Multi-stage pump cooperative variable flow control method and system based on edge calculation

Abstract

The invention discloses a multi-stage pump collaborative variable flow control method and system based on edge computing, which are applied to the technical field of edge computing and provide basic data for subsequent load prediction by acquiring real-time operating parameters and environmental parameters of a second-stage pump set in a cold supply system. The real-time operation parameters and the environment parameters are input into a preset load prediction model, and target operation parameters of a second-stage pump set are determined and comprise the number of started pump sets and the operation frequency; based on the target operation parameters of the second-stage pump set, the linkage control parameters, matched with the output flow of the second-stage pump set, of the first-stage pump set are calculated, and flow collaboration between the two stages of pump sets is ensured; and according to the target operation parameters and the linkage control parameters, the operation states of the first-stage pump set and the second-stage pump set are synchronously adjusted, so that the total flow of the first-stage pump set and the total flow of the second-stage pump set tend to be consistent. Therefore, the method has the advantage that the system energy efficiency and the operation economy are improved.

Description

Technical Field

[0001] This invention relates to the field of edge computing technology, and in particular to a multi-stage pump collaborative variable flow control method and system based on edge computing. Background Technology

[0002] With the rapid development of modern industry, especially large buildings and data centers, higher demands are being placed on the stability, energy efficiency, and operating costs of cooling systems. The primary and secondary pump technologies in central air conditioning chilled water systems have undergone multiple innovations and optimizations, evolving from traditional constant-speed pump systems to intelligent variable frequency control systems. These technological advancements have significantly improved the energy efficiency and reliability of the systems.

[0003] However, in actual operation, achieving coordinated variable flow control between primary and secondary pumps in a cooling system remains a pressing technical challenge. In existing technologies, primary pumps typically operate at a fixed frequency, while secondary pumps operate at a variable frequency. This operating mode makes it difficult to achieve precise flow matching between the two pump sets. This leads to a problem where large-flow bypass phenomena easily occur in the bypass lines of the cooling system, a condition known as unsuitable high-flow bypass in the supply and demand pipe. This unbalanced flow not only causes significant energy loss and reduces the overall energy efficiency of the system but may also affect the stability and comfort of the terminal cooling effect. Furthermore, traditional control strategies often struggle to maximize cooling efficiency and minimize refrigeration costs when dealing with complex load changes and external environmental parameters. Therefore, the industry has long faced the challenge of precisely coordinating and controlling the number and frequency of primary and secondary pumps to eliminate unbalanced flow in bypass lines, thereby further improving system energy efficiency and operational economy.

[0004] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, the present invention aims to provide a multi-stage pump collaborative variable flow control method and system based on edge computing, which solves the problem of unbalanced flow and energy loss in the bypass pipeline caused by the difficulty in accurately matching the flow of the primary and secondary pumps in the prior art, and has the advantages of improving system energy efficiency and operating economy.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, a multi-stage pump collaborative variable flow control method based on edge computing is provided for a cooling system, the cooling system comprising a first-stage pump group and a second-stage pump group, the method comprising the following steps: S1: Obtain the real-time operating parameters and environmental parameters of the second-stage pump unit in the cooling system; S2: Input the real-time operating parameters and environmental parameters into the preset load prediction model to determine the target operating parameters of the second-stage pump group. The target operating parameters include the number of pump groups that are turned on and the operating frequency. S3: Based on the target operating parameters of the second-stage pump set, calculate the linkage control parameters of the first-stage pump set that match the output flow of the second-stage pump set; S4: Based on the target operating parameters and the linkage control parameters, synchronously adjust the operating states of the first-stage pump group and the second-stage pump group so that the total flow rate of the first-stage pump group and the total flow rate of the second-stage pump group tend to be consistent.

[0007] Furthermore, the real-time operating parameters include the flow rate, inlet and outlet water temperature, and pressure of the second-stage pump set; the environmental parameters include outdoor meteorological parameters.

[0008] Furthermore, step S2 includes: S21: Input the real-time operating parameters and environmental parameters into the load forecasting model to perform positive load forecasting and obtain the end-point load demand value; S22: Perform reverse reasoning calculations based on the terminal load demand value to determine the number of pumps to be started and the operating frequency of the second-stage pump set.

[0009] Furthermore, step S3 includes: S31: Calculate the total output flow of the second-stage pump set under the stated terminal load demand value based on the number of pumps started and the operating frequency of the second-stage pump set; S32: Based on the total output flow rate, calculate the flow balance point of the first-stage pump set under the same load demand, and use the number of pump sets turned on and the operating frequency corresponding to the flow balance point as the linkage control parameters.

[0010] Furthermore, step S4 includes: S41: Based on the target operating parameters and the linkage control parameters, adjust the power output of the first-stage pump group and the second-stage pump group in real time to eliminate the unbalanced flow in the bypass pipeline of the cooling system and maintain the unbalanced flow within a preset threshold range.

