Intelligent production control system for permanent magnet shielding pump
By introducing industrial control robots and a hierarchical distributed architecture, the problems of insufficient intelligent decision-making, lagging fault detection, and inadequate data utilization in the permanent magnet shielded pump control system have been solved, realizing intelligent operation and maintenance and efficient operation of the system.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUIMAO ELECTRONIC COMPONENT KUNSHAN CO LTD
- Filing Date
- 2026-04-27
- Publication Date
- 2026-06-05
AI Technical Summary
Existing permanent magnet shielded pump control systems have shortcomings in terms of limited intelligent decision-making capabilities, lagging fault detection, lack of data value mining, strong network dependence, and one-way human-machine interaction, resulting in low system operating efficiency, delayed fault response, insufficient data utilization, and reliance on manual experience for operation and maintenance.
It adopts a hierarchical distributed architecture, introduces industrial control robots as the processing layer, and combines multi-pump linkage and redundant control logic, fault diagnosis and protection logic, data recording and analysis modules, and self-learning and continuous evolution modules to achieve intelligent decision-making, early fault identification, in-depth data mining, and two-way human-machine interaction.
It has improved the system's intelligent decision-making level, realized early warning of faults and in-depth data value mining, reduced network dependence risk, built an efficient human-computer interaction mode, and significantly improved the operating efficiency and reliability of pump sets.
Smart Images
Figure CN122148571A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of shielded pump control technology, and in particular to an intelligent production control system for a permanent magnet shielded pump. Background Technology
[0002] Permanent magnet shielded pumps are leak-free pumps that completely isolate the motor stator and rotor through a shielding sleeve. They are widely used in nuclear industry, petrochemical industry, refrigeration and air conditioning, aerospace and other fields with extremely high requirements for sealing and reliability. With the improvement of industrial automation, pump control systems have evolved from manual operation to automated control based on PLCs and frequency converters. In recent years, the application of Internet of Things (IoT) technology has made remote monitoring possible, and some high-end systems have begun to introduce data acquisition and centralized monitoring functions.
[0003] The existing system architecture is as follows: Field equipment layer: This includes sensors (pressure, temperature, flow, vibration, etc.) and actuators (electric valves, frequency converters) installed on each permanent magnet shielded pump. The sensors collect the pump's operating parameters and transmit them to the centralized control layer via fieldbus (such as Modbus RTU).
[0004] Centralized Control Layer: An industrial control computer (or high-performance PLC, such as Siemens S7-1500) serves as the main controller, communicating with the frequency converters of each pump via Ethernet. The main controller incorporates a PID control algorithm and multi-pump rotation logic, adjusting the speed of each pump based on the main pipe pressure setpoint, and implementing simple timed rotation and fault switching (e.g., automatically starting the standby pump when the currently running pump fails). The main controller is also responsible for uploading data to the remote monitoring layer.
[0005] Remote monitoring layer: The computer-based monitoring software (such as WinCC configuration software) installed in the central control room is connected to the main controller via Ethernet to display real-time data, historical trends and alarm records. Operators can modify the setting parameters or manually start and stop the pump group through the software.
[0006] Although the current solution has evolved to the stage of automated control, there is still considerable room for improvement in areas such as multi-pump collaborative control and predictive maintenance.
[0007] 1. Limited intelligent decision-making capabilities: The centralized control layer (PLC or industrial computer) has limited computing power and can only execute preset simple logic (such as PID, timed rotation), and cannot run complex optimization algorithms (such as efficiency optimization based on performance curves, multi-variable constraint solving), which means that multi-pump collaboration can only remain at the rule level and cannot achieve true global optimization.
[0008] 2. Delayed fault detection: Fault detection only uses threshold comparison (such as alarm when the temperature exceeds 80℃), and cannot perform spectrum analysis and pattern recognition on high-frequency signals such as vibration and current. It is difficult to detect fault signs such as early bearing wear and the beginning of rotor imbalance. The alarm is often triggered only when the fault has caused obvious damage.
[0009] 3. Its data value has not been explored: The data is only used for real-time display and simple recording, without in-depth mining, and it is impossible to identify energy consumption optimization space or predict the remaining life of equipment from historical data.
[0010] 4. Strong network dependence: The centralized controller and field devices are connected by a wired bus. Once the controller fails or the network is interrupted, the entire system may be paralyzed.
[0011] 5. One-way human-computer interaction: Remote monitoring can only display and manually intervene in one direction, lacking intelligent decision-making assistance. Operators rely on personal experience to handle alarm information. Summary of the Invention
[0012] The purpose of this application is to provide an intelligent production control system for permanent magnet shielded pumps, which can solve the technical problems of existing permanent magnet shielded pump control systems, such as limited intelligent decision-making capabilities, lagging fault detection, lack of data value mining, strong network dependence, and one-way human-machine interaction.
[0013] To achieve the above objectives, this application provides the following solution: This application provides an intelligent production control system for permanent magnet shielded pumps, which includes: an equipment layer, a control layer, a processing layer, and a monitoring and management layer.
[0014] The equipment layer includes: a permanent magnet shielded pump unit, sensors and actuators installed on the pump body and pipelines; the sensors are used to collect the operating parameters of the permanent magnet shielded pump equipment in real time and transmit them to the control layer; the actuators are used to receive the first control command from the control layer and execute the corresponding actions; the sensors include: a thermocurrent device, a thermocouple temperature sensor, a direct application temperature sensor, a containment pressure sensor, a vibration sensor, a shaft displacement sensor, a power sensor, an inlet pressure sensor, an outlet pressure sensor, and a flow meter. The actuators include: an inlet electric valve, an outlet electric valve, a frequency converter, and an alarm device; the frequency converter and the alarm device are located in the field control cabinet.
[0015] The control layer consists of a field control cabinet, which has a built-in PLC. The PLC is used to receive the operating parameters, preprocess the operating parameters and upload them to the processing layer; it is also used to receive the second control command from the processing layer, parse it and generate a first control command to send to the actuator.
[0016] The processing layer, which is an industrial control robot, is used to receive pre-processed parameters, generate second control commands for intelligent control of the permanent magnet shielded pump unit through intelligent analysis and decision-making, and send them to the control layer; it is also used to send operating parameters, alarm information, diagnostic results and report data to the monitoring management layer.
[0017] The monitoring management layer is used to receive the operating parameters, alarm information, diagnostic results and report data, and to perform visualization and push reminders; it is also used to receive manual operation instructions and send them to the processing layer to realize remote monitoring and remote intervention.
[0018] Optionally, the industrial control robot includes: a multi-pump linkage and redundancy control logic module, a fault diagnosis and protection logic module, a data recording and analysis module, a self-learning and continuous evolution module, and a two-way communication interface with the monitoring and management layer.
[0019] The multi-pump linkage and redundancy control logic module is used to dynamically allocate the output of the permanent magnet shielded pump unit according to the real-time system load, the performance curves of each permanent magnet shielded pump and its health status, so as to achieve optimal energy efficiency operation, and to perform redundancy switching of start-up and shutdown when equipment is abnormal, so as to ensure continuous and stable operation of the system.
