data security. The redundant backup unit takes over the faulty equipment within 50ms. Combined with the visualization monitoring and manual intervention functions of the HMI configuration unit, a control system that combines reliability and flexibility is formed.

Claims

1. A water pump control method based on edge-cloud smart pumping station, characterized in that, Includes the following steps: S1. End-point data acquisition steps: Real-time acquisition of pump operating parameters, medium status parameters, and environmental parameters is achieved through various types of sensors deployed on the pipelines and equipment of the smart pump room system; the pump operating parameters include pump speed, inlet and outlet pressure, motor current and torque; the medium status parameters include the temperature, humidity, flow rate and pressure of the medium in the pipeline; the environmental parameters include temperature, humidity and air conditioning load related data in the pump room; S2. Data preprocessing and edge transmission steps: The edge control unit cleans, filters, standardizes, and compresses the collected multi-source heterogeneous data to remove abnormal data. Based on ModbusTCP and OPCUA protocols, a real-time communication link is established between the edge control unit and the end sensors and water pump controller. The pre-processed data is divided into real-time control data and non-real-time analysis data. The real-time control data is transmitted to the edge computing module through a millisecond-level transmission channel, while the non-real-time analysis data is uploaded to the cloud platform through a standard API interface. S3. Edge Real-Time Control Steps: The edge computing module calls the pre-loaded basic control model and calculates the target operating parameters of the water pump by combining real-time control data; the basic control model is a fusion model built based on PID algorithm and PLCopen motion control standard, which supports 32 real-time multi-task parallel processing and has a scan cycle as low as 250μs; the edge computing module sends the target operating parameters to the water pump controller through the EtherCAT / Profinet master station interface to control the water pump speed, start and stop status and valve opening, so as to achieve preliminary constant regulation of medium pressure and flow rate; S4. Cloud Model Optimization Steps: The cloud platform receives non-real-time analysis data, combines it with historical operating data to construct a big data training set, and uses cloud AI computing power to iteratively optimize the basic control model to generate an adaptive control model. The iterative optimization process includes feature extraction, operating condition clustering, and parameter tuning, so that the model can match the water pump control requirements under different air conditioning loads and ambient temperatures. The optimized adaptive control model is deployed to the edge computing module and updated through containerization. S5. Closed-loop dynamic adjustment steps: The edge computing module obtains the medium state parameters after the water pump is controlled through the real-time feedback channel and compares them with the preset constant control threshold. If the deviation value exceeds the allowable range, the adaptive control model is called to recalculate the target operating parameters and execute the control operation, forming a closed-loop adjustment mechanism of "acquisition-analysis-control-feedback".

2. The method according to claim 1, characterized in that, In the terminal data acquisition step, multiple types of sensors are hot-swappable and support dynamic configuration of data acquisition frequency; among them, the acquisition frequency of pressure sensor and flow sensor is not less than 100Hz, the acquisition frequency of temperature and humidity sensor is 10-50Hz, and the acquisition frequency of water pump operating parameters is synchronized with water pump control cycle.

3. The method according to claim 1, characterized in that, In the edge real-time control step, the edge computing module has a built-in RT-Linux operating system and adopts a multi-core CPU and NPU collaborative architecture. The CPU core is used to execute real-time control tasks, and the NPU core is used to call the adaptive control model to perform inference calculations, so as to realize the millisecond-level distribution of control strategies.

4. The method according to claim 1, characterized in that, In the cloud-based model optimization step, historical operating data includes at least three months of pump operating parameters, medium status parameters, and corresponding air conditioning load data; the big data training set is stored in XML standard format, supporting data interaction with MES and ERP systems to achieve cross-platform data sharing.

5. A water pump control system for an edge-cloud-based smart pumping station, characterized in that, include: S1. End-point sensing layer: includes pressure sensor, temperature and humidity sensor, flow meter and motor status sensor. All sensors are equipped with standard industrial communication interfaces for collecting multi-source data on pump operation and surrounding environment. S2. Edge Control Layer: This layer includes an edge control unit, an edge computing module, and a water pump controller. The edge control unit integrates an EtherCAT / Profinet master station, supporting the parsing of over a hundred industrial control protocols. The edge computing module is equipped with the MatriXControl programming platform and the MatriXFactory data engine, supporting IEC61131-3 standard programming language and C / C++ algorithm development. The water pump controller is connected to the edge computing module via a real-time bus, receiving control commands and driving the water pump to operate. S3. Cloud Management Layer: Includes a cloud server, an AI model training module, and a data storage module; the cloud server provides a web-based management interface, supporting device status monitoring and control parameter configuration; the AI ​​model training module is built based on 6 TOPS-level NPU computing power and is used to achieve iterative optimization of the control model; the data storage module uses a distributed database to store historical running data and optimized model files; S4. Communication Transmission Layer: Includes industrial Ethernet, 4G / WIFI wireless communication module and VPN encrypted channel; the industrial Ethernet is used for real-time data transmission between the edge control layer and the end perception layer, the wireless communication module is used for cloud uploading of non-real-time data, and the VPN encrypted channel ensures the security of cross-network data transmission.

6. The system according to claim 5, characterized in that, The edge computing module of the edge control layer supports containerized deployment and can simultaneously load the basic control model, the adaptive control model, and the predictive maintenance model. The predictive maintenance model analyzes abnormal fluctuations in the pump motor current and torque to provide early warning of pump failures.

7. The system according to claim 5, characterized in that, The end-sensing layer also includes a visualization HMI configuration unit, which is connected to the edge computing module to display the pump's operating status, medium parameters, and control strategy execution in real time, and supports the input of manual intervention control commands.

8. The system according to claim 5, characterized in that, The cloud management layer also includes a permission management module, which assigns different operation permissions based on roles to achieve hierarchical control over operations such as control parameter modification, model update, and fault handling.

9. The system according to claim 5, characterized in that, The edge control unit adopts the MX6300Pro edge intelligent controller, which is equipped with an 8-core ARM processor and 6 TOPSNPU, supports 32-axis synchronous control, and can connect up to 128 slave devices to meet the multi-pump collaborative control requirements of large-scale intelligent pumping stations.

10. The system according to claim 5, characterized in that, The system also includes a redundant backup unit, which is deployed in parallel with the edge control module. When the main control module fails, the redundant unit can take over the control task within 50ms, ensuring the continuity of system operation.

Images