[0011] Furthermore, the cooling system also includes cooling pumps and chiller units, and the first-stage pump set includes a primary chiller pump; the primary chiller pump, the cooling pump and the chiller unit are connected in a one-to-one correspondence so that each chiller unit can operate independently with variable flow rate; The second-stage pump set adopts a redundant design, which includes an operating pump that meets the normal operation requirements of the cooling system and at least one standby pump. The cooling system also includes a cooling tower system, which adopts a grouped parallel design to allow for grouped maintenance of each group of cooling towers without shutting down the cooling system.

[0012] Furthermore, the chiller unit includes a cold storage fixed-frequency dual-condition chiller unit and a direct chiller unit; the method also includes the following steps: S5: When the terminal cooling load exceeds the preset load threshold, or when the direct cooling unit malfunctions, control the cold storage fixed frequency dual-condition unit to connect to the direct cooling network for cooling. The cooling system further includes a cold storage tank; the method further includes: S6: During off-peak electricity price periods, control the cold storage fixed-frequency dual-condition unit to perform cold storage and store the cold energy in the cold storage tank; S7: During peak electricity price periods, release the cold energy in the cold storage tank.

[0013] Furthermore, the cooling system is controlled through a distributed edge computing system architecture, which includes energy management software and edge computing modules corresponding to the chiller unit, the first-stage pump unit, the second-stage pump unit, the cooling pump, the cooling tower, and the cold storage tank, respectively; the method also includes the following steps: S8: Local data processing and logic control are performed through each of the aforementioned edge computing modules; S9: The energy management software works in collaboration with each edge computing module to perform global energy optimization management.

[0014] Furthermore, step S8 includes: S81: Establish a synchronous clock reference between the first-stage pump set and the second-stage pump set; S82: Generate a predictive control sequence containing the evolution trend of the operating state within a future preset time period based on the target operating parameters and the linkage control parameters; S83: The predictive control sequence is synchronously distributed to the edge computing modules corresponding to the first-stage pump group and the second-stage pump group; S84: When data packet loss or delay occurs in the communication link between each edge computing module, each edge computing module autonomously executes local control output according to the evolution trend in the predictive control sequence to maintain the flow coordination between the first-stage pump group and the second-stage pump group during the communication anomaly.

[0015] Secondly, a multi-stage pump collaborative variable flow control device based on edge computing, the device comprising: Parameter acquisition module: acquires the real-time operating parameters and environmental parameters of the second-stage pump unit in the cooling system; Load forecasting module: Input the real-time operating parameters and environmental parameters into a preset load forecasting model to determine the target operating parameters of the second-stage pump set. The target operating parameters include the number of pump sets in operation and the operating frequency. Linkage calculation module: Based on the target operating parameters of the second-stage pump set, calculates the linkage control parameters of the first-stage pump set that match the output flow of the second-stage pump set; Coordinated adjustment module: Based on the target operating parameters and the linkage control parameters, synchronously adjust the operating status of the first-stage pump group and the second-stage pump group so that the total flow rate of the first-stage pump group and the total flow rate of the second-stage pump group tend to be consistent.

[0016] Compared to existing technologies, the advantages of this invention are as follows: By acquiring real-time operating parameters and environmental parameters of the second-stage pump unit in the cooling system, basic data is provided for subsequent load forecasting; by inputting the real-time operating parameters and environmental parameters into a preset load forecasting model, the target operating parameters of the second-stage pump unit are determined, including the number of pump units in operation and the operating frequency, thus realizing intelligent prediction and planning of the operating status of the second-stage pump unit; based on the target operating parameters of the second-stage pump unit, the linkage control parameters of the first-stage pump unit that match the output flow of the second-stage pump unit are calculated, ensuring flow coordination between the two pump units; according to the target operating parameters and linkage control parameters, the operating status of the first-stage pump unit and the second-stage pump unit are synchronously adjusted so that the total flow of the first-stage pump unit tends to be consistent with the total flow of the second-stage pump unit, solving the problem of unbalanced flow and energy loss in the bypass pipeline caused by the difficulty in accurately matching the flow of the first-stage and second-stage pumps in the prior art. Therefore, this application has the advantages of improving system energy efficiency and operating economy. Attached Figure Description

[0017] Figure 1 The flowchart of a multi-stage pump collaborative variable flow control method based on edge computing provided by the present invention is shown.

[0018] Figure 2 The present invention provides a structural diagram of a multi-stage pump collaborative variable flow control system based on edge computing.

[0019] Figure 3 This is a schematic diagram of a multi-stage pump collaborative variable flow control system based on edge computing provided by the present invention.