[0020] The fault diagnosis and protection logic module is used to extract fault features from operating parameters, identify fault types and fault probabilities through a deep learning model, output graded protection instructions and alarm information, and establish a fault knowledge base for diagnosis and learning.
[0021] The data recording and analysis module is used to store historical operating data in time series, statistically analyze energy consumption, efficiency, and runtime, predict the remaining lifespan of the equipment, and automatically generate operation and maintenance reports and energy efficiency analysis reports.
[0022] The self-learning and continuous evolution module is used to perform online incremental training of the control algorithm and diagnostic model based on new fault cases and operating parameters, and to retain historical experience by using an elastic weight consolidation mechanism to achieve continuous system optimization.
[0023] The bidirectional communication interface with the monitoring management layer is used to push operating parameters, alarm information, diagnostic results and report data to the monitoring management layer, and to receive manual control commands, parameter settings and query requests, and to prioritize the execution of manual control commands.
[0024] Optionally, the multi-pump linkage and redundancy control logic module includes: The load sensing submodule is used to calculate the total system load demand based on real-time pressure, flow rate, and set target values.
[0025] The efficiency optimization submodule is used to combine the performance curves of the permanent magnet shielded pumps with the health status of each permanent magnet shielded pump to solve for the optimal pump combination and speed distribution, so as to achieve the optimal system energy efficiency.
[0026] The redundancy switching submodule is used to perform a smooth start-stop switching when the permanent magnet shielded pump unit fails, ensuring continuous system operation.
[0027] Optionally, the fault diagnosis and protection logic module includes: The feature extraction submodule is used to extract fault feature vectors from the operating parameters.
[0028] The deep learning submodule is used to identify the fault feature vector using a CNN or LSTM model and output the fault type and fault probability.
[0029] The protection decision submodule is used to output graded protection commands and alarm information based on the fault type and the fault probability.
[0030] The fault knowledge base stores fault characteristics, diagnostic results, and processing records, providing data support for diagnosis and learning.
[0031] Optionally, the data recording and analysis module includes: The historical data storage submodule is used to store operating parameters, alarm information, operation logs, and historical data of equipment status using a time-series database.
[0032] The statistical analysis submodule is used to perform statistical calculations on runtime, energy consumption, efficiency, and reliability, and to predict the remaining service life based on the equipment degradation trend.
[0033] The report generation submodule is used to automatically generate daily, weekly, monthly, and annual reports as well as energy efficiency analysis reports.
[0034] Optionally, the self-learning and continuous evolution module includes: The online model update submodule is used to periodically trigger model and algorithm updates based on newly added fault cases and operating parameters.
[0035] The incremental training submodule is used to perform online incremental training on the multi-pump linkage and redundancy control logic module and the fault diagnosis and protection logic module.
[0036] Optionally, the bidirectional communication interface with the monitoring management layer includes: The data delivery submodule is used to push operating parameters, alarm information, diagnostic results, and report data to the monitoring management layer.
[0037] The instruction receiving submodule is used to receive manual control instructions, parameter settings and query requests issued by the monitoring management layer, and to execute the manual control instructions first.
[0038] Optionally, the intelligent production control system for the permanent magnet shielded pump also includes an AI model.
[0039] The AI model is connected to the device layer, the control layer, the processing layer, and the monitoring and management layer, respectively, to provide AI functions.
[0040] Optionally, the control layer and the processing layer communicate via industrial Wi-Fi. Each control cabinet has an independent IP address, supporting separate addressing by the processing layer. When the network is interrupted, the control cabinet can independently execute local shutdown and protection logic.
[0041] Optionally, the intelligent production control system for the permanent magnet shielded pump can calculate the pump head, motor output power, pump efficiency, shaft power, and power consumption of the permanent magnet shielded pump.
[0042] The formula for calculating the head of the permanent magnet shielded pump is as follows: ; ; ; Where H is the head of the permanent magnet shielded pump, in meters; The outlet pressure of the permanent magnet shielded pump is expressed in kPa. The inlet pressure of the permanent magnet shielded pump is kPa. The density of the medium is kg / m³. 3 ; The height difference between the inlet pressure sensor positions, in meters (m). v1 is the height difference between the outlet pressure sensor positions, in meters; v2 is the inlet velocity, in meters per second; v3 is the outlet velocity, in meters per second; g is the acceleration due to gravity, in meters per second. 2 Q represents flow rate, m 3 / h; D1 is the inlet diameter, mm; D2 is the outlet diameter, mm.
[0043] The formula for calculating the output power of the motor is: ; in, The output power of the motor is expressed in KW. This represents the input power of the motor, in kilowatts (kW). For motor efficiency.
[0044] The formula for calculating the efficiency of the permanent magnet shielded pump is as follows: ; in, The efficiency of the permanent magnet shielded pump.
[0045] The formula for calculating the shaft power is: ; in, Shaft power, KW; This refers to the specific gravity of the medium.
[0046] The formula for calculating the power consumption of the permanent magnet shielded pump is as follows: ; Where E is the power consumption of the permanent magnet shielded pump, in kWh; and t is the time, in hours.
[0047] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides an intelligent production control system for permanent magnet shielded pumps, featuring a four-layer architecture: equipment layer, control layer, processing layer, and monitoring and management layer. The equipment layer utilizes sensors to collect real-time operating parameters of the pump and pipelines, and actuators receive commands to execute actions. This enables accurate acquisition of equipment operating status and rapid response to control demands, ensuring real-time data acquisition and reliable execution. The control layer employs a field control cabinet with a built-in PLC, which preprocesses operating parameters and performs command parsing and forwarding, reducing the processing pressure on upper layers and enabling local basic data processing and command relay, thus improving system response efficiency and stability. The processing layer uses industrial control robots for intelligent analysis and decision-making, leveraging powerful computing capabilities to implement complex algorithms and intelligent judgments. This overcomes the computing power limitations of traditional controllers, enabling multi-pump collaborative optimization, early fault identification, and in-depth data mining, shifting from passive control to proactive intelligent regulation. The monitoring and management layer provides visualized display and push notifications of operating data, alarm information, diagnostic results, and reports. It also supports remote human intervention, constructing a two-way human-machine interaction mode, reducing reliance on operator experience and improving operational convenience and management efficiency. This system effectively improves the intelligent decision-making level of the permanent magnet shielded pump control system, realizes early warning of faults and in-depth data value mining, reduces the risk of dependence on system network and centralized controller, builds two-way efficient human-machine interaction, and significantly improves the pump group's operating efficiency, reliability and intelligent operation and maintenance level. Attached Figure Description
[0048] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0049] Figure 1 A schematic diagram of an intelligent production control system for a permanent magnet shielded pump provided in one embodiment of this application; Figure 2 A schematic diagram of the overall framework of an intelligent production control system for a permanent magnet shielded pump provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an industrial control robot provided in one embodiment of this application; Figure 4 A schematic diagram of a hierarchical AI capability architecture provided in an embodiment of this application; Figure 5 This is a schematic diagram of a fault diagnosis AI process provided in an embodiment of this application; Figure 6 A schematic diagram illustrating the specific workflow of the fault diagnosis and protection logic module AI provided in an embodiment of this application; Figure 7 A schematic diagram illustrating the specific workflow of the multi-pump linkage and redundancy control module AI provided in an embodiment of this application; Figure 8 This is a schematic diagram illustrating the specific workflow of the self-learning and continuous evolution module AI provided in one embodiment of this application.
