Energy-saving control method for cooling tower of central air conditioner

Abstract

The invention provides an energy-saving control method for a central air conditioner cooling tower, which comprises the following steps: when detecting that the current wet bulb temperature deviates from an expected environment temperature curve, acquiring a corrected environment temperature curve corresponding to the expected environment temperature curve; a current control chart of the cooling tower and a real-time power running chart of the central air conditioning system are obtained, the current control chart comprises a main control cooling device and a plurality of controlled cooling devices which are arranged in a topological mode, the main control cooling device is in communication connection with the controlled cooling devices, and the real-time power running chart comprises a plurality of energy consumption assemblies; connecting edges between the energy consumption assemblies, wherein the connecting edges are in an energy consumption transfer relationship between the energy consumption assemblies; adjusting the current control chart based on the corrected environment temperature curve and the real-time power running chart to obtain a target control chart; and controlling a controlled cooling device of the cooling tower through the target control chart. The operation efficiency of the cooling equipment group is improved, and the redundant energy consumption of the cooling tower is reduced.

Description

Technical Field

[0001] The present invention relates to the field of central air-conditioning automatic control, and particularly to an energy-saving control method for a central air-conditioning cooling tower. Background Art

[0002] As a key device for building energy consumption, the operation efficiency of a central air-conditioning cooling tower directly affects the energy consumption of the overall system. Traditional control methods for central air-conditioning cooling towers usually adjust equipment based on fixed environmental parameters or preset temperature curves. However, in actual operation, the energy consumption nodes of the system and the dynamic fluctuations of the environmental temperature will cause the operating state of the system to deviate from the preset model. The existing technology lacks a dynamic correction mechanism for the environmental temperature curve, and there is insufficient coordination between equipment topology control and real-time energy consumption transfer relationships, which easily leads to a decrease in the operating efficiency of the cooling equipment group and energy waste. Summary of the Invention

[0003] An embodiment of the present invention provides an energy-saving control method for a central air-conditioning cooling tower, aiming to solve the problem that the energy consumption cannot be accurately controlled during the control process of the existing central air-conditioning cooling tower. By establishing a coupled analysis model of a dynamically corrected environmental temperature curve and a real-time energy consumption transfer topology, the cooling tower equipment group can adaptively adjust the control strategy according to the environmental parameter offset, solve the problem of energy efficiency imbalance caused by environmental temperature fluctuations in traditional methods, improve the operating efficiency of the cooling equipment group, and reduce the redundant energy consumption of the cooling tower.

[0004] In a first aspect, an embodiment of the present invention provides an energy-saving control method for a central air-conditioning cooling tower, the method comprising the following steps: When it is detected that the current wet-bulb temperature deviates from the expected environmental temperature curve, obtain the corrected environmental temperature curve corresponding to the expected environmental temperature curve; and obtain the current control diagram of the cooling tower and the real-time power operation diagram of the central air-conditioning system, the current control diagram includes a main controlled cooling device and a plurality of controlled cooling devices set by topology, the main controlled cooling device is communicatively connected to the controlled cooling devices, the real-time power operation diagram includes a plurality of energy consumption components, and connection edges between the energy consumption components, the connection edges are the energy consumption transfer relationships between the energy consumption components; Based on the corrected environmental temperature curve and the real-time power operation diagram, adjust the current control diagram to obtain a target control diagram; Control the controlled cooling devices of the cooling tower through the target control diagram.

[0005] Optionally, before the step of obtaining the corrected environmental temperature curve corresponding to the expected environmental temperature curve when it is detected that the current wet-bulb temperature deviates from the expected environmental temperature curve, the method further includes: Determine the expected environmental temperature curve at the location of the cooling tower according to the weather data of the day; Collect the current wet-bulb temperature; When the current wet-bulb temperature deviates from the expected ambient temperature curve by more than a preset deviation value, perform temperature curve prediction based on the historical wet-bulb temperature of the day to obtain a predicted wet-bulb temperature curve; Calculate the correlation between the predicted wet-bulb temperature curve and the expected ambient temperature curve, and when the correlation is less than the correlation threshold, determine that the wet-bulb temperature deviates from the expected ambient temperature curve.