[0020] In the diagram: 201, Parameter Acquisition Module; 202, Load Prediction Module; 203, Linkage Calculation Module; 204, Coordinated Regulation Module. Detailed Implementation

[0021] The present invention will now be further described with reference to the accompanying drawings and specific embodiments: This application proposes a multi-stage pump collaborative variable flow control method based on edge computing, applied to a large-scale cooling system comprising at least one first-stage pump group and one second-stage pump group. The core workflow of this method begins with a comprehensive understanding of the cooling system's status, followed by intelligent prediction and precise calculation to ultimately achieve synchronized and collaborative operation of the two pump groups. This aims to solve the problem of flow mismatch between the first and second-stage pumps, leading to energy waste in bypass pipelines, which is a common issue in traditional control methods.

[0022] Specifically, the method includes the following steps: First, acquire the real-time operating parameters of the second-stage pump unit in the cooling system, as well as environmental parameters affecting the cooling load. Acquiring this data is the foundation for all subsequent decisions. Then, using the acquired real-time operating parameters and environmental parameters as input, a pre-defined load prediction model is provided. This model, through complex calculations, determines the target operating parameters that the second-stage pump unit should achieve to meet the current and future terminal cooling load demands. These parameters specifically represent the number of pumps that need to be activated and the operating frequency of each pump. After determining the operating targets of the second-stage pump unit, it is necessary to ensure that the first-stage pump unit, as the cold source supplier, can closely cooperate with it. Therefore, the next step is to calculate and determine the linkage control parameters of the first-stage pump unit that perfectly match the expected output flow of the second-stage pump unit based on the target operating parameters of the second-stage pump unit. Finally, according to the calculated target operating parameters of the second-stage pump unit and the linkage control parameters of the first-stage pump unit, the control system issues commands to the actuators of both pump units to synchronously adjust the operating states of the first and second-stage pump units. The ultimate goal of this synchronization adjustment is to make the total flow generated by the first-stage pump set consistent with the total flow delivered to the end by the second-stage pump set, thereby minimizing or even eliminating the unbalanced flow in the bypass pipeline and improving the energy efficiency of the entire cooling system.

[0023] In one specific embodiment, the load forecasting model can be a deep learning-based neural network model, such as a Long Short-Term Memory (LSTM) network or a Gated Recurrent Unit (GRU). Before deployment, the model is trained using historical operating data. This training data includes the flow rate, inlet and outlet water temperatures, pressures, and corresponding outdoor dry-bulb temperature, wet-bulb temperature, and relative humidity of the second-stage pump unit over a past period, as well as the actual terminal cooling load data for the same period. The model structure comprises multiple input layers, each receiving different types of real-time operating parameters and environmental parameters. This input data undergoes nonlinear transformation and feature extraction within a multi-layer neural network, ultimately generating a predicted value at the output layer—the terminal cooling load demand at a future point in time (e.g., 15 or 30 minutes from now). During training, the model adjusts its internal weights and biases using a backpropagation algorithm to minimize the error between the predicted and actual values.

[0024] In traditional cooling systems, the primary pump typically operates at a constant flow rate, while the secondary pump adjusts its flow rate according to changes in the terminal load. An inherent drawback of this model is the frequent discrepancy between the constant water supply of the primary pump and the fluctuating demand of the secondary pump. When the secondary pump's flow demand decreases, excess chilled water flows directly back to the primary pump's inlet through the bypass pipe between the primary and secondary pumps, creating an ineffective cycle. This water flow does not participate in the cooling exchange at the terminal, but the transportation process consumes a significant amount of pump power, resulting in substantial energy waste. The method proposed in this application establishes a predictive, interconnected control mechanism, enabling the primary pump's flow rate to follow the secondary pump's demand changes in real time, fundamentally solving the flow mismatch problem and achieving efficient operation under all operating conditions.

[0025] Furthermore, in order for the load forecasting model to make accurate judgments, the input parameters need to be comprehensive and representative. In one specific implementation, the acquired real-time operating parameters specifically include the flow rate, inlet and outlet water temperatures, and pressure of the second-stage pump unit. Simultaneously, the acquired environmental parameters include outdoor meteorological parameters.

[0026] In detail, the flow rate of the second-stage pump unit is the most direct indicator of the amount of cooling capacity being delivered to the end-user. It is typically monitored continuously using a flow meter, such as an ultrasonic or electromagnetic flow meter, installed on the main outlet pipe of the second-stage pump unit. The inlet and outlet water temperatures—the temperatures of the main outlet and return pipes of the second-stage pump unit—are multiplied by the flow rate to directly calculate the actual cooling load consumed at the end-user. This temperature data is collected in real-time by high-precision temperature sensors, such as PT100 platinum resistance thermometers, installed on the corresponding pipes. Pressure, especially the pressure difference between the inlet and outlet of the second-stage pump unit, reflects the hydraulic resistance characteristics of the end-user network and whether the head provided by the pump unit is appropriate. Pressure data is measured by a pressure transmitter. These internal operating parameters together constitute a detailed characterization of the end-user load demand.