[0050] Figure label: 1. Thermocurrent device; 2. Thermocouple temperature sensor; 3. Directly applied temperature sensor; 4. Containment pressure sensor; 5. Vibration sensor; 6. Shaft displacement sensor; 7. Power sensor; 8. Inlet electric valve; 9. Inlet pressure sensor; 10. Outlet pressure sensor; 11. Flow meter; 12. Outlet electric valve; 13. Control cabinet. Detailed Implementation
[0051] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.
[0052] This application proposes an intelligent production control system for permanent magnet shielded pumps. Through technologies such as intelligent robots, the Internet of Things, and automated control, it achieves super-intelligence: human-machine collaboration, intelligent decision-making, data-driven management, and interconnectivity. It solves a series of problems caused by the primary technical issue of traditional shielded pumps—"strong dependence and lagging information perception"—and changes the traditional management model of shielded pumps.
[0053] The existing permanent magnet shielded pump management solution has limited computing power in the centralized control layer (PLC or industrial computer), which can only execute simple preset logic (such as PID regulation and timed rotation), and cannot perform complex real-time analysis and optimization decisions, resulting in low overall system operating efficiency and delayed fault response.
[0054] Due to the limited computing power of PLCs, they cannot run complex optimization algorithms (such as efficiency optimization based on performance curves and multivariable constraint solving). Multi-pump control can only use simple PID regulation or fixed rule rotation, and cannot dynamically allocate the output of each pump according to the real-time load. This results in some pumps operating in the inefficient zone for a long time, high total system energy consumption, and uneven wear of each pump.
[0055] Due to limited computing power, fault detection can only use threshold comparison (such as alarms when the temperature exceeds 80°C). It cannot perform spectrum analysis and pattern recognition on high-frequency signals such as vibration and current, and cannot detect fault signs such as early bearing wear and the beginning of rotor imbalance. Often, alarms are only triggered when the fault has already caused obvious damage, resulting in frequent unplanned downtime, high maintenance costs, and shortened equipment life.
[0056] Due to the lack of a central intelligent decision-making unit, the data is only used for real-time display and simple recording, without in-depth analysis. It is impossible to identify energy consumption optimization space from historical data, predict the remaining lifespan of equipment, and achieve predictive maintenance and continuous optimization.
[0057] The remote monitoring layer can only display information in one direction and allow manual intervention. It cannot form intelligent collaboration with the on-site control system. When faced with alarm information, operators lack decision support and rely on personal experience to handle the situation. This results in low emergency response efficiency and may cause losses due to misjudgment.
[0058] Secondary drawbacks: The lack of accurate data support and scientific energy efficiency management methods makes it impossible to quantify and optimize energy waste. Issues: Operational quality heavily relies on the personal experience of senior technicians; personnel changes can easily lead to fluctuations in management level, and novices are prone to misjudgment.
[0059] To address the aforementioned main shortcomings (lack of a central intelligent decision-making unit), the purpose of this application is to provide an intelligent production control system for permanent magnet shielded pumps. This system introduces a high-performance industrial control robot as the core of the processing layer, constructing a four-layer architecture: "equipment layer - control layer - processing layer - monitoring and management layer," to achieve the following objectives: The robot's powerful computing capabilities are used to run a multi-pump linkage optimization algorithm, which dynamically allocates the output of each pump according to the real-time load, thereby achieving optimal system-level energy efficiency.
[0060] By leveraging the robot's deep learning capabilities to perform real-time analysis of multi-dimensional sensor data, early warning and accurate diagnosis of faults can be achieved, transforming reactive maintenance into predictive maintenance.
[0061] By leveraging the robot's data mining capabilities, control strategies and diagnostic models can be continuously optimized, enabling the system to learn and evolve continuously.
[0062] Establish a remote operation and maintenance model that integrates human and machine collaboration, where robots provide intelligent decision support and humans can intervene remotely, thereby improving emergency response efficiency.
[0063] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0064] In one exemplary embodiment, such as Figure 1 As shown, an intelligent production control system for permanent magnet shielded pumps is provided. This application adopts a layered distributed architecture, which includes, from bottom to top, the equipment layer, control layer, processing layer and monitoring and management layer. Each layer has a clear function and standard interface, realizing the separation and collaboration of data acquisition, local control, intelligent decision-making and human-machine interaction.
[0065] The equipment layer includes: a permanent magnet shielded pump unit, sensors and actuators installed on the pump body and pipelines; the sensors are used to collect the operating parameters of the permanent magnet shielded pump equipment in real time and transmit them to the control layer; the actuators are used to receive the first control command from the control layer and execute the corresponding actions; the sensors include: a thermocurrent device, a thermocouple temperature sensor, a direct application temperature sensor, a containment pressure sensor, a vibration sensor, a shaft displacement sensor, a power sensor, an inlet pressure sensor, an outlet pressure sensor, and a flow meter. The actuators include: an inlet electric valve, an outlet electric valve, a frequency converter, and an alarm device; the frequency converter and the alarm device are located in the field control cabinet.
[0066] The control layer consists of a field control cabinet, which has a built-in PLC. The PLC is used to receive the operating parameters, preprocess the operating parameters and upload them to the processing layer; it is also used to receive the second control command from the processing layer, parse it and generate a first control command to send to the actuator.
[0067] The processing layer, which is an industrial control robot, is used to receive pre-processed parameters, generate second control commands for intelligent control of the permanent magnet shielded pump unit through intelligent analysis and decision-making, and send them to the control layer; it is also used to send operating parameters, alarm information, diagnostic results and report data to the monitoring management layer.
[0068] The monitoring management layer is used to receive the operating parameters, alarm information, diagnostic results and report data, and to perform visualization and push reminders; it is also used to receive manual operation instructions and send them to the processing layer to realize remote monitoring and remote intervention.
[0069] like Figure 2 As shown, the permanent magnet shielded pump equipment includes many sensors, including: Thermocurrent device 1: located on the motor stator winding; alarm and shutdown due to excessive temperature in the winding caused by circulating fluid loss or overload; Thermocouple temperature sensor 2: located in the circulation pipeline; alarm and shutdown due to insufficient circulating fluid, magnetic decoupling, and temperature rise; Direct application temperature sensor 3: located on the sealed housing or tank; alarm and shutdown due to insufficient circulating fluid, magnetic drive decoupling, and temperature rise; Containment pressure sensor 4: located on the secondary housing; shutdown due to pressure rise caused by containment leakage; Vibration sensor 5: located on the housing corresponding to the bearing; alarm and shutdown due to vibration; Shaft displacement sensor 6: located on the pump casing; alarm and shutdown due to radial or axial wear of the sliding bearing; Power sensor 7: located in the motor circuit; alarm and shutdown due to excessive or insufficient power during operation, overload, or single-phase adjustment; Inlet electric valve 8: located on the inlet pipeline, causing the valve to open or close, thereby controlling the flow of fluid. Inlet pressure sensor 9 and outlet pressure sensor 10: These sensors can sense and measure the liquid pressure values at the pump inlet and outlet and in the pipeline in real time, ensuring that the system operates within the set safety range. Flow meter 11: Installed on the outlet pipeline, it measures the pump flow rate in real time. Outlet electric valve 12: This valve can precisely control the flow rate and pressure at the pump inlet, ensuring stable operation of the pump under design conditions. Control cabinet 13: This cabinet is responsible for collecting field data, executing robot commands, and making independent decisions in emergency situations.
[0070] Specifically, the equipment layer includes permanent magnet shielded pump units and various parameter acquisition devices (thermocurrent devices, thermocouple temperature sensors, direct application temperature sensors, containment pressure sensors, vibration sensors, shaft displacement sensors, power sensors, inlet pressure sensors, outlet pressure sensors, and flow meters) installed on each permanent magnet shielded pump and pipeline, as well as actuators (including inlet electric valves, outlet electric valves, frequency converters, and alarm devices; the frequency converters and alarm devices are located in the field control cabinet). Sensors are used to collect pump operating parameters in real time (such as speed, temperature, pressure, flow rate, vibration, power, displacement, etc.). All sensors output standard 4-20mA analog or digital signals, which are connected to the corresponding field control cabinets via shielded cables. The sampling frequency is not less than 1kHz (not less than 10kHz for vibration sensors) to ensure that high-frequency fault characteristics are not lost. Electric valves can receive commands from the control layer to realize valve opening and closing or opening degree adjustment.
[0071] The control layer consists of the control cabinet on the permanent magnet shielded pump unit. The main function of the control cabinet is: Data acquisition: The PLC acquires sensor signals through the analog input module, performs preprocessing such as filtering and normalization, and packages the data after adding a timestamp.
[0072] Local control: The PLC can directly execute local safety logic such as emergency stop, ensuring equipment safety even when the network is interrupted.
[0073] Wireless communication: The pre-processed data is uploaded to the industrial control robot in the processing layer in real time via the wireless module; at the same time, the control commands issued by the industrial control robot are received, parsed, and controlled through analog or digital output to control the frequency converter, valve opening, or alarm device.
[0074] Actuator drive: Adjusts the inverter output frequency according to instructions to change the pump speed; controls the audible and visual alarm; controls the electric valve switch.
[0075] An industrial Wi-Fi network (supporting 802.11ac / ax protocol) is used between the control layer and the processing layer to ensure high bandwidth and low latency transmission. Each control cabinet has an independent IP address, allowing industrial control robots to be addressed individually.
[0076] The core of the processing layer is an industrial control robot, whose hardware configuration includes a high-performance CPU (such as an Intel Xeon or equivalent), a GPU (NVIDIA Tesla T4 or Jetson AGX), large-capacity memory and solid-state drives, as well as a wireless network card. The industrial control robot establishes a connection with all control cabinets through an industrial Wi-Fi AP, receives data uploaded from each cabinet in real time, and performs centralized analysis and decision-making.
[0077] As the central intelligent decision-making unit in the processing layer, the industrial control robot achieves intelligent control of the permanent magnet shielded pump unit through the collaborative work of four core program modules. The core design philosophy is to enable the industrial control robot to possess both "real-time response capability" and "long-term learning capability." Like an experienced engineer, it can instantly identify anomalies and take measures, while also accumulating experience through daily work, becoming increasingly professional. The control cabinet in the control layer sends sensor data (pressure, flow, temperature, vibration, etc.) to the industrial control robot via a wireless network. The industrial control robot receives this raw data and processes it (each of the four modules performs its specific function). The fused instructions are sent to the corresponding control cabinet via the wireless network, and the PLC in the control cabinet executes the instructions (frequency adjustment, pump start / stop, alarm, etc.). The self-learning module periodically "looks back," updating the model with accumulated data. The updated model is then fed back into the fault diagnosis and multi-pump linkage modules, making the system increasingly usable.
[0078] like Figure 3 As shown, the industrial control robot includes: a multi-pump linkage and redundancy control logic module, a fault diagnosis and protection logic module, a data recording and analysis module, a self-learning and continuous evolution module, and a two-way communication interface with the monitoring and management layer.
[0079] The multi-pump linkage and redundancy control logic module is used to dynamically allocate the output of the permanent magnet shielded pump unit according to the real-time system load, the performance curves of each permanent magnet shielded pump and its health status, so as to achieve optimal energy efficiency operation, and to perform redundancy switching of start-up and shutdown when equipment is abnormal, so as to ensure continuous and stable operation of the system.
[0080] The multi-pump linkage and redundancy control logic module, based on the real-time performance curves of each pump and intelligent optimization algorithms, dynamically calculates the optimal operating scheme of the system, realizing on-demand allocation, energy-saving operation, and automatic redundancy switching, including: The load sensing submodule is used to calculate the total system load demand based on real-time pressure, flow rate, and set target values.
[0081] The efficiency optimization submodule is used to combine the performance curves of the permanent magnet shielded pumps with the health status of each permanent magnet shielded pump to solve for the optimal pump combination and speed distribution, so as to achieve the optimal system energy efficiency.
[0082] The redundancy switching submodule is used to perform a smooth start-stop switching when the permanent magnet shielded pump unit fails, ensuring continuous system operation.
[0083] Specifically, the multi-pump linkage and redundancy control logic module addresses the core questions: How much flow does the system require? How to allocate it most efficiently? What to do if a pump fails? The load sensing submodule acts like a dispatcher, taking in the actual pressure, actual flow, and the user's desired values, and outputting the total system load demand through an algorithm. The efficiency optimization submodule acts like an actuary, calculating "how to match these pumps for maximum energy efficiency." This module takes into account the performance curve of each pump (data measured at the factory), the total system load demand (from the previous module), and the health status of each pump (from the fault diagnosis module). Based on the fundamental laws of fluid mechanics, higher rotational speed results in higher flow rate: Q∝n (directly proportional), and higher rotational speed results in higher head: H∝n. 2 (Proportional to the square), the higher the rotational speed, the greater the power consumption: P∝n 3 (Proportional to the cube). The system will calculate which pump is the most energy-efficient and determine which pumps should be turned on and which should be turned off. The redundancy switching submodule acts like a safety officer, smoothly switching to the standby pump when a pump malfunctions. For example, if the fault diagnosis module receives an alarm signal (bearing wear on pump #3), the module will use a "start first, then stop" approach, slowly starting the standby pump. Once the faulty pump has completely stopped, the switchover is complete, and updated operating instructions are output.
[0084] The fault diagnosis and protection logic module is used to extract fault features from operating parameters, identify fault types and fault probabilities through a deep learning model, output graded protection instructions and alarm information, and establish a fault knowledge base for diagnosis and learning.
[0085] The fault diagnosis and protection logic module utilizes deep learning models (CNN / LSTM) to analyze multi-dimensional sensor data (vibration, temperature, pressure, current, etc.) to achieve early fault warning, accurate diagnosis, and graded protection, including: The feature extraction submodule is used to extract fault feature vectors from the operating parameters.
[0086] The deep learning submodule is used to identify the fault feature vector using a CNN or LSTM model and output the fault type and fault probability.
[0087] The protection decision submodule is used to output graded protection commands and alarm information based on the fault type and the fault probability.
[0088] The fault knowledge base stores fault characteristics, diagnostic results, and processing records, providing data support for diagnosis and learning.
[0089] Specifically, the core problem addressed by the fault diagnosis and protection logic module is: Is the equipment currently healthy? Will it malfunction? And what if a problem occurs? The feature extraction submodule, like a skilled traditional Chinese medicine practitioner, extracts useful features from raw data, obtaining pump vibration signals, temperature signals, pressure, flow rate, current, etc. For example, it extracts the vibration waveform, analyzes the frequencies contained within it, and outputs a multi-dimensional feature vector. Next, the deep learning submodule, like an experienced expert, determines "what kind of fault it is" based on the features. The feature vector from the feature extraction module is input, and using a CNN (Convolutional Neural Network), the vibration spectrum is treated as a "photograph." The CNN automatically learns from this photograph what a "broken bearing" or "unbalanced rotor" looks like. During training, it is given a large amount of labeled fault data, summarizes patterns, and outputs the probability of various fault types. The protection decision-making submodule acts like a commander, determining "what action to take" based on the diagnostic results. It takes the fault probability distribution from the previous module as input and calculates a comprehensive health status, with different faults having different weights. Bearing problems have a higher weight because they are more serious. This module outputs protection commands (sent to the control cabinet for execution) and alarm information (sent to the monitoring management layer for display). The fault knowledge base acts like a medical record, recording the characteristics, diagnostic results, and actual handling of each fault. This information is periodically analyzed by the self-learning module to update the model. When a new fault is encountered, it can check if there are similar ones in the past.
[0090] The data recording and analysis module is used to store historical operating data in time series, statistically analyze energy consumption, efficiency, and runtime, predict the remaining lifespan of the equipment, and automatically generate operation and maintenance reports and energy efficiency analysis reports.
[0091] The data recording and analysis module includes: The historical data storage submodule is used to store operating parameters, alarm information, operation logs, and historical data of equipment status using a time-series database.
[0092] The statistical analysis submodule is used to perform statistical calculations on runtime, energy consumption, efficiency, and reliability, and to predict the remaining service life based on the equipment degradation trend.
[0093] The report generation submodule is used to automatically generate daily, weekly, monthly, and annual reports as well as energy efficiency analysis reports.
[0094] Specifically, the data recording and analysis module addresses the core questions of how such a vast amount of data is generated, stored, and used. The historical data storage submodule acts like an archivist, categorizing and storing massive amounts of data. It stores high-frequency data, key data, and minute-level data, and generates daily statistics. The statistical analysis submodule acts like a data analyst, calculating useful metrics from the data. It provides statistics on pump runtime, energy consumption, efficiency, and reliability. The report generation submodule acts like a secretary, regularly compiling the analysis results into reports, generating daily, weekly, monthly, and annual reports.
[0095] The data recording and analysis module is responsible for storing historical data, performing statistical analysis, generating reports, and providing data support for other modules. It uses a time-series database to store all sensor data, supporting high-concurrency writing and fast querying; predicts remaining effective lifespan based on equipment degradation models (such as temperature trends and vibration trends); automatically generates daily operation reports, including pump runtime, start / stop counts, energy consumption statistics, average efficiency, alarm records, etc.; and periodically performs energy efficiency analysis to identify inefficient periods and equipment, and proposes optimization suggestions.
[0096] The self-learning and continuous evolution module is used to perform online incremental training of the control algorithm and diagnostic model based on new fault cases and operating parameters, and to retain historical experience by using an elastic weight consolidation mechanism to achieve continuous system optimization.
[0097] The self-learning and continuous evolution module utilizes newly added operational data and fault cases to perform online incremental training on the multi-pump linkage algorithm parameters and deep learning diagnostic model, thereby achieving the system's self-learning and continuous evolution, including: The online model update submodule is used to periodically trigger model and algorithm updates based on newly added fault cases and operating parameters.
[0098] The incremental training submodule is used to perform online incremental training on the multi-pump linkage and redundancy control logic module and the fault diagnosis and protection logic module.
[0099] Specifically, the core problem addressed by the self-learning and continuous evolution module is: how can the system become smarter with use? The online model update submodule acts like a class monitor, regularly checking whether the knowledge needs updating. It updates weekly: after adding more than 10 fault cases, it found a decrease in diagnostic accuracy. The incremental training submodule acts like a student reviewing old knowledge, updating old knowledge with new knowledge. For the fault diagnosis model, it continues training with new fault cases, but care must be taken to avoid "learning new things and forgetting old ones." The "elastic weight consolidation" technique uses parameters important to the old tasks, with smaller changes during updates. For the multi-pump linkage model: pump performance changes over time (wear and tear). New data is used to correct the performance curve, and it is found that the decision-making is not the most energy-efficient under certain operating conditions, so the algorithm weights are adjusted. The updated model is output (and returned to the fault diagnosis and multi-pump linkage modules for use).
[0100] The bidirectional communication interface with the monitoring management layer is used to push operating parameters, alarm information, diagnostic results and report data to the monitoring management layer, and to receive manual control commands, parameter settings and query requests, and to prioritize the execution of manual control commands.
[0101] The bidirectional communication interface with the monitoring management layer includes: The data delivery submodule is used to push operating parameters, alarm information, diagnostic results, and report data to the monitoring management layer.
[0102] The instruction receiving submodule is used to receive manual control instructions, parameter settings and query requests issued by the monitoring management layer, and to execute the manual control instructions first.
[0103] Specifically, the two-way communication interface with the monitoring and management system is responsible for communication between the robot and external devices (computers, mobile phones). It sends out real-time data such as pressure, flow rate, and temperature; alarm information including which device is faulty, its severity, and suggested solutions; diagnostic results such as health scores and predicted remaining lifespan; and reports such as daily and weekly reports. It receives manual control commands: manually starting and stopping a pump, setting the frequency; parameter settings such as modifying alarm thresholds and pressure setpoints; and query requests such as retrieving historical data and viewing reports.
[0104] As an optional implementation, the intelligent production control system for the permanent magnet shielded pump also includes an AI model.
[0105] The AI model is connected to the device layer, the control layer, the processing layer, and the monitoring and management layer, respectively, to provide AI functions.
[0106] like Figure 4 As shown, it clearly demonstrates the top-level capabilities of AI technology in the system and its applications at different levels, with an overall structure divided into two layers: I. Top Level: AI Capability Overview (Core Capability Layer).
[0107] It is the source of the entire architecture's capabilities, encompassing four core AI capabilities: Pattern recognition: used for classifying, recognizing, and extracting features from data, signals, and features.
[0108] Trend prediction: Predicting future states and trends based on historical data.
[0109] Decision optimization: Output the optimal decision solution based on the analysis results.
[0110] Self-learning: Continuously optimize models and algorithms through data iteration to upgrade capabilities.
[0111] II. Bottom Layer: Layered Implementation of AI Capabilities (Application Layer)
[0112] The top-level AI capabilities empower the system at four different system levels, each with a specific AI application focus: Device layer: AI-assisted preprocessing.
[0113] For sensors / actuators, AI is used to preprocess raw data, such as cleaning, noise reduction, and feature extraction, to provide high-quality data input for upper-level analysis.
[0114] Control layer: AI-assisted local judgment.
[0115] For control cabinets, AI-assisted edge-side judgment, local decision-making, and real-time control are achieved, reducing cloud dependence and improving response speed.
[0116] Processing layer: AI core deep analysis.
[0117] It is the core AI hub of the entire system, undertaking core analytical tasks such as deep data mining, complex logical operations, and core model reasoning, and is the core carrier layer of AI capabilities.
[0118] Monitoring layer: This refers to the monitoring and management layer, where AI assists in human-computer interaction.
[0119] For monitoring and maintenance scenarios, AI-assisted visualization monitoring, anomaly alarms, and intelligent interaction are achieved, improving the efficiency of human-machine collaboration.
[0120] Between layers: AI-assisted intelligent communication.
[0121] For communication links between system levels, AI is used to optimize data transmission, achieve intelligent routing, ensure communication reliability, and connect data flows at all levels.
[0122] Among these, fault diagnosis is the most core and in-depth application of AI models, such as... Figure 5As shown, vibration sensors, temperature sensors, pressure sensors, flow sensors, current sensors, etc., are used to collect real-time, multi-dimensional operating data of the equipment as raw data for diagnosis.
[0123] Edge AI Preliminary Screening and Anomaly Detection: The raw data collected at the edge is initially screened and anomaly detection is performed: if the data is determined to be normal, it is not uploaded to save bandwidth and the process ends directly; if abnormal data is detected, the abnormal data is uploaded to the control cabinet AI.
[0124] The control cabinet uses a lightweight AI-powered judgment system to perform rapid analysis of abnormal data, making two core judgments: ① whether it is an emergency fault; ② whether it can be handled locally. If it is determined to be an emergency fault, it directly triggers local protection actions and the process ends; if it is a non-emergency fault, it uploads the data to the core AI of the industrial control robot.
[0125] Robot Core AI Deep Diagnosis: As the core analysis hub of the entire process, deep diagnosis is completed in four steps: Step 1: Feature extraction, which uses an autoencoder to automatically learn fault features from the raw data.
[0126] The second step is deep learning diagnostics, which involves building a fusion model and simultaneously analyzing spatial features (spectrum) and temporal features (trends).
[0127] Step 3: Knowledge graph reasoning, using association analysis to locate the root cause of the failure.
[0128] Step 4: Decision output, clarifying the fault type, severity, and recommended handling measures.
[0129] Monitoring AI Human-Computer Interaction: Based on the diagnostic results of core AI, generate readable fault reports and recommend targeted maintenance solutions for maintenance personnel, completing the closed loop of the entire fault diagnosis process.
[0130] More specifically, the integration of AI models into the four modules of industrial control robots is shown below: 1. Fault Diagnosis and Protection Logic Module – The module with the deepest integration of AI.
[0131] AI technology can perform deep learning, for example, in vibration spectrum analysis, entering an automatic fault-learning spectrum mode, which can detect bearing wear and rotor imbalance earlier than the human eye. It can also be used for time-series trend analysis, capturing the changes in temperature and vibration over time, and predicting "it will break down in a few days".
[0132] AI technology also features anomaly detection algorithms that pre-screen data, quickly identify data points that deviate from the norm, filter normal data, and focus on anomalies.
[0133] The specific workflow of this AI module is as follows: Figure 6 As shown.
[0134] 2. Multi-pump linkage and redundant control module - AI-assisted optimization.
[0135] AI is mainly used to improve optimization results: AI technology uses neural network fitting to correct performance curves, adjusting the pump's performance curves based on actual operating data to adapt to equipment aging and wear; and it performs load forecasting to predict demand changes in the next 5-15 minutes, allowing for advance adjustments and faster response.
[0136] The specific workflow of this AI module is as follows: Figure 7 As shown.
[0137] 3. Self-learning and continuous evolution module – AI achieves self-improvement.
[0138] This module applies AI technology: AI technology enables incremental learning, online model updates, and continuous training of the model with new data. The more the system is used, the more accurate it becomes. AI can actively learn, perform data annotation, identify uncertain samples, and require manual confirmation, thus reducing the cost of manual annotation.
[0139] The specific workflow of this AI module is as follows: Figure 8 As shown.
[0140] 4. Data Recording and Analysis Module – AI-Assisted Analysis.
[0141] AI technology can be used for cluster analysis to classify operating conditions, automatically identify different operating conditions, and analyze the performance of each condition in a targeted manner.
[0142] AI technology can also predict trends, calculate the remaining lifespan of equipment, prepare spare parts in advance, and arrange maintenance.
[0143] As an optional implementation, the intelligent production control system for the permanent magnet shielded pump can calculate the pump head, motor output power, pump efficiency, shaft power, and power consumption of the permanent magnet shielded pump.
[0144] The formula for calculating the head of the permanent magnet shielded pump is as follows: ; ; ; Where H is the head of the permanent magnet shielded pump, in meters; The outlet pressure of the permanent magnet shielded pump is expressed in kPa. The inlet pressure of the permanent magnet shielded pump is kPa. The density of the medium is kg / m³. 3 ; The height difference between the inlet pressure sensor positions, in meters (m). v1 is the height difference between the outlet pressure sensor positions, in meters; v2 is the inlet velocity, in meters per second; v3 is the outlet velocity, in meters per second; g is the acceleration due to gravity, in meters per second. 2 Q represents flow rate, m 3 / h; D1 is the inlet diameter, mm; D2 is the outlet diameter, mm.
[0145] The formula for calculating the output power of the motor is: ; in, The output power of the motor is expressed in KW. This represents the input power of the motor, in kilowatts (kW). For motor efficiency.
[0146] The formula for calculating the efficiency of the permanent magnet shielded pump is as follows: ; in, The efficiency of the permanent magnet shielded pump.
[0147] The formula for calculating the shaft power is: ; in, Shaft power, KW; This refers to the specific gravity of the medium.
[0148] The formula for calculating the power consumption of the permanent magnet shielded pump is as follows: ; Where E is the power consumption of the permanent magnet shielded pump, in kWh; and t is the time, in hours.