[0006] Optionally, the step of determining the expected ambient temperature curve at the location of the cooling tower according to the weather data of the day specifically includes: Obtain the multi-day historical wet-bulb temperature curve at the location of the cooling tower, and obtain the multi-day historical weather data corresponding to the multi-day historical wet-bulb temperature curve in terms of date; Determine a plurality of representative weather data based on the multi-day historical weather data, and determine a plurality of representative wet-bulb temperature curves according to the multi-day historical wet-bulb temperature curve, and the representative weather data and the representative wet-bulb temperature curves correspond one by one; Associate the representative weather data with the representative wet-bulb temperature curves; According to the similarity between the weather data of the day and the representative weather data, determine the representative wet-bulb temperature curve corresponding to the representative weather data with the highest similarity as the expected ambient temperature curve at the location of the cooling tower.

[0007] Optionally, the step of determining a plurality of representative weather data based on the multi-day historical weather data and determining a plurality of representative wet-bulb temperature curves according to the multi-day historical wet-bulb temperature curve specifically includes: Cluster the multi-day historical weather data to obtain historical weather data clustering clusters, and determine the clustering center of the weather data clustering cluster as the representative weather data corresponding to the weather data clustering cluster, and each historical weather data clustering cluster includes at least one day of historical weather data; Based on the corresponding relationship between the multi-day historical wet-bulb temperature curve and the multi-day historical weather data in terms of the date, determine historical wet-bulb temperature curve clustering clusters in the multi-day historical wet-bulb temperature curve, and determine the clustering center of the historical wet-bulb temperature curve clustering cluster as the representative wet-bulb temperature curve corresponding to the historical wet-bulb temperature curve clustering cluster, and each historical wet-bulb temperature curve clustering cluster includes at least one day of historical wet-bulb temperature curve, and the historical wet-bulb temperature curve clustering clusters and the weather data clustering clusters correspond one by one.

[0008] Optionally, the step of obtaining the corrected ambient temperature curve corresponding to the expected ambient temperature curve specifically includes: Obtain the deviation value between the current wet-bulb temperature and the expected ambient temperature curve, and obtain the fluctuation range of the wet-bulb temperature in the same period of history; Based on the deviation value and the fluctuation range of the wet-bulb temperature in the same period of history, construct a compensation coefficient matrix; Based on the compensation coefficient matrix, perform compensation calculation on the expected ambient temperature curve to obtain a corrected ambient temperature curve corresponding to the expected ambient temperature curve.

[0009] Optionally, the step of performing compensation calculation on the expected ambient temperature curve based on the compensation coefficient matrix to obtain a corrected ambient temperature curve corresponding to the expected ambient temperature curve specifically includes: Determine the predicted value of the wet-bulb temperature according to the predicted wet-bulb temperature curve, and determine the first confidence level corresponding to the predicted value of the wet-bulb temperature according to the compensation coefficient matrix; Collect the average measured wet-bulb temperature within a preset radius around the location where the cooling tower is located, and determine the second confidence level corresponding to the average measured wet-bulb temperature according to the health status of the sensor; Calculate the heat reflux influence amount according to the air outlet wind speed and exhaust gas temperature of the cooling tower, and determine the third confidence level corresponding to the heat reflux influence amount according to the compensation coefficient matrix; Based on the predicted value of the wet-bulb temperature and the corresponding first confidence level, the measured wet-bulb temperature and the corresponding second confidence level, the heat reflux influence amount and the corresponding third confidence level, determine the corrected wet-bulb temperature; Based on the corrected wet-bulb temperature, process the expected ambient temperature curve and perform compensation calculation to obtain a corrected ambient temperature curve corresponding to the expected ambient temperature curve.

[0010] Optionally, the step of obtaining the current control diagram of the cooling tower and the real-time power operation diagram of the central air-conditioning system specifically includes: Determine the main control relationship between the main control cooling equipment and the controlled cooling equipment, and according to the main control relationship, determine the current control diagram of the cooling tower; Based on the energy consumption component list of the central air-conditioning system, map each energy consumption component to a node in the graph structure, and generate connection edges between upstream and downstream nodes to obtain the real-time power operation diagram of the central air-conditioning system.

[0011] Optionally, the main control relationship between the main control cooling equipment and the controlled cooling equipment of the cooling tower is a dynamically changing relationship. The step of determining the main control relationship between the main control cooling equipment and the controlled cooling equipment and determining the current control diagram of the cooling tower according to the main control relationship specifically includes: Poll the device nodes in the cooling tower control network through a preset communication protocol, and determine the physical address and communication port configuration of the main control cooling device based on the control authority identifier in the device response message. Analyze the device attribute parameters of each controlled cooling device, where the device attribute parameters include device ID, rated power threshold, and real-time operating frequency. Map the main control cooling device and the controlled cooling devices to a topology diagram to obtain the current control diagram of the cooling tower.

[0012] Optionally, the step of mapping each energy consumption component to a node in the graph structure and generating connection edges between upstream and downstream nodes based on the energy consumption component list of the central air-conditioning system to obtain the real-time power operation diagram of the central air-conditioning system specifically includes: Based on the energy consumption component list of the central air-conditioning system, obtain the operation parameters of each energy consumption component within a set sampling period. Based on the fluid network topology model, map each energy consumption component to a node in the graph structure, and establish a bidirectional connection edge when there is a physical pipeline connection between two energy consumption components to obtain an initial operation diagram. According to the real-time pipeline flow rate, maximum design flow rate of the pipeline, real-time inlet and outlet water temperature difference, reference temperature difference, and adjustment coefficient between two energy consumption components corresponding to the bidirectional connection edge, determine the weight value corresponding to the bidirectional connection edge. In the initial operation diagram, add the corresponding weight value to the bidirectional connection edge to obtain the real-time power operation diagram of the central air-conditioning system.

[0013] Optionally, the step of adjusting the current control diagram based on the corrected environmental temperature curve and the real-time power operation diagram to obtain the target control diagram specifically includes: Based on the trained graph convolutional model, with the goal of minimizing the overall energy consumption of the real-time power operation diagram, perform prediction processing on the corrected environmental temperature curve and the real-time power operation diagram to obtain the target control diagram.

[0014] In an embodiment of the present invention, when it is detected that the current wet-bulb temperature deviates from the expected ambient temperature curve, a corrected ambient temperature curve corresponding to the expected ambient temperature curve is obtained; and, the current control diagram of the cooling tower and the real-time power operation diagram of the central air-conditioning system are obtained. The current control diagram includes a main controlled cooling device and multiple controlled cooling devices set in a topology. The main controlled cooling device is communicatively connected to the controlled cooling devices. The real-time power operation diagram includes multiple energy consumption components and connection edges between the energy consumption components. The connection edges represent the energy consumption transfer relationships between the energy consumption components. Based on the corrected ambient temperature curve and the real-time power operation diagram, the current control diagram is adjusted to obtain a target control diagram; and the controlled cooling devices of the cooling tower are controlled through the target control diagram. By establishing a coupled analysis model of a dynamically corrected ambient temperature curve and a real-time energy consumption transfer topology, the cooling tower equipment group can adaptively adjust the control strategy according to the ambient parameter offset, solving the problem of energy efficiency imbalance caused by ambient temperature fluctuations in traditional methods, improving the operation efficiency of the cooling device group, and reducing the redundant energy consumption of the cooling tower. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 is a flowchart of an energy-saving control method for a central air-conditioning cooling tower provided by an embodiment of the present invention; Figure 2 is a flowchart of another energy-saving control method for a central air-conditioning cooling tower provided by an embodiment of the present invention; Figure 3 is a specific flowchart of step S10 provided by an embodiment of the present invention; Figure 4 is a specific flowchart of step S102 provided by an embodiment of the present invention; Figure 5 is a specific flowchart of step S1 provided by an embodiment of the present invention; Figure 6 is a specific flowchart of step S13 provided by an embodiment of the present invention; Figure 7 is a specific flowchart of another step S1 provided by an embodiment of the present invention; Figure 8 is a specific flowchart of step S14 provided by an embodiment of the present invention; Figure 9It is a specific flowchart of step S15 provided by an embodiment of the present invention. Detailed implementation manners

[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present invention.

[0018] As Figure 1 shown Figure 1 It is a flowchart of an energy-saving control method for a central air-conditioning cooling tower provided by an embodiment of the present invention. The energy-saving control method for the central air-conditioning cooling tower includes the steps of: S1. When it is detected that the current wet-bulb temperature deviates from the expected ambient temperature curve, obtain the corrected ambient temperature curve corresponding to the expected ambient temperature curve; and obtain the current control diagram of the cooling tower and the real-time power operation diagram of the central air-conditioning system.

[0019] S2. Based on the corrected ambient temperature curve and the real-time power operation diagram, adjust the current control diagram to obtain the target control diagram.

[0020] S3. Control the controlled cooling equipment of the cooling tower through the target control diagram.

[0021] In the embodiment of the present invention, the current control diagram includes a main control cooling equipment and a plurality of controlled cooling equipment set by topology. The main control cooling equipment is communicatively connected to the controlled cooling equipment. The real-time power operation diagram includes a plurality of energy consumption components and the connection edges between the energy consumption components. The connection edges are the energy consumption transfer relationships between the energy consumption components.

[0022] Specifically, the main control cooling equipment serves as a control center, responsible for receiving environmental parameters, analyzing the energy consumption topology relationship, and generating global control instructions. It has a built-in dynamic correction algorithm for comparing the deviation between the real-time wet-bulb temperature and the expected ambient temperature curve.

[0023] There are multiple controlled cooling equipment, which can also be called a group of controlled cooling equipment, consisting of multiple distributed cooling equipment, connected to the main control equipment through a communication link, executing the control instructions issued by the main control equipment, and feeding back the real-time operation status.

[0024] The real-time power operation diagram describes the dynamic association relationship of the system energy consumption components in a graph structure, including energy consumption component nodes (such as fans, water pumps) and their connection edges (weight parameters of the energy consumption transfer path), used to characterize the energy interaction intensity between devices.

[0025] The corrected ambient temperature curve is generated based on historical weather data and an environmental parameter prediction model. When the wet-bulb temperature deviation is detected, the curve shape can be dynamically updated through interpolation or fitting algorithms to ensure its matching with the current environmental conditions.

[0026] The master device continuously monitors the real-time wet-bulb temperature. When the detected value deviates from the threshold range of the expected ambient temperature curve, the correction process of the expected ambient temperature curve is triggered.

[0027] Call the correction algorithm to generate the corrected ambient temperature curve, and simultaneously obtain the current cooling tower control diagram (including device topology status) and the real-time power operation diagram (including energy consumption component association relationship).

[0028] Based on the corrected ambient temperature curve, the master device analyzes the load distribution of each energy consumption component in the real-time power operation diagram and the transfer efficiency of the connection edges to identify high-energy consumption nodes or low-efficiency transfer paths. It can calculate the load redistribution plan for the controlled device group in combination with the communication and control capabilities of the device topology, and preferentially adjust the operating parameters of the devices associated with the low-efficiency energy consumption paths.

[0029] According to the adjustment results, reconstruct the target topology structure of the current control diagram to generate a target control diagram including device start-stop logic and speed / flow adjustment instructions.

[0030] The master device issues the target control diagram instructions to the corresponding controlled devices and simultaneously updates the energy consumption transfer relationship in the real-time power operation diagram to form a closed-loop control.

[0031] After the controlled devices execute the new instructions, the master device continuously collects environmental parameters and system energy consumption data to verify the energy efficiency improvement effect.

[0032] If the ambient wet-bulb temperature deviates again or the verification of the energy consumption transfer efficiency does not meet the expectations, repeat the above process for iterative optimization.

[0033] Through dynamic correction and the energy consumption dynamic diagram model, combined with the closed-loop collaborative control mechanism in the dynamic process, the system's rapid response and optimal energy efficiency adjustment under environmental parameter fluctuations are realized, overcoming the control lag and energy consumption waste problems caused by the rigid environmental model in traditional methods.

[0034] In an embodiment of the present invention, when it is detected that the current wet-bulb temperature deviates from the expected ambient temperature curve, a corrected ambient temperature curve corresponding to the expected ambient temperature curve is obtained; and, the current control diagram of the cooling tower and the real-time power operation diagram of the central air-conditioning system are obtained. The current control diagram includes a main controlled cooling device and multiple controlled cooling devices set by topology. The main controlled cooling device is communicatively connected to the controlled cooling devices. The real-time power operation diagram includes multiple energy consumption components and connection edges between the energy consumption components. The connection edges represent the energy consumption transfer relationships between the energy consumption components. Based on the corrected ambient temperature curve and the real-time power operation diagram, the current control diagram is adjusted to obtain a target control diagram; through the target control diagram, the controlled cooling devices of the cooling tower are controlled. The present invention establishes a coupled analysis model of a dynamically corrected ambient temperature curve and a real-time energy consumption transfer topology, enabling the cooling tower equipment group to adaptively adjust the control strategy according to the ambient parameter offset, solving the problem of energy efficiency imbalance caused by ambient temperature fluctuations in traditional methods, improving the operation efficiency of the cooling equipment group, and reducing the redundant energy consumption of the cooling tower.

[0035] It can be understood that in the specific implementation of the present application, it involves relevant data such as temperature data, weather data, equipment data, and user data. When the embodiments in the present application are applied to specific products or technologies, user permission or consent needs to be obtained, and the collection, use, and processing of relevant data, as well as the training, deployment, and invocation of algorithm models, all need to comply with the relevant laws, regulations, and standards of relevant countries and regions.

[0036] Optionally, before step S1, the method further includes: S10. Determine the expected ambient temperature curve at the location of the cooling tower according to the weather data of the day.

[0037] S20. Collect the current wet-bulb temperature.

[0038] S30. When the current wet-bulb temperature deviates from the expected ambient temperature curve by more than a preset deviation value, perform temperature curve prediction based on the historical wet-bulb temperature of the day to obtain a predicted wet-bulb temperature curve.

[0039] S40. Calculate the correlation between the predicted wet-bulb temperature curve and the expected ambient temperature curve, and when the correlation is less than the correlation threshold, determine that the wet-bulb temperature deviates from the expected ambient temperature curve.

[0040] In an embodiment of the present invention, meteorological forecast data or a historical weather database can be accessed to obtain the weather data of the day near the location of the cooling tower, and then the expected ambient temperature curve at the location of the cooling tower is generated through a time series analysis algorithm. The curve form of the expected ambient temperature curve is related to the daily temperature change trend.

[0041] The current wet-bulb temperature can be collected by a wet-bulb temperature acquisition unit, which is a high-precision sensor group deployed around the cooling tower. It collects and calibrates the current wet-bulb temperature in real time, and transmits its data to the main control cooling equipment through a communication interface.

[0042] Based on a prediction model with a sliding time window (such as ARIMA or LSTM), a short-term predicted wet-bulb temperature curve (the predicted wet-bulb temperature curve for the current day) can be generated according to the historical wet-bulb temperature data of the current day, serving as a reference benchmark for the dynamic changes in the environment.

[0043] Through a correlation coefficient calculation engine (such as the Pearson correlation coefficient algorithm), the morphological matching degree between the predicted wet-bulb temperature curve and the expected ambient temperature curve is quantified, and a deviation determination is triggered based on a preset correlation threshold.

[0044] Specifically, when the main control device starts the system, by parsing the meteorological data of the current day (such as the maximum / minimum temperature, humidity change rate), combined with the historical weather patterns of the same type, an initial expected ambient temperature curve is generated and set as the reference benchmark for the environmental parameters, that is, the expected ambient temperature curve.

[0045] The wet-bulb temperature acquisition unit uploads the measured value of the current wet-bulb temperature to the main control device at a fixed sampling period (such as 2 minutes).

[0046] The main control device calculates the difference between the current wet-bulb temperature and the theoretical value at the corresponding time point of the expected ambient temperature curve. If the difference exceeds the preset deviation value (such as ±1.5°C) within N consecutive sampling periods, it is determined that a significant deviation has occurred in the environmental parameters.

[0047] When a deviation is detected, the historical wet-bulb temperature data of the current day (such as the sequence in the past 2 hours) is extracted, and the predicted wet-bulb temperature curve for a future period of time (such as the next hour) is deduced through the prediction model.

[0048] The correlation coefficient between the predicted wet-bulb temperature curve and the expected ambient temperature curve in the same time interval is calculated through a correlation analysis algorithm. If the correlation coefficient is lower than the correlation threshold (such as 0.7), it is confirmed that the deviation of the environmental parameters is valid, and the process of obtaining the corrected ambient temperature curve is triggered.

[0049] If the correlation coefficient is higher than the threshold, it is considered that the current deviation belongs to short-term fluctuations, and the main control device can maintain the original expected ambient temperature curve unchanged and only compensate for the deviation through local parameter fine-tuning.

[0050] Through predicting environmental parameters, the present invention realizes the dynamic generation of the expected environmental temperature curve and the multi-level verification of the deviation criterion (difference determination + correlation analysis), avoiding frequent switching of control caused by misjudgment of a single threshold. At the same time, a double verification mechanism of a prediction model and a correlation coefficient is adopted, which not only improves the ability to predict the trend change of environmental parameters, but also ensures the rigor of the correction trigger condition, further reducing the risk of system misoperation.

[0051] Optionally, step S10 specifically includes: S101. Obtain the multi-day historical wet-bulb temperature curve of the location where the cooling tower is located, and obtain the multi-day historical weather data corresponding to the multi-day historical wet-bulb temperature curve in terms of date.

[0052] S102. Determine multiple representative weather data based on the multi-day historical weather data, and determine multiple representative wet-bulb temperature curves according to the multi-day historical wet-bulb temperature curve.

[0053] Among them, the representative weather data and the representative wet-bulb temperature curve correspond one by one.

[0054] S103. Associate the representative weather data with the representative wet-bulb temperature curve.

[0055] S104. According to the similarity between the current-day weather data and the representative weather data, determine the representative wet-bulb temperature curve corresponding to the representative weather data with the highest similarity as the expected environmental temperature curve of the location where the cooling tower is located.

[0056] In the embodiment of the present invention, the multi-day historical wet-bulb temperature curve of the location where the cooling tower is located and the multi-day historical weather data (including parameters such as temperature, humidity, and wind speed) associated with its date can be stored and managed to form a weather database.

[0057] An unsupervised clustering algorithm (such as K-means or DBSCAN) is used to cluster the historical weather data according to the similarity of meteorological characteristics, and a representative set of representative weather data is extracted.

[0058] Map the clustered representative weather data to the historical wet-bulb temperature curve of the corresponding date, and generate a representative wet-bulb temperature curve through a curve shape matching algorithm (such as dynamic time warping DTW) to establish an association relationship library of "weather pattern - wet-bulb curve".

[0059] Through a multi-dimensional feature similarity calculation model (such as Euclidean distance or cosine similarity), quantify the matching degree between the current-day weather data and each representative weather data, and select the representative wet-bulb temperature curve corresponding to the most matching representative weather data as the expected environmental temperature curve of the location where the cooling tower is located.

[0060] Specifically, in the initialization stage, multi-day historical wet-bulb temperature curves and corresponding multi-day historical weather data for a preset time period (such as the past 30 days) are loaded from the environmental weather database.

[0061] Perform feature standardization processing on the multi-day historical weather data, divide clustering clusters according to the distribution law of meteorological parameters, extract the central point data of each cluster as representative weather data, and mark the date range to which it belongs.

[0062] Extract multi-day historical wet-bulb temperature curves corresponding to the dates of each representative weather data from the environmental parameter spatio-temporal database, and eliminate noise interference through curve smoothing and normalization processing.

[0063] Perform morphological fusion (such as weighted average or principal component analysis) on multiple historical wet-bulb temperature curves within the same clustering cluster to generate a representative wet-bulb temperature curve for this weather pattern, and complete the binding storage of "weather pattern - wet-bulb curve".

[0064] After the main control device obtains the current-day weather data, perform multi-dimensional feature comparison between it and all representative weather data through similarity matching, calculate the similarity score and sort.

[0065] Select the representative weather data with the highest similarity, extract the corresponding representative wet-bulb temperature curve from the association relationship library of "weather pattern - wet-bulb curve", and use it as the benchmark output of the expected environmental temperature curve for the current day.

[0066] In a possible embodiment, if the actual wet-bulb temperature of the current day continuously deviates from the expected curve and triggers the correction process of the expected environmental temperature curve, the system automatically records this deviation event and updates the wet-bulb curve association relationship under this weather pattern in the historical data storage unit to achieve self-learning optimization.

[0067] The present invention solves the problem of insufficient accuracy of the expected curve caused by the traditional method relying on a single meteorological forecast data through the historical pattern clustering and multi-dimensional similarity matching mechanism. By mining the implicit association relationship between the weather pattern and the wet-bulb temperature curve in the historical data, the generation accuracy of the expected environmental temperature curve is significantly improved, thereby laying a reliable data foundation for the optimization of subsequent control strategies. At the same time, the self-learning calibration mechanism enhances the system's adaptability to long-term environmental changes.

[0068] Optionally, step S102 specifically includes: S1021. Cluster the multi-day historical weather data to obtain historical weather data clustering clusters, and determine the clustering center of the weather data clustering cluster as the representative weather data corresponding to the weather data clustering cluster.

[0069] Wherein, each of the historical weather data clustering clusters includes at least one day of historical weather data.

[0070] S1022. Based on the corresponding relationship between the multi-day wet-bulb temperature curves and the multi-day historical weather data in terms of date, in the multi-day historical wet-bulb temperature curves, determine the historical wet-bulb temperature curve clustering clusters, and determine the clustering center of the historical wet-bulb temperature curve clustering cluster as the representative wet-bulb temperature curve corresponding to the historical wet-bulb temperature curve clustering cluster.

[0071] Wherein, each of the historical wet-bulb temperature curve clustering clusters includes at least one day's historical wet-bulb temperature curve, and the historical wet-bulb temperature curve clustering clusters correspond one-to-one with the weather data clustering clusters.

[0072] In the embodiments of the present invention, a clustering algorithm (such as K-means) can be used to divide the feature space of the multi-day historical weather data, generate several historical weather data clustering clusters, and define the clustering center data of each clustering cluster as the representative weather data of the cluster.

[0073] Based on the date correspondence, perform secondary clustering on the multi-day historical wet-bulb temperature curves associated with the same weather data clustering cluster to generate historical wet-bulb temperature curve clustering clusters, and use the clustering center curve of the clustering cluster as the representative wet-bulb temperature curve of the cluster.

[0074] Establish a strict one-to-one correspondence between the weather data clustering clusters and the historical wet-bulb temperature curve clustering clusters to ensure that each representative weather data is only mapped to one representative wet-bulb temperature curve, and form an association relationship library of "weather pattern - wet-bulb curve".

[0075] The historical weather data can be feature-extracted (such as the daily change rates of temperature, humidity, and wind speed) by a weather data clustