Water source heat pump system energy-saving optimization method and system based on data driving

Abstract

The invention relates to the technical field of heat supply management, and discloses a water source heat pump system energy-saving optimization method and system based on data driving. The method comprises the following steps: carrying out moving average filtering and normalization processing on collected cooling and heating load signals of a water source heat pump to generate standard load signals; a short-term load prediction model is constructed based on the change rate of the standard load signal, and a target load of a next period is obtained; inputting the target load into a load-flow mapping function obtained by historical efficient sample training, outputting a target flow set value, and performing constraint by using a load interval flow safety boundary associated with the function; and finally, converting the target flow into a frequency instruction in combination with a preset water pump performance curve, and issuing the frequency instruction to a water pump variable-frequency controller for execution. According to the technical scheme, closed-loop linkage of accurate load prediction and water pump frequency conversion control is achieved in a data driving mode, and the overall operation energy efficiency of the water source heat pump system is effectively improved on the premise that safe operation of system flow is guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of heating management technology, and in particular relates to a data-driven energy-saving optimization method and system for water source heat pump systems. Background Technology

[0002] Water source heat pump systems, as a highly efficient and energy-saving solution for supplying cold and heat sources, have been widely used in the field of building heating, ventilation, and air conditioning. Their energy-saving potential largely depends on the ability of key fluid transport equipment, such as water pumps, to precisely adjust flow rates according to changes in the building's terminal load, thus matching actual needs and avoiding inefficient operating conditions such as "high flow rate, small temperature difference."

[0003] Currently, the industry typically relies on simple PID control strategies based on indirect parameters such as temperature or temperature difference, or fixed, phased operation modes set based on experience, for flow control of water source heat pump systems. These traditional methods have significant limitations: on the one hand, they are slow and imprecise in responding to dynamic load changes, making it difficult to achieve real-time optimization of flow settings, resulting in high pump energy consumption; on the other hand, the control process often lacks direct constraints on the system's flow safety boundary, and operators often rely on experience or conservative settings to avoid risks such as pump cavitation and pipeline overpressure, which further limits the potential for improving system energy efficiency.

[0004] With the development of IoT and big data technologies, the collection of heat pump system operation data has become more convenient. How to effectively mine and utilize this massive amount of historical operation data, establish a data-driven precision control model, and thus achieve adaptive optimization of flow and minimization of pump energy consumption while ensuring system safety has become a key technical problem that urgently needs to be solved in the field of energy-saving optimization of water source heat pump systems. Summary of the Invention

[0005] The purpose of this invention is to provide a data-driven energy-saving optimization method and system for water source heat pump systems, in order to solve the technical problems in the prior art, such as relying on indirect parameters and empirical rules for pump frequency conversion control, lack of accurate load-flow mapping relationship, inability to achieve optimal system energy efficiency under the premise of ensuring flow safety, and lag in dynamic load response, which leads to high pump energy consumption.

[0006] To achieve the above objectives, a data-driven energy-saving optimization method for a water source heat pump system is provided in a first aspect of the present invention, comprising: Acquire target water source heat pump cooling and heating load signal data, perform moving average filtering and normalization processing on the target water source heat pump cooling and heating load signal data, and output standard load signal data; The rate of change is calculated based on the difference between adjacent sampling points in the previous time window of the standard load signal data at the current moment, and a short-term load prediction model is constructed based on the rate of change to calculate the target load data for the next control cycle. The target load data is input into a preset load-flow mapping function, and the target flow setting data is output. The load-flow mapping function is trained by load and high-efficiency flow sample pairs collected during historical operation, and is associated with flow safety boundaries divided according to load intervals. The flow safety boundaries include the minimum allowable flow and the maximum allowable flow under the corresponding load. Based on the target flow rate setting data and the preset water pump performance curve parameters, the target water source heat pump frequency command data is calculated through the mapping relationship between flow rate and frequency, and the target water source heat pump frequency command data is sent to the water pump frequency converter.

[0007] Furthermore, it also includes, The length of the sliding window is determined based on the sampling frequency of the target water source heat pump cooling and heating load signal data and the preset smoothing duration. Based on the sliding window length, a subset of data within the current window is extracted from the target water source heat pump cooling and heating load signal data; Calculate the arithmetic mean of all data points in the data subset, and use the result as the filtered load data at the current sampling time; The sliding window is moved along the time axis, and the steps of extracting a subset of data and calculating the arithmetic mean are repeated to output the load signal data.

[0008] Furthermore, the process of determining the sliding window length based on the sampling frequency of the target water source heat pump heating and cooling load signal data and the preset smoothing duration also includes: The sampling frequency for acquiring the target water source heat pump cooling and heating load signal data; Multiply the preset smoothing duration by the sampling frequency to obtain the number of data points contained in the window; The number of data points is used as the length of the sliding window.

[0009] Furthermore, the target water source heat pump cooling and heating load signal data is normalized, including: Determine the maximum and minimum values ​​in the filtered load signal data, and set the upper and lower limits of the normalization target; Based on the linear mapping relationship, several data points in the filtered load signal data are mapped to the interval defined by the target upper limit and the target lower limit to generate the standard load signal data.

[0010] Furthermore, the rate of change is calculated based on the difference between adjacent sampling points of the standard load signal data within the previous time window at the current moment, including: Extract a subset of data from the standard load signal data corresponding to the previous time window at the current moment; Calculate the difference between each pair of adjacent data points in the data subset to obtain a difference sequence; The sampling time interval for obtaining the standard load signal data; Divide each difference in the difference sequence by the sampling time interval to generate a rate of change sequence.

[0011] Furthermore, it also includes: A short-term load forecasting model is constructed based on this rate of change to calculate the target load data for the next control period, including: Linear regression analysis was performed based on the rate of change sequence to obtain the slope parameter; The slope parameter is added to the standard load signal data at the current time to obtain the predicted load data; The predicted load data is determined as the target load data for the next control cycle.

[0012] Furthermore, it also includes: Obtain the historical operating load of the target water source heat pump, determine the load range, and divide the load range into several consecutive load intervals; For the load range, the minimum and maximum allowable flow rates corresponding to the load range are read from the preset configuration file; Establish a mapping table containing all load intervals and their corresponding minimum and maximum allowable flow values, the mapping table constituting the flow safety boundary; When the load-flow mapping function performs mapping calculations, the target load data is matched with the load intervals in the mapping table. If the target load data falls within a certain load interval, the minimum allowable flow value and the maximum allowable flow value corresponding to that interval are used as the flow safety constraint boundaries under the target load data.

[0013] Furthermore, based on the target flow rate setting data and the pre-stored water pump performance curve parameters, the target water source heat pump frequency command data is calculated, including: The pump performance curve parameters are called, which define the relationship between the pump operating frequency, flow rate and head; Using the target flow rate setting data as the target flow rate input, and combining it with the currently required head value, the corresponding relationship is queried to obtain the pump operating frequency; The operating frequency of the water pump is used as the frequency command data of the target water source heat pump.

[0014] Furthermore, sending the target water source heat pump frequency command data to the water pump frequency converter includes: A data communication link is established with the water pump frequency converter through a preset communication protocol; The target water source heat pump frequency command data is encoded into a command data packet conforming to the communication protocol format; The instruction data packet is transmitted to the water pump frequency converter through the data communication link.

[0015] Secondly, the present invention also provides a data-driven energy-saving optimization system for a water source heat pump system, which applies any of the data-driven energy-saving optimization methods for water source heat pump systems, including: The data preprocessing module is used to acquire target water source heat pump cooling and heating load signal data, perform moving average filtering and normalization processing on the target water source heat pump cooling and heating load signal data, and output standard load signal data. The load forecasting module is used to calculate the rate of change of the standard load signal data based on the difference between adjacent sampling points in the previous time window at the current moment, and to construct a short-term load forecasting model based on the rate of change to calculate the target load data for the next control cycle. The load-flow mapping module is used to input the target load data into a preset load-flow mapping function and output target flow setting data. The load-flow mapping function is trained by load and high-efficiency flow sample pairs collected during historical operation and is associated with flow safety boundaries divided according to load intervals. The flow safety boundaries include the minimum allowable flow and the maximum allowable flow under the corresponding load. The instruction output module is used to calculate the target water source heat pump frequency instruction data based on the target flow rate setting data and preset water pump performance curve parameters, through the mapping relationship between flow rate and frequency, and send the target water source heat pump frequency instruction data to the water pump frequency converter.

[0016] The beneficial technical effects of the present invention are at least as follows: The data-driven energy-saving optimization method and system for water source heat pump systems provided by this invention have the following outstanding advantages: Firstly, by applying moving average filtering and normalization to the acquired hot and cold load signals, high-frequency noise and dimensional differences in the original data can be effectively eliminated, generating smooth and standard load signals and improving data quality. Based on this, a short-term load prediction model is constructed by calculating the rate of change obtained from the difference between adjacent sampling points. This enables accurate prediction of the load trend in the next control cycle, freeing the pump flow setting from the limitations of relying on hysteresis temperature parameters or empirical rules. It achieves feedforward optimization control based on real-time load prediction, significantly improving the system's response speed and control foresight to dynamic loads.

[0017] Secondly, by introducing a load-flow mapping function trained from historical high-efficiency operating data and associating it with flow safety boundaries (including minimum and maximum allowable flow rates) divided according to load intervals, the flow setting not only pursues optimal energy efficiency but also strictly ensures that the pump operates within its hydraulic safety range. This method transforms traditional empirical safety margin control into data-driven precise boundary constraints, preventing operational risks such as pipeline overpressure or pump cavitation while avoiding energy waste caused by overly conservative settings. This mechanism realizes the transformation from "empirical safety control" to "data-driven safety optimization," maximizing the energy-saving potential of the pump while ensuring the long-term safe and reliable operation of the system.

[0018] Third, a complete closed-loop control chain is established, from load forecasting to flow setting, then to generating frequency commands based on pump performance curves, and finally issuing and executing them. The entire control process is entirely data-driven, and the core mapping function and performance curves support online updates and calibrations based on actual operating data, enabling the control strategy to adapt to performance degradation or equipment replacement in the water source heat pump system. The system possesses continuous self-improvement capabilities from "sensing, optimizing, and executing" to "data feedback and model updating," ensuring control effectiveness, adaptability, and engineering robustness under long-term operation, and providing a reliable technical solution for achieving intelligent energy-saving operation of the water source heat pump system throughout its entire lifecycle. Attached Figure Description

[0019] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the working steps of a data-driven energy-saving optimization method for a water source heat pump system, as disclosed in one embodiment of the present invention. Figure 2 This is a diagram illustrating the specific working steps for outputting standard load signal data according to one embodiment of the present invention; Figure 3This is a schematic diagram of a data-driven energy-saving optimization device for a water source heat pump system, as disclosed in one embodiment of the present invention. Detailed Implementation

[0021] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0022] Example 1 refer to Figures 1-2 This invention provides an embodiment of a data-driven energy-saving optimization method for a water source heat pump system, comprising: S1. Obtain target water source heat pump cooling and heating load signal data, perform moving average filtering and normalization processing on the target water source heat pump cooling and heating load signal data, and output standard load signal data. S2. Calculate the rate of change based on the difference between adjacent sampling points in the previous time window of the standard load signal data at the current moment, and construct a short-term load prediction model based on the rate of change to calculate the target load data for the next control cycle. S3. Input the target load data into a preset load-flow mapping function and output the target flow setting data. The load-flow mapping function is trained by load and high-efficiency flow sample pairs collected during historical operation and is associated with flow safety boundaries divided according to load intervals. The flow safety boundaries include the minimum allowable flow and the maximum allowable flow under the corresponding load. S4. Based on the target flow rate setting data and the preset water pump performance curve parameters, calculate the target water source heat pump frequency command data through the mapping relationship between flow rate and frequency, and send the target water source heat pump frequency command data to the water pump frequency converter.

[0023] In this embodiment, the target water source heat pump system uses pre-set load sensors (such as a combination of temperature, flow, and power sensors) to collect real-time cooling and heating load signal data. The collected data includes load data from both the evaporator and condenser sides of the water source heat pump. The collection frequency is set to once per minute, and the collection duration covers the entire system operating cycle (e.g., 24 hours) to ensure that the collected data comprehensively reflects the dynamic changes in the system load. The collected target water source heat pump cooling and heating load signal data may contain abnormal fluctuations caused by sensor errors and external interference (such as power grid fluctuations or sudden changes in ambient temperature). Therefore, it is necessary to perform moving average filtering to eliminate data noise and improve data accuracy.

[0024] After filtering, the filtered load signal data is normalized. Since the range of values ​​of the collected load signal data may vary greatly (e.g., the difference between winter heating load and summer cooling load values ​​is large), normalization can map all load data to a uniform value range (preferably [0,1] in this embodiment), avoiding the impact of large numerical differences on the accuracy of subsequent model prediction and mapping calculation, and outputting standardized load signal data.