[0027] Outdoor meteorological parameters, as key external factors influencing a building's cooling load, are equally indispensable. These parameters typically include the dry-bulb temperature, wet-bulb temperature, and relative humidity of the outdoor air. Dry-bulb temperature directly affects heat transfer through the building envelope; relative humidity affects human comfort perception, thus influencing indoor temperature settings; and wet-bulb temperature is a core indicator for evaluating the cooling tower's heat dissipation capacity, indirectly affecting the efficiency of the entire refrigeration cycle. This meteorological data can be collected by automatic weather stations located outside the building and transmitted to the control system via communication networks. By combining these internal operating parameters with external environmental parameters, the load forecasting model can establish a more complete and robust input vector, thereby significantly improving the accuracy of predicting future load change trends.

[0028] In the above method, the step of determining the target operating parameters of the second-stage pump set, i.e., step S2, can be further decomposed into a refined process that includes forward prediction and backward reasoning. Specifically, this step includes: First, inputting the acquired real-time operating parameters and environmental parameters into the load prediction model to perform forward load prediction calculations, the result of which is a clear end-point load demand value. Then, based on this predicted end-point load demand value, backward reasoning calculations are performed to finally determine the specific number of second-stage pump sets that need to be activated and the operating frequency that each pump should maintain in order to meet this load demand.

[0029] This process can be understood as a two-stage decision-making process. The first stage is predicting the future, i.e., positive load forecasting. The load forecasting model plays a central role here. This model is not a simple linear regression, but a complex model trained on a large amount of historical data, such as an adaptive artificial intelligence inference model. This model can learn and understand the nonlinear, time-varying relationship between various operating and environmental parameters and the terminal cooling load. For example, the model can identify the typical daily variation curve of the terminal cooling load under a specific outdoor temperature and humidity, along with changes in the density of people in the office building. When new real-time data is input, the model can predict the average cooling load demand within a short time window in the future, such as the next 15 or 30 minutes, based on the complex mapping relationship it has established internally. This demand value can be expressed in a specific physical quantity, such as how many kilowatt-hours or how many refrigeration tons.

[0030] The second stage is planning, or backward reasoning. After obtaining a clear load demand value, the control system needs to translate it into specific operating instructions for the pump set. This backward reasoning process is an optimization solution process. The control system internally stores the performance curve data of each pump in the second-stage pump set, which describes the relationship between the flow rate, head, and power consumption of the pump at different operating frequencies. The goal of the backward reasoning algorithm is to find a pump set operation combination—that is, a combination of the number of pumps operating and the operating frequency—that minimizes the total energy consumption of the entire pump set, while meeting the predicted end-load demand value. For example, for a medium load, the algorithm might compare the energy consumption of operating three pumps, each running at 40 Hz, versus operating two pumps, each running at 55 Hz, and select the one with lower energy consumption as the final target operating parameters. Through this combination of forward prediction and backward reasoning, it is ensured that the operating state of the second-stage pump set not only meets the demand but also meets the demand in the most economical way.

[0031] After determining the target operating parameters of the second-stage pump set, the step of calculating the linkage control parameters of the first-stage pump set, i.e., step S3, can also be broken down into two closely linked sub-steps. Specifically, this step includes: First, based on the number of second-stage pump sets in operation and their operating frequency determined in the previous step, accurately calculate the total output flow rate that the second-stage pump sets will generate when meeting the predicted end-load demand. Then, using this calculated total output flow rate as the benchmark and target, calculate the flow balance point of the first-stage pump set under the same load demand, and use the number of first-stage pump sets in operation and their operating frequency corresponding to reaching this flow balance point as the final output linkage control parameters.

[0032] The core idea of ​​this process is demand-driven supply and flow matching. By using the target operating parameters of the second-stage pump set—namely, the number of pumps in operation and the operating frequency—combined with the performance curve of each pump, the total flow rate that the second-stage pump set will deliver to the terminal network under the current network resistance can be calculated very accurately. For example, if it is determined that two pumps of model A will be in operation at a frequency of 50 Hz, referring to the performance curve of pump A, we can find that the flow rate of a single pump is Q at the current head. Therefore, the total output flow rate is 2 multiplied by Q. This total output flow rate represents the chilled water flow rate required by the terminal in the next control cycle.

[0033] The calculated total output flow rate of the second-stage pump set becomes a clear instruction to the first-stage pump set. Upon receiving this target flow rate value, the control system's linkage calculation module executes an optimization calculation process similar to the reverse reasoning of the second-stage pump set, but this time the optimization target is the first-stage pump set. The calculation module queries the performance curves of each pump in the first-stage pump set to find an operating combination that can generate the target flow rate with the lowest energy consumption. This combination—the number of first-stage pumps that need to be activated and their respective operating frequencies—is the so-called flow balance point. Using the operating parameters corresponding to this balance point as linkage control parameters ensures that the amount of water flowing out from the cold source side is exactly equal to the amount of water needed at the terminal. This precise flow matching is key to eliminating unbalanced flow in bypass pipelines and achieving system energy savings.

[0034] Finally, the step of synchronously adjusting the operating status of the first-stage pump group and the second-stage pump group, namely step S4, is specifically executed as follows: based on the determined target operating parameters of the second-stage pump group and the linkage control parameters of the first-stage pump group, the power output of the first-stage pump group and the second-stage pump group is adjusted in real time through the control system. Its direct purpose is to eliminate the unbalanced flow in the bypass pipeline of the cooling system and keep the unbalanced flow within a preset minimum threshold range.

[0035] This step is the closed-loop execution link of the entire control strategy. The control system sends the calculated number of operating pumps and frequency commands to the frequency converters controlling the first-stage and second-stage pump groups, respectively. Upon receiving the commands, the frequency converters immediately adjust the power frequency supplied to the pump motors, thereby changing the motor speed and precisely controlling the pump's power output and flow rate. This adjustment is real-time and continuous. To verify the adjustment effect, a bidirectional flow meter is installed on the bypass line in the cooling system to monitor the magnitude and direction of the unbalanced flow in real time. The reading of this flow meter is sent back to the control system as a feedback signal.

[0036] Ideally, the flow rate in the bypass line should be zero when the total flow rate of the first-stage pump set is exactly equal to that of the second-stage pump set. However, in actual operation, achieving zero flow is difficult due to minute fluctuations in network resistance or slight control delays. Therefore, an acceptable threshold range is set; for example, the absolute value of the unbalanced flow rate is required to not exceed five percent of the total flow rate of the second-stage pump set. The control system continuously compares the actual measured value of the bypass flow rate with this threshold. If the measured value exceeds the threshold range, the control system makes fine adjustments, such as slightly increasing or decreasing the operating frequency of the first-stage pump set, until the unbalanced flow rate returns to the allowable range. This dynamic adjustment based on real-time feedback ensures close coordination between the two pump sets throughout the entire operation, physically eliminating large-scale energy waste.

[0037] To support the efficient and reliable operation of the aforementioned control methods, the physical design of the cooling system also needs corresponding optimization. In a preferred embodiment, the cooling system further includes cooling pumps and chiller units, with the first-stage pump group specifically consisting of primary chilled pumps. The primary chilled pumps, cooling pumps, and chiller units are connected in a one-to-one correspondence, allowing each chiller unit to operate independently with variable flow rates. Simultaneously, the second-stage pump group employs a redundant design, meaning that in addition to the operating pumps meeting the normal operating requirements of the cooling system, at least one standby pump is provided. Furthermore, the cooling system includes a cooling tower system, which adopts a grouped parallel design, allowing for grouped maintenance of each cooling tower without interrupting the operation of the entire cooling system.

[0038] Specifically, the one-to-one connection between the chilled water pumps, cooling pumps, and chiller units means that each chiller unit has its own dedicated chilled water circulation pump and cooling water circulation pump. The biggest advantage of this design is that it eliminates the hydraulic coupling between the chiller units. When the system starts or stops a chiller unit according to load requirements, its corresponding pumps also start and stop accordingly, preventing hydraulic shock to other operating units. More importantly, each unit can independently adjust its chilled water and cooling water flow rates according to its own load rate, ensuring it always operates within its most efficient range. This is crucial for improving the overall energy efficiency ratio of the entire chiller plant.

[0039] The redundant design of the secondary pump set typically employs an N+1 configuration, where N represents the number of operating pumps required to meet the maximum design load. This additional standby pump provides a robust guarantee for system reliability. When any operating secondary pump fails, the control system automatically detects the anomaly and immediately activates the standby pump, seamlessly taking over the work of the failed pump and ensuring uninterrupted cooling service to the end users. This is particularly important for locations with extremely high requirements for cooling continuity, such as data centers and precision manufacturing workshops.

[0040] The parallel design of cooling tower systems significantly improves system maintainability. For example, a large system might have twenty cooling towers, which can be divided into three groups, such as eight, eight, and four towers. When maintenance work such as cleaning, repair, or packing replacement is needed in one group, that group can simply be isolated from the main system via valves, while the other two groups can continue to operate and handle the overall system's heat dissipation load. This avoids the predicament of having to shut down the entire cooling system for cooling tower maintenance, ensuring the continuity of production or operation.

[0041] To further enhance the economic efficiency and resilience of the cooling system under extreme conditions, the selection and operation strategies of the chillers have been specially designed. In one specific embodiment, the chillers include a cold storage fixed-frequency dual-mode unit and a direct cooling unit. Correspondingly, the control method also includes an emergency and peak-shaving control logic: when the terminal cooling load exceeds a preset load threshold, or when the direct cooling unit malfunctions, the control system will control the cold storage fixed-frequency dual-mode unit to connect to the direct cooling network for cooling. Furthermore, the cooling system is equipped with a large-capacity cold storage tank and operates under the following strategy: during off-peak electricity price periods, the cold storage fixed-frequency dual-mode unit stores the generated cooling energy in the cold storage tank; during peak electricity price periods, the pre-stored cooling energy in the cold storage tank is released to meet cooling demand.

[0042] The core of this configuration and strategy lies in utilizing cold storage technology to achieve peak shaving and valley filling. Cold storage fixed-frequency dual-mode units can directly supply cooling to end-users in normal mode, or in cold storage mode, they can generate cooling capacity by cooling at lower temperatures. Direct cooling units primarily handle the daily base load. When extreme high temperatures occur in summer, causing a surge in end-user cooling load beyond the total capacity of the direct cooling units, the cold storage units can immediately switch to direct cooling mode, serving as a powerful backup to ensure cooling demand is met. Similarly, if a direct cooling unit unexpectedly shuts down, the cold storage units can quickly take over, ensuring system reliability.

[0043] Even more economically valuable is leveraging the peak-valley difference in electricity prices. During off-peak hours, such as at night, electricity costs are lower. The control system takes advantage of this opportunity to activate the cold storage chiller unit, operating in high-efficiency mode to lower the water temperature in the storage tank to a low level, such as 4 degrees Celsius, thus storing a significant amount of cold energy as the sensible heat of the water. During the day, especially during peak electricity prices, the control system reduces or even stops the operation of high-energy-consuming chillers, instead using the low-temperature water in the storage tank to cool the terminals. In this way, a large amount of electricity consumption is shifted from high-price periods to low-price periods. Although the total cooling capacity remains the same, the total electricity cost can be significantly reduced, resulting in significant economic benefits and contributing to peak shaving and valley filling of the power grid.

[0044] To achieve the aforementioned complex collaborative control, optimization, and multi-mode operation strategies, the method proposed in this application employs a distributed edge computing system architecture for control. This architecture includes upper-layer energy management software and a series of edge computing modules distributed across the device layer. These modules correspond to key equipment such as chillers, first-stage pumps, second-stage pumps, cooling pumps, cooling towers, and cold storage tanks. Under this architecture, the control method also includes two levels of operation: localized data processing and logic control through each edge computing module; and global energy optimization management through the collaborative work of the energy management software and the various edge computing modules.

[0045] This distributed architecture combines the advantages of centralized and decentralized control. Each edge computing module is a powerful, small controller installed close to the device it controls. It is responsible for high-speed, real-time data acquisition and closed-loop control of the devices under its jurisdiction. For example, the edge computing module corresponding to a pump unit will directly connect to the pump's frequency converter, pressure and flow sensors, and perform local control tasks such as frequency regulation with millisecond-level response speed. This design solves a large number of real-time control tasks locally on-site, greatly reducing the bandwidth requirements and latency sensitivity of the central communication network, and ensuring the stability and rapid response capability of the basic control.

[0046] The energy management software acts as the brain of the entire system. It doesn't directly participate in the instantaneous control of equipment, but rather handles higher-level, non-real-time optimization decisions requiring global information. The energy management software collects and aggregates data from all edge computing modules, performs long-term trend analysis and energy efficiency assessments, and runs complex optimization algorithms, such as the aforementioned load forecasting model and optimization logic. Its calculation results, such as the optimal operating mode and target parameters for each equipment group, are sent as high-level instructions to the corresponding edge computing modules. Upon receiving these instructions, the edge computing modules execute them as target values ​​for local control. This cloud-edge collaborative working mode ensures both real-time reliability of the underlying control and global energy optimization of the entire cooling system.

[0047] In a distributed edge computing architecture, the specific implementation of local data processing and logic control, especially in handling communication anomalies, demonstrates its advanced design and robustness. Specifically, the steps for local control via edge computing modules include: First, establishing a synchronized clock reference between the edge computing modules of the first-level and second-level pump groups. Second, generating a predictive control sequence containing the evolution trend of the operating state over a preset future time period based on the target operating parameters and linkage control parameters calculated by the upper layer. Next, synchronously distributing this predictive control sequence to the edge computing modules corresponding to the first and second-level pump groups respectively. Finally, when data packet loss or significant delay occurs in the communication links between the edge computing modules, each edge computing module can autonomously execute local control outputs based on the evolution trend in its locally stored predictive control sequence, thus maintaining traffic coordination between the first and second-level pump groups even during communication anomalies.

[0048] The core of this series of operations lies in enhancing the system's autonomous operation capability through prediction and contingency planning. Establishing a synchronized clock reference is a prerequisite for achieving coordinated actions. This can be achieved through methods such as network time protocols, ensuring that the timestamps of all edge modules are highly consistent, enabling them to execute predetermined control actions at the same time.

[0049] Generating and distributing predictive control sequences is crucial. After calculating the pump unit operating parameters for the next period, the energy management software doesn't simply issue a static setpoint, but rather a dynamic sequence of commands with a timeline. For example, the command might be: "For the first five minutes, maintain the frequency at 45 Hz; for the second five minutes, linearly increase to 48 Hz; for the third five minutes, maintain 48 Hz." This sequence is a pre-planning of the future evolution of the operating state.

[0050] The most crucial part lies in the autonomous execution logic in case of communication failures. Under normal circumstances, the edge module executes according to this sequence and is always ready to receive new sequence updates. However, if the network connection between the edge module and the energy management software is interrupted, or if data transmission encounters serious problems, the edge module will not stop working or enter an unsafe state due to the lack of new instructions. Instead, it will initiate an autonomous operation mode and continue to execute control according to the last predicted control sequence stored locally. Because the edge modules of both the first-stage and second-stage pump groups receive matching and synchronized control sequences, they can still coordinate and maintain flow balance even when the connection with the energy management software is lost. This design greatly improves the reliability and stability of the system in the face of network fluctuations or failures, avoiding the risk of the entire system going out of control or a sudden drop in energy efficiency due to communication problems.

[0051] Please refer to Figure 2 , Figure 3 This application provides a multi-stage pump collaborative variable flow control device based on edge computing, the device comprising: Parameter acquisition module 201: Acquires real-time operating parameters and environmental parameters of the second-stage pump unit in the cooling system; Load forecasting module 202: Inputs real-time operating parameters and environmental parameters into a preset load forecasting model to determine the target operating parameters of the second-stage pump set. The target operating parameters include the number of pump sets in operation and the operating frequency. Linkage calculation module 203: Based on the target operating parameters of the second-stage pump set, calculates the linkage control parameters of the first-stage pump set that match the output flow of the second-stage pump set; Coordinated adjustment module 204: Based on the target operating parameters and linkage control parameters, synchronously adjust the operating status of the first-stage pump group and the second-stage pump group so that the total flow of the first-stage pump group and the total flow of the second-stage pump group tend to be consistent.

[0052] The parameter acquisition module 201 is used to acquire real-time operating parameters and environmental parameters of the second-stage pump unit in the cooling system. Physically, this module can consist of a series of sensors, such as flow meters, temperature sensors, pressure transmitters, weather stations, and data acquisition interfaces and communication units connected to these sensors.

[0053] The load forecasting module 202 is responsible for inputting the real-time operating parameters and environmental parameters collected by the parameter acquisition module into the preset load forecasting model, and determining the target operating parameters of the second-stage pump set through calculation, including the number of pump sets to be started and the operating frequency. This module usually exists in software form and runs on an energy management software server or a dedicated edge computing server. Its core is a trained artificial intelligence algorithm or a complex physical model.

[0054] The linkage calculation module 203 calculates the linkage control parameters of the first-stage pump set that match the output flow of the second-stage pump set, based on the target operating parameters of the second-stage pump set determined by the load prediction module. This module is also implemented in software, performing optimization calculations based on pump performance curves to ensure accurate matching of supply and demand flow rates.

[0055] The coordinated adjustment module 204, based on the parameters output by the load prediction module and the linkage calculation module, synchronously adjusts the operating states of the first-stage pump group and the second-stage pump group to make their total flow rates nearly identical. This module consists of edge computing modules distributed on-site, corresponding to each pump group, and the actuators such as frequency converters they control. It is responsible for translating the calculation results into actual physical control actions. These modules work together to form a complete, intelligent coordinated control system.

[0056] For those skilled in the art, various other corresponding changes and modifications can be made based on the technical solutions and concepts described above, and all such changes and modifications should fall within the protection scope of the claims of this invention.

Claims

1. A multi-stage pump cooperative variable flow control method based on edge computing, used in a cooling system, the cooling system comprising a first-stage pump group and a second-stage pump group, characterized in that, The method includes the following steps: S1: Obtain the real-time operating parameters and environmental parameters of the second-stage pump unit in the cooling system; S2: Input the real-time operating parameters and environmental parameters into the preset load prediction model to determine the target operating parameters of the second-stage pump group. The target operating parameters include the number of pump groups that are turned on and the operating frequency. S3: Based on the target operating parameters of the second-stage pump set, calculate the linkage control parameters of the first-stage pump set that match the output flow of the second-stage pump set; S4: Based on the target operating parameters and the linkage control parameters, synchronously adjust the operating states of the first-stage pump group and the second-stage pump group so that the total flow rate of the first-stage pump group and the total flow rate of the second-stage pump group tend to be consistent.

2. The multi-stage pump collaborative variable flow control method based on edge computing according to claim 1, characterized in that, The real-time operating parameters include the flow rate, inlet and outlet water temperature, and pressure of the second-stage pump set; the environmental parameters include outdoor meteorological parameters.

3. The multi-stage pump collaborative variable flow control method based on edge computing according to claim 1, characterized in that, Step S2 includes: S21: Input the real-time operating parameters and environmental parameters into the load forecasting model to perform positive load forecasting and obtain the end-point load demand value; S22: Perform reverse reasoning calculations based on the terminal load demand value to determine the number of pumps to be started and the operating frequency of the second-stage pump set.

4. The multi-stage pump collaborative variable flow control method based on edge computing according to claim 3, characterized in that, Step S3 includes: S31: Calculate the total output flow of the second-stage pump set under the stated terminal load demand value based on the number of pumps started and the operating frequency of the second-stage pump set; S32: Based on the total output flow rate, calculate the flow balance point of the first-stage pump set under the same load demand, and use the number of pump sets turned on and the operating frequency corresponding to the flow balance point as the linkage control parameters.

5. The multi-stage pump collaborative variable flow control method based on edge computing according to claim 1, characterized in that, Step S4 includes: S41: Based on the target operating parameters and the linkage control parameters, adjust the power output of the first-stage pump group and the second-stage pump group in real time to eliminate the unbalanced flow in the bypass pipeline of the cooling system and maintain the unbalanced flow within a preset threshold range.

6. The multi-stage pump collaborative variable flow control method based on edge computing according to claim 1, characterized in that, The cooling system also includes cooling pumps and chiller units. The first-stage pump set includes a primary chiller pump. The primary chiller pump, the cooling pump, and the chiller unit are connected in a one-to-one correspondence manner so that each chiller unit can operate independently with variable flow rate. The second-stage pump set adopts a redundant design, which includes an operating pump that meets the normal operation requirements of the cooling system and at least one standby pump. The cooling system also includes a cooling tower system, which adopts a grouped parallel design to allow for grouped maintenance of each group of cooling towers without shutting down the cooling system.

7. The multi-stage pump cooperative variable flow control method based on edge computing according to claim 6, characterized in that, The chiller unit includes a cold storage constant-frequency dual-condition chiller unit and a direct chiller unit; the method also includes the following steps: S5: When the terminal cooling load exceeds the preset load threshold, or when the direct cooling unit malfunctions, control the cold storage fixed frequency dual-condition unit to connect to the direct cooling network for cooling. The cooling system further includes a cold storage tank; the method further includes: S6: During off-peak electricity price periods, control the cold storage fixed-frequency dual-condition unit to perform cold storage and store the cold energy in the cold storage tank; S7: During peak electricity price periods, release the cold energy in the cold storage tank.

8. The multi-stage pump collaborative variable flow control method based on edge computing according to claim 1, characterized in that, The cooling system is controlled through a distributed edge computing system architecture, which includes energy management software and edge computing modules corresponding to the chiller unit, the first-stage pump unit, the second-stage pump unit, the cooling pump, the cooling tower, and the cold storage tank, respectively; the method also includes the following steps: S8: Local data processing and logic control are performed through each of the aforementioned edge computing modules; S9: The energy management software works in collaboration with each edge computing module to perform global energy optimization management.

9. A multi-stage pump collaborative variable flow control method based on edge computing according to claim 8, characterized in that, Step S8 includes: S81: Establish a synchronous clock reference between the first-stage pump set and the second-stage pump set; S82: Generate a predictive control sequence containing the evolution trend of the operating state within a future preset time period based on the target operating parameters and the linkage control parameters; S83: The predictive control sequence is synchronously distributed to the edge computing modules corresponding to the first-stage pump group and the second-stage pump group; S84: When data packet loss or delay occurs in the communication link between each edge computing module, each edge computing module autonomously executes local control output according to the evolution trend in the predictive control sequence to maintain the flow coordination between the first-stage pump group and the second-stage pump group during the communication anomaly.

10. A multi-stage pump collaborative variable flow control device based on edge computing, characterized in that, The device includes: Parameter acquisition module: acquires the real-time operating parameters and environmental parameters of the second-stage pump unit in the cooling system; Load forecasting module: Input the real-time operating parameters and environmental parameters into a preset load forecasting model to determine the target operating parameters of the second-stage pump set. The target operating parameters include the number of pump sets in operation and the operating frequency. Linkage calculation module: Based on the target operating parameters of the second-stage pump set, calculates the linkage control parameters of the first-stage pump set that match the output flow of the second-stage pump set; Coordinated adjustment module: Based on the target operating parameters and the linkage control parameters, synchronously adjust the operating status of the first-stage pump group and the second-stage pump group so that the total flow rate of the first-stage pump group and the total flow rate of the second-stage pump group tend to be consistent.

Images