[0149] The monitoring and management layer includes computer-based monitoring software and a mobile app. It communicates with the industrial control robot via an industrial Wi-Fi network, enabling real-time monitoring and displaying the system topology, real-time parameters of each pump (pressure, flow rate, temperature, vibration, etc.), and operating status (running / stopping / fault / early warning) graphically. It also supports alarm push notifications, receiving alarm information from the robot, with pop-up alerts on the computer and push notifications (supporting sound and vibration) on the mobile app. Remote control is supported, allowing switching between manual and automatic modes. In manual mode, operators can remotely start / stop individual pumps, adjust frequencies, and modify setpoints via the software or app. The industrial control robot prioritizes manual control commands and resumes autonomous control in automatic mode. Historical queries and reports are supported, allowing users to query historical operating data, alarm records, and operation logs, and automatically generate operating reports and energy consumption analysis reports. Parameter settings are also supported, allowing users to configure system operating parameters (pressure setpoints, alarm thresholds, PID parameters, etc.) and configure equipment information.
[0150] Remote monitoring through collaboration between mobile app and PC monitoring software provides a graphical monitoring interface, supports alarm push notifications, remote control, historical data queries, and report generation, enabling unattended or minimally staffed operation and maintenance modes.
[0151] In this application, the industrial control robot serves as the core of the processing layer, and its key role is reflected in the following aspects: Global Intelligent Decision Center: It gathers real-time data from all pumps, runs complex optimization algorithms (multi-pump linkage) and deep learning models (fault diagnosis), and makes globally optimal decisions, far exceeding the simple logic of traditional PLCs.
[0152] Fault prediction and diagnosis expert: Utilizing deep learning to analyze high-frequency data such as vibration and temperature, identify early fault characteristics (such as the sidebands of bearing wear), enabling early warnings several days in advance and transforming reactive maintenance into predictive maintenance.
[0153] Data value mining engine: Continuously stores and analyzes historical data, automatically generates energy efficiency reports and lifespan predictions, and provides decision support for operation and maintenance.
[0154] Self-learning evolution: Through the self-learning module, the system continuously optimizes control strategies and diagnostic models using new data, becoming increasingly intelligent with use and adapting to equipment aging and changes in operating conditions.
[0155] The intelligent interface for human-machine collaboration: On the one hand, it autonomously controls the operation of the pump group; on the other hand, it receives and prioritizes the execution of remote human commands. It provides diagnostic suggestions during alarms to assist human decision-making, forming an efficient human-machine collaboration mode. In other words, the industrial control robot has a dual-mode operation mechanism: the system supports automatic / manual mode switching. The robot autonomously controls the operation of the pump group according to preset logic, while simultaneously receiving and prioritizing remote manual intervention commands from the monitoring and management layer, achieving seamless integration of automatic and manual operation.
[0156] This application uses an industrial control robot as the core of the processing layer, and runs a multi-pump linkage optimization algorithm, a deep learning fault diagnosis model, and a self-learning module, which has the following advantages: Advantage 1: The overall energy efficiency of the system is significantly improved.
[0157] Traditional solution: Multiple pumps use simple PID or timed rotation, each pump is adjusted independently, which cannot be dynamically allocated according to real-time load. Some pumps deviate from the high-efficiency range for a long time, resulting in high energy consumption.
[0158] The proposed solution is a multi-pump linkage module that uses the performance curves of each pump and intelligent optimization algorithms to solve for the optimal combination of the system in real time, so that each pump can operate in the high-efficiency range as much as possible.
[0159] According to the pump similarity law, power is directly proportional to the cube of the rotational speed, and energy saving is significant during low-frequency operation. Comprehensive calculations show that this application can achieve system energy savings of 15% to 30% compared to traditional solutions, while also balancing wear on each pump and extending its overall lifespan.
[0160] Advantage 2: Faults are detected earlier, and unplanned downtime is greatly reduced.
[0161] Traditional solutions can only trigger threshold alarms. By the time parameters exceed the limits, the fault has already occurred or even caused damage, making it a reactive repair.
[0162] The proposed solution: The fault diagnosis module utilizes deep learning to analyze early features such as vibration spectrum and temperature change rate, and can issue early warnings at the nascent stage of a fault (such as the early stage of bearing wear).
[0163] Taking bearing wear as an example, the early warning system can provide warnings several days to weeks earlier than traditional alarms, allowing maintenance personnel to schedule repairs in advance and avoid unexpected downtime. Unplanned downtime is expected to be reduced by more than 50%, and maintenance costs by 30% to 50%.
[0164] Advantage 3: The system has self-learning and continuous optimization capabilities.
[0165] Traditional approach: The control logic is fixed and cannot learn from historical data to improve.
[0166] The proposed solution involves a self-learning module that continuously updates the parameters of the multi-pump linkage algorithm and the deep learning diagnostic model using new data.
[0167] The more the system is used, the smarter it becomes. Its control strategies and diagnostic accuracy improve over time, adapting to equipment aging and changes in operating conditions, and always maintaining optimal performance.
[0168] Advantage 4: High efficiency of human-machine collaboration and reduced operation and maintenance costs.
[0169] Traditional solutions require 24-hour staffing or frequent inspections, and emergency response relies on on-site handling, resulting in low efficiency.
[0170] This proposed solution involves a robot that operates autonomously, monitors and sends real-time status updates and alarms to the management team, and allows operators to monitor the system anytime, anywhere via a mobile app and intervene remotely.
[0171] This enables fewer or even no staff on duty, reducing labor costs by more than 50%; and shortening emergency response time from hours to minutes.
[0172] Advantage 5: The system has high reliability and strong adaptability.
[0173] Traditional solutions rely on a single controller and wired network, which poses a high risk of single point of failure and makes expansion difficult.
[0174] The proposed solution features a control cabinet capable of independent local operation, which can be degraded to maintain operation during network outages; wireless communication facilitates flexible expansion with new pump units; and robot malfunctions do not affect basic local control.
[0175] The system is robust, suitable for complex industrial environments, and easy to expand and upgrade.
[0176] Advantage 6: Data-driven decision-making and refined management.
[0177] Traditional approach: Only simple reports are provided, and the value of the data is not explored.
[0178] This application proposes a solution that automatically generates reports such as energy efficiency analysis, lifespan prediction, and operation and maintenance recommendations, providing data support for equipment management and optimization.
[0179] To achieve a shift from experience-driven to data-driven approaches and improve overall management efficiency.
[0180] In summary, this application fundamentally solves the pain point of insufficient intelligence in existing systems by introducing industrial control into robots as the core of intelligent decision-making, and has significant advantages in energy saving, safety, reliability, and operation and maintenance.
[0181] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0182] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. An intelligent production control system for a permanent magnet shielded pump, characterized in that, The intelligent production control system for the permanent magnet shielded pump includes: an equipment layer, a control layer, a processing layer, and a monitoring and management layer. The equipment layer includes: a permanent magnet shielded pump unit, sensors and actuators installed on the pump body and pipelines; the sensors are used to collect the operating parameters of the permanent magnet shielded pump equipment in real time and transmit them to the control layer; the actuators are used to receive the first control command from the control layer and execute the corresponding actions; the sensors include: a thermocouple temperature sensor, a direct application temperature sensor, a containment pressure sensor, a vibration sensor, a shaft displacement sensor, a power sensor, an inlet pressure sensor, an outlet pressure sensor, and a flow meter; the actuators include: an inlet electric valve, an outlet electric valve, a frequency converter, and an alarm device; the frequency converter and the alarm device are located in the field control cabinet; The control layer consists of a field control cabinet, which has a built-in PLC. The control cabinet is used to receive the operating parameters, preprocess the operating parameters and upload them to the processing layer; it is also used to receive the second control command from the processing layer, parse it and generate a first control command to send to the actuator. The processing layer, which is an industrial control robot, is used to receive pre-processed parameters, generate second control commands for intelligent control of the permanent magnet shielded pump unit through intelligent analysis and decision-making, and send them to the control layer; it is also used to send operating parameters, alarm information, diagnostic results and report data to the monitoring and management layer. The monitoring management layer is used to receive the operating parameters, alarm information, diagnostic results and report data, and to perform visualization and push reminders; it is also used to receive manual operation instructions and send them to the processing layer to realize remote monitoring and remote intervention.
2. The intelligent production control system for permanent magnet shielded pumps according to claim 1, characterized in that, The industrial control robot includes: a multi-pump linkage and redundancy control logic module, a fault diagnosis and protection logic module, a data recording and analysis module, a self-learning and continuous evolution module, and a two-way communication interface with the monitoring and management layer. The multi-pump linkage and redundancy control logic module is used to dynamically allocate the output of the permanent magnet shielded pump unit according to the real-time load of the system, the performance curves of each permanent magnet shielded pump and the health status, so as to achieve optimal energy efficiency operation, and to perform redundancy switching of start-up and shutdown when the equipment is abnormal, so as to ensure the continuous and stable operation of the system. The fault diagnosis and protection logic module is used to extract fault features from operating parameters, identify fault types and fault probabilities through a deep learning model, output graded protection instructions and alarm information, and establish a fault knowledge base for diagnosis and learning. The data recording and analysis module is used to store historical operating data in time series, statistically analyze energy consumption, efficiency, and operating time, predict the remaining lifespan of the equipment, and automatically generate operation and maintenance reports and energy efficiency analysis reports. The self-learning and continuous evolution module is used to perform online incremental training of the control algorithm and diagnostic model based on new fault cases and operating parameters, and to retain historical experience by using an elastic weight consolidation mechanism to achieve continuous system optimization. The bidirectional communication interface with the monitoring management layer is used to push operating parameters, alarm information, diagnostic results and report data to the monitoring management layer, and to receive manual control commands, parameter settings and query requests, and to prioritize the execution of manual control commands.
3. The intelligent production control system for permanent magnet shielded pumps according to claim 2, characterized in that, The multi-pump linkage and redundancy control logic module includes: The load sensing submodule is used to calculate the total system load demand based on real-time pressure, flow rate, and set target values. The efficiency optimization submodule is used to combine the performance curves of permanent magnet shielded pumps with the health status of each permanent magnet shielded pump to solve for the optimal pump combination and speed distribution, so as to achieve the optimal system energy efficiency. The redundancy switching submodule is used to perform a smooth start-stop switching when the permanent magnet shielded pump unit fails, ensuring continuous system operation.
4. The intelligent production control system for permanent magnet shielded pumps according to claim 2, characterized in that, The fault diagnosis and protection logic module includes: A feature extraction submodule is used to extract fault feature vectors from the operating parameters; The deep learning submodule is used to identify the fault feature vector using a CNN or LSTM model and output the fault type and fault probability. The protection decision submodule is used to output graded protection instructions and alarm information according to the fault type and the fault probability; The fault knowledge base stores fault characteristics, diagnostic results, and processing records, providing data support for diagnosis and learning.
5. The intelligent production control system for permanent magnet shielded pumps according to claim 2, characterized in that, The data recording and analysis module includes: The historical data storage submodule is used to store operating parameters, alarm information, operation logs, and historical equipment status data using a time-series database. The statistical analysis submodule is used to perform statistical calculations on runtime, energy consumption, efficiency, and reliability, and predict the remaining service life based on equipment degradation trends. The report generation submodule is used to automatically generate daily, weekly, monthly, and annual reports as well as energy efficiency analysis reports.
6. The intelligent production control system for permanent magnet shielded pumps according to claim 2, characterized in that, The self-learning and continuous evolution module includes: The online model update submodule is used to periodically trigger model and algorithm updates based on newly added fault cases and operating parameters; The incremental training submodule is used to perform online incremental training on the multi-pump linkage and redundancy control logic module and the fault diagnosis and protection logic module.
7. The intelligent production control system for permanent magnet shielded pumps according to claim 2, characterized in that, The bidirectional communication interface with the monitoring management layer includes: The data delivery submodule is used to push operating parameters, alarm information, diagnostic results and report data to the monitoring management layer; The instruction receiving submodule is used to receive manual control instructions, parameter settings and query requests issued by the monitoring management layer, and to execute the manual control instructions first.
8. The intelligent production control system for permanent magnet shielded pumps according to claim 1, characterized in that, The intelligent production control system for the permanent magnet shielded pump also includes: an AI model; The AI model is connected to the device layer, the control layer, the processing layer, and the monitoring and management layer, respectively, to provide AI functions.
9. The intelligent production control system for permanent magnet shielded pumps according to claim 1, characterized in that, The control layer and the processing layer communicate via industrial Wi-Fi. Each control cabinet has an independent IP address, supporting separate addressing by the processing layer. When the network is interrupted, the control cabinet can independently execute local shutdown and protection logic.
10. The intelligent production control system for permanent magnet shielded pumps according to claim 1, characterized in that, The intelligent production control system for the permanent magnet shielded pump can calculate the pump head, motor output power, pump efficiency, shaft power, and power consumption of the permanent magnet shielded pump. The formula for calculating the head of the permanent magnet shielded pump is as follows: ； ； ； Where H is the head of the permanent magnet shielded pump, in meters; The outlet pressure of the permanent magnet shielded pump is expressed in kPa. The inlet pressure of the permanent magnet shielded pump is kPa. The density of the medium is kg / m³. 3 ; The height difference between the inlet pressure sensor positions, in meters (m). v1 is the height difference between the outlet pressure sensor positions, in meters; v2 is the inlet velocity, in meters per second; v3 is the outlet velocity, in meters per second; g is the acceleration due to gravity, in meters per second. 2 Q represents flow rate, m 3 / h; D1 is the inlet diameter, mm; D2 is the outlet diameter, mm; The formula for calculating the output power of the motor is: ； in, The output power of the motor is expressed in KW. This represents the input power of the motor, in kilowatts (kW). For motor efficiency; The formula for calculating the efficiency of the permanent magnet shielded pump is as follows: ； in, The efficiency of the permanent magnet shielded pump; The formula for calculating the shaft power is: ； in, Shaft power, KW; Specific gravity of the medium; The formula for calculating the power consumption of the permanent magnet shielded pump is as follows: ； Where E is the power consumption of the permanent magnet shielded pump, in kWh; and t is the time, in hours.