[0025] Based on step S1, further steps include: S11. Determine the sliding window length based on the sampling frequency of the target water source heat pump cooling and heating load signal data and the preset smoothing duration; S12. Based on the length of the sliding window, extract a subset of data within the current window from the target water source heat pump cooling and heating load signal data; S13. Calculate the arithmetic mean of all data points in the data subset, and use the calculation result as the filtered load data at the current sampling time; S14. Move the sliding window along the time axis, repeat the steps of extracting a subset of data and calculating the arithmetic mean, and output the load signal data.

[0026] In this embodiment, based on the standard load signal data output by S1, a time window length is set (preferably 30 minutes in this embodiment, corresponding to 30 sampling points, matching the sampling frequency of 1 time / minute), and the standard load signal data within the previous time window at the current moment is extracted as the analysis sample. The difference between adjacent sampling points in this sample is calculated, that is, the standard load data of the next sampling point minus the standard load data of the previous sampling point, to obtain the load difference between adjacent sampling points. Then, this difference is divided by the sampling time interval (1 minute in this embodiment) to obtain the load change rate between each adjacent sampling point, thereby generating a load change rate sequence.

[0027] Based on the generated load change rate sequence, a short-term load forecasting model is constructed (in this embodiment, a linear regression model is preferred due to its simple structure, high computational efficiency, and suitability for rapid forecasting in real-time control scenarios). By fitting and analyzing the change rate sequence, model parameters are determined, and this model is then used to calculate the target load data for the next control cycle (preferably 5 minutes in this embodiment). This forecasting method, based on historical load change patterns, can quickly capture dynamic load trends, ensuring a high degree of match between the predicted target load data and actual load demand, thus providing an accurate basis for subsequent flow rate settings.

[0028] Furthermore, the process of determining the sliding window length based on the sampling frequency of the target water source heat pump heating and cooling load signal data and the preset smoothing duration also includes: The sampling frequency for acquiring the target water source heat pump cooling and heating load signal data; Multiply the preset smoothing duration by the sampling frequency to obtain the number of data points contained in the window; The number of data points is used as the length of the sliding window.

[0029] Furthermore, the target water source heat pump cooling and heating load signal data is normalized, including: Determine the maximum and minimum values ​​in the filtered load signal data, and set the upper and lower limits of the normalization target; Based on the linear mapping relationship, several data points in the filtered load signal data are mapped to the interval defined by the target upper limit and the target lower limit to generate the standard load signal data.

[0030] Furthermore, the rate of change is calculated based on the difference between adjacent sampling points of the standard load signal data within the previous time window at the current moment, including: Extract a subset of data from the standard load signal data corresponding to the previous time window at the current moment; Calculate the difference between each pair of adjacent data points in the data subset to obtain a difference sequence; The sampling time interval for obtaining the standard load signal data; Divide each difference in the difference sequence by the sampling time interval to generate a rate of change sequence.

[0031] Based on step S2, the change rate is calculated by differentiating adjacent sampling points of the standard load signal data within the previous time window at the current moment, and a short-term load forecasting model is constructed based on this change rate to calculate the target load data for the next control cycle. The method also includes the following steps: The sliding window length is determined by considering the sampling frequency of the target water source heat pump's heating and cooling load signal data and the preset smoothing duration. This ensures that the filtering effect can effectively eliminate data noise without losing the dynamic characteristics of the load data. In this embodiment, the sampling frequency of the load signal data is 1 time / minute, and the preset smoothing duration is 5 minutes (i.e., smoothing the load fluctuation at the current moment by averaging the load data over 5 minutes). The sliding window length is determined accordingly.

[0032] Extract a subset of data within the current window. Based on the determined sliding window length, extract load data (a total of 5 data points) from the real-time collected target water source heat pump heating and cooling load signal data, including the current sampling time and the previous 4 sampling times. This subset forms the data within the current window. For example, if the current sampling time is minute t, extract the 5 load data points at minutes t-4, t-3, t-2, t-1, and t as the data subset of the current window.

[0033] The filtered load data of the current window is calculated using the arithmetic mean method. The arithmetic mean of all 5 data points in the data subset within the current window is calculated using the following formula: Filtered load data = (sum of all data point values ​​in the data subset) / number of data points (5 in this embodiment). The calculated arithmetic mean is used as the filtered load data at the current sampling time (minute t).

[0034] The filtering process is executed iteratively, moving the sliding window along the time axis by one sampling time point (i.e., one minute) each time. Steps two and three are repeated: after each movement, a subset of data within the new window is extracted (e.g., five data points from minute t-3 to minute t+1), and the arithmetic mean of this subset is calculated as the filtered load data for minute t+1. This process is repeated until all collected load signal data has been filtered, and finally, the filtered load signal data is output for subsequent normalization processing.

[0035] Furthermore, it also includes: Based on this rate of change, a short-term load forecasting model is constructed to calculate the target load data for the next control cycle, including: Linear regression analysis was performed based on the rate of change sequence to obtain the slope parameter; The slope parameter is added to the standard load signal data at the current time to obtain the predicted load data; The predicted load data is determined as the target load data for the next control cycle.

[0036] Preferably, in some embodiments, Furthermore, it also includes: Obtain the historical operating load of the target water source heat pump, determine the load range, and divide the load range into several consecutive load intervals; For the load range, the minimum and maximum allowable flow rates corresponding to the load range are read from the preset configuration file; Establish a mapping table containing all load intervals and their corresponding minimum and maximum allowable flow values, the mapping table constituting the flow safety boundary; When the load-flow mapping function performs mapping calculations, the target load data is matched with the load intervals in the mapping table. If the target load data falls within a certain load interval, the minimum allowable flow value and the maximum allowable flow value corresponding to that interval are used as the flow safety constraint boundaries under the target load data.

[0037] A load-flow mapping function is pre-constructed. The construction process is as follows: Load data and high-efficiency flow data under corresponding operating conditions are collected from the historical operation period of the water source heat pump system (covering at least one complete cooling season and one complete heating season) to form several load-high-efficiency flow sample pairs; the sample pairs are cleaned to remove abnormal samples (such as samples under equipment failure conditions or samples with flow rates exceeding the safe range); machine learning algorithms (support vector regression algorithm is preferred in this embodiment) are used to train the cleaned sample pairs to obtain the initial load-flow mapping function; subsequently, in combination with the system operation safety requirements, flow safety boundaries are divided according to the load range, and the flow safety boundaries are associated with the initial mapping function to complete the construction of the load-flow mapping function.

[0038] The flow safety boundary is divided according to load intervals. Each load interval corresponds to a unique minimum and maximum allowable flow rate. The minimum allowable flow rate ensures that the water source heat pump evaporator and condenser will not experience scaling, freezing, or cracking due to insufficient flow. The maximum allowable flow rate prevents excessive pump energy consumption and system pressure exceeding limits due to excessive flow. The target load data calculated in step 2 is input into the load-flow mapping function. The function adjusts the constraints based on the load interval where the target load data is located, combined with the flow safety boundary, and outputs target flow setting data that meets the requirements for efficient and safe system operation.

[0039] Furthermore, based on the target flow rate setting data and the pre-stored water pump performance curve parameters, the target water source heat pump frequency command data is calculated, including: The pump performance curve parameters are called, which define the relationship between the pump operating frequency, flow rate and head; Using the target flow rate setting data as the target flow rate input, and combining it with the currently required head value, the corresponding relationship is queried to obtain the pump operating frequency; The operating frequency of the water pump is used as the frequency command data of the target water source heat pump.

[0040] In this embodiment, linear regression analysis is performed on the rate of change sequence. Using the generated load change rate sequence as the analysis object, a linear regression algorithm is employed for fitting analysis to construct a linear regression model. The expression of the linear regression model is: r = k × x + b, where x is the sampling point number, r is the load change rate, k is the slope parameter (reflecting the overall trend of the change rate), and b is the intercept parameter. The rate of change sequence is fitted using the least squares method to calculate the slope parameter k and the intercept parameter b. In this embodiment, the slope parameter k obtained after fitting is preferably 0.001-0.005 to ensure that the model can accurately capture the trend of the load change rate.

[0041] The second step is to determine the final value of the slope parameter. Since the intercept parameter b mainly reflects the influence of random errors and has a relatively small impact on load forecasting, the slope parameter k is mainly used as the basis for judging the load change trend in subsequent forecasts. The slope parameter k obtained by fitting is used as the final trend parameter. If the slope parameter k is positive, it indicates that the load is on an upward trend; if it is negative, it indicates that the load is on a downward trend; if it is close to 0, it indicates that the load is stabilizing.

[0042] The third step is to calculate the predicted load data. The slope parameter k is added to the standard load signal data L_t' at the current time to obtain the predicted load data for the next control cycle. The calculation formula is: Among them, the standard load signal data at the current moment. This refers to the data point at the current sampling time in the output standard load signal sequence; This is the slope parameter obtained from linear regression fitting. For example, the standard load data at the current time. slope parameter Then predict the load data (The value after normalization).

[0043] The fourth step is to determine the target load data. This involves using the calculated predicted load data... The target load data for the next control cycle is determined. In this embodiment, the next control cycle is set to 5 minutes, which matches the response speed of the pump frequency converter control, ensuring that the predicted target load data can promptly guide subsequent flow rate settings and frequency adjustments. If the predicted load data... If the load exceeds the normalized target range [0,1], it will be subjected to a limiting process, which will correct it to 0 (if less than 0) or 1 (if greater than 1) to ensure the rationality of the target load data and provide accurate input for subsequent load-flow mapping.

[0044] Furthermore, sending the target water source heat pump frequency command data to the water pump frequency converter includes: A data communication link is established with the water pump frequency converter through a preset communication protocol; The target water source heat pump frequency command data is encoded into a command data packet conforming to the communication protocol format; The instruction data packet is transmitted to the water pump frequency converter through the data communication link.

[0045] In this embodiment, a data communication link needs to be established first. The system pre-sets the communication protocol (preferably Modbus-RTU communication protocol in this embodiment, which features stable transmission, strong anti-interference capability, and suitability for industrial control scenarios). A bidirectional data communication link is established with the water pump frequency converter through a communication module in the system (such as an RS485 communication module). Before establishing the communication link, the communication parameters are configured, including baud rate (9600bps in this embodiment), data bits (8 bits), stop bits (1 bit), and parity bit (no parity), to ensure that the communication parameters of the system communication module and the water pump frequency converter are consistent, thus avoiding communication failure.

[0046] The obtained target water source heat pump frequency command data (e.g., 42Hz) is encoded according to the preset Modbus-RTU communication protocol format to generate a command data packet. The format of the command data packet includes: slave address (the unique address of the water pump frequency converter, set to 0x01 in this embodiment), function code (the function code used to control the water pump frequency, 0x06 in this embodiment), register address (the register address storing the frequency command, 0x0001 in this embodiment), frequency command data (42Hz converted to hexadecimal data, i.e., 0x002A), and checksum (used to verify the integrity of the data packet and avoid errors during data transmission).

[0047] The encoded instruction data packets are sent to the pump frequency converter via the established data communication link. During data transmission, the system monitors the communication status in real time. If a data packet transmission failure is detected (such as a checksum mismatch or communication timeout), the instruction data packet is resent, up to a maximum of three times. If all three attempts fail, a communication fault alarm signal is issued, reminding staff to check the communication line, the status of the pump frequency converter, etc., to ensure successful transmission of the instruction data packets.

[0048] Upon receiving the instruction data packet, the pump frequency converter decodes it, extracts the target frequency instruction data, and adjusts the pump's operating frequency accordingly. After adjustment, the pump frequency converter sends feedback to the system (e.g., frequency adjusted to 42Hz, normal operation). The system receives the feedback and completes the frequency instruction sending process. If the feedback indicates adjustment failure, the above steps are repeated to ensure the pump's operating frequency meets the target requirements.

[0049] refer to Figure 3 Secondly, the present invention also provides a data-driven energy-saving optimization system for a water source heat pump system, which applies any of the data-driven energy-saving optimization methods for water source heat pump systems, including: The data preprocessing module is used to acquire target water source heat pump cooling and heating load signal data, perform moving average filtering and normalization processing on the target water source heat pump cooling and heating load signal data, and output standard load signal data. The load forecasting module is used to calculate the rate of change of the standard load signal data based on the difference between adjacent sampling points in the previous time window at the current moment, and to construct a short-term load forecasting model based on the rate of change to calculate the target load data for the next control cycle. The load-flow mapping module is used to input the target load data into a preset load-flow mapping function and output target flow setting data. The load-flow mapping function is trained by load and high-efficiency flow sample pairs collected during historical operation and is associated with flow safety boundaries divided according to load intervals. The flow safety boundaries include the minimum allowable flow and the maximum allowable flow under the corresponding load. The instruction output module is used to calculate the target water source heat pump frequency instruction data based on the target flow rate setting data and preset water pump performance curve parameters, through the mapping relationship between flow rate and frequency, and send the target water source heat pump frequency instruction data to the water pump frequency converter.

[0050] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the data-driven energy-saving optimization method for a water source heat pump system as described in any one of the claims.

[0051] This embodiment provides a data-driven energy-saving optimization method and system for a water source heat pump system, which has the following beneficial effects: This embodiment provides a data-driven energy-saving optimization method and system for a water source heat pump system. The technical solution provided by this invention significantly improves the energy-saving effect and control intelligence level of the water source heat pump system through a data-driven approach.

[0052] Specifically, the raw signals of the collected heating and cooling loads were first preprocessed: a moving average filtering algorithm was used to remove high-frequency interference components from the signals, and then normalization was used to eliminate the influence of different dimensions, ultimately resulting in a standardized load signal sequence. This process improved the data quality and laid the foundation for subsequent analysis.

[0053] In the load forecasting stage, based on the standardized load signal, the instantaneous load change trend is captured by calculating the differential rate of change between adjacent sampling points. The short-term forecasting model built using this rate of change can predict the direction and magnitude of load changes one control cycle in advance, providing forward-looking guidance for pump flow regulation. This forecasting method based on real-time load change rate has a faster response characteristic compared to traditional control strategies that rely on temperature feedback.

[0054] In the flow optimization setting phase, the system calls a pre-trained load-flow mapping function. This function is trained using high-efficiency operating data collected from historical operations, reflecting the correspondence between load and flow under optimal energy efficiency conditions. Simultaneously, minimum and maximum allowable flow rates are set for each load interval, forming flow safety boundary constraints. When predicting load input, the system first obtains the theoretically optimal flow rate through the mapping function, then constrains it within the safety boundaries of the corresponding load interval, ultimately outputting a target flow rate setting that ensures both energy efficiency and safety.

[0055] In the control command generation stage, based on the obtained target flow rate setpoint and the pre-stored pump performance curve parameters, the corresponding pump operating frequency command is calculated using the flow rate-frequency correspondence. This frequency command is then sent to the pump frequency converter for execution via a standard communication protocol, achieving precise adjustment of the pump speed.

[0056] The entire control process forms a data-driven closed loop: load data is processed and predicted to guide flow optimization settings, and then converted into frequency commands to control pump operation. Key parameters in the system (such as the load-flow mapping function) can be updated online based on actual operating data, enabling the control strategy to adapt to changes in system performance. This design ensures that the system maintains an optimized operating state throughout its entire lifecycle, maximizing energy savings while ensuring safety.

[0057] In the description of this specification, the references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0058] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0059] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A data-driven energy-saving optimization method for a water source heat pump system, characterized in that, include: Acquire target water source heat pump cooling and heating load signal data, perform moving average filtering and normalization processing on the target water source heat pump cooling and heating load signal data, and output standard load signal data; The rate of change is calculated based on the difference between adjacent sampling points in the previous time window of the standard load signal data at the current moment, and a short-term load prediction model is constructed based on the rate of change to calculate the target load data for the next control cycle. The target load data is input into a preset load-flow mapping function, and the target flow setting data is output. The load-flow mapping function is trained by load and high-efficiency flow sample pairs collected during historical operation, and is associated with flow safety boundaries divided according to load intervals. The flow safety boundaries include the minimum allowable flow and the maximum allowable flow under the corresponding load. Based on the target flow rate setting data and the preset water pump performance curve parameters, the target water source heat pump frequency command data is calculated through the mapping relationship between flow rate and frequency, and the target water source heat pump frequency command data is sent to the water pump frequency converter.

2. The data-driven energy-saving optimization method for a water source heat pump system according to claim 1, characterized in that, It also includes, The length of the sliding window is determined based on the sampling frequency of the target water source heat pump cooling and heating load signal data and the preset smoothing duration. Based on the sliding window length, a subset of data within the current window is extracted from the target water source heat pump cooling and heating load signal data; Calculate the arithmetic mean of all data points in the data subset, and use the result as the filtered load data at the current sampling time; The sliding window is moved along the time axis, and the steps of extracting a subset of data and calculating the arithmetic mean are repeated to output the load signal data.

3. The data-driven energy-saving optimization method for a water source heat pump system according to claim 2, characterized in that, The process of determining the sliding window length based on the sampling frequency of the target water source heat pump heating and cooling load signal data and the preset smoothing duration also includes: The sampling frequency for acquiring the target water source heat pump cooling and heating load signal data; Multiply the preset smoothing duration by the sampling frequency to obtain the number of data points contained in the window; The number of data points is used as the length of the sliding window.

4. The data-driven energy-saving optimization method for a water source heat pump system according to claim 1, characterized in that, The target water source heat pump cooling and heating load signal data is normalized, including: Determine the maximum and minimum values ​​in the filtered load signal data, and set the upper and lower limits of the normalization target; Based on the linear mapping relationship, several data points in the filtered load signal data are mapped to the interval defined by the target upper limit and the target lower limit to generate the standard load signal data.

5. The data-driven energy-saving optimization method for a water source heat pump system according to claim 1, characterized in that, The rate of change is calculated based on the difference between adjacent sampling points of the standard load signal data within the previous time window at the current moment, including: Extract a subset of data from the standard load signal data corresponding to the previous time window at the current moment; Calculate the difference between each pair of adjacent data points in the data subset to obtain a difference sequence; The sampling time interval for obtaining the standard load signal data; Divide each difference in the difference sequence by the sampling time interval to generate a rate of change sequence.

6. The data-driven energy-saving optimization method for a water source heat pump system according to claim 5, characterized in that, Also includes: Based on this rate of change, a short-term load forecasting model is constructed to calculate the target load data for the next control cycle, including: Linear regression analysis was performed based on the rate of change sequence to obtain the slope parameter; The slope parameter is added to the standard load signal data at the current time to obtain the predicted load data; The predicted load data is determined as the target load data for the next control cycle.

7. The data-driven energy-saving optimization method for a water source heat pump system according to claim 1, characterized in that, Also includes: Obtain the historical operating load of the target water source heat pump, determine the load range, and divide the load range into several consecutive load intervals; For the load range, the minimum and maximum allowable flow rates corresponding to the load range are read from the preset configuration file; Establish a mapping table containing all load intervals and their corresponding minimum and maximum allowable flow values, the mapping table constituting the flow safety boundary; When the load-flow mapping function performs mapping calculations, the target load data is matched with the load intervals in the mapping table. If the target load data falls within a certain load interval, the minimum allowable flow value and the maximum allowable flow value corresponding to that interval are used as the flow safety constraint boundaries under the target load data.

8. The data-driven energy-saving optimization method for a water source heat pump system according to claim 1, characterized in that, Based on the target flow rate setting data and the pre-stored water pump performance curve parameters, the target water source heat pump frequency command data is calculated, including: The pump performance curve parameters are called, which define the relationship between the pump operating frequency, flow rate and head; Using the target flow rate setting data as the target flow rate input, and combining it with the currently required head value, the corresponding relationship is queried to obtain the pump operating frequency; The operating frequency of the water pump is used as the frequency command data of the target water source heat pump.

9. The data-driven energy-saving optimization method for a water source heat pump system according to claim 1, characterized in that, Sending the target water source heat pump frequency command data to the water pump frequency converter includes: A data communication link with the water pump frequency converter is established through a preset communication protocol; The target water source heat pump frequency command data is encoded into a command data packet conforming to the communication protocol format; The instruction data packet is transmitted to the water pump frequency converter through the data communication link.

10. A data-driven energy-saving optimization system for a water source heat pump system, employing the data-driven energy-saving optimization method for a water source heat pump system as described in any one of claims 1-9, characterized in that, include: The data preprocessing module is used to acquire target water source heat pump cooling and heating load signal data, perform moving average filtering and normalization processing on the target water source heat pump cooling and heating load signal data, and output standard load signal data. The load forecasting module is used to calculate the rate of change of the standard load signal data based on the difference between adjacent sampling points in the previous time window at the current moment, and to construct a short-term load forecasting model based on the rate of change to calculate the target load data for the next control cycle. The load-flow mapping module is used to input the target load data into a preset load-flow mapping function and output target flow setting data. The load-flow mapping function is trained by load and high-efficiency flow sample pairs collected during historical operation and is associated with flow safety boundaries divided according to load intervals. The flow safety boundaries include the minimum allowable flow and the maximum allowable flow under the corresponding load. The instruction output module is used to calculate the target water source heat pump frequency instruction data based on the target flow rate setting data and preset water pump performance curve parameters, through the mapping relationship between flow rate and frequency, and send the target water source heat pump frequency instruction data to the water pump frequency converter.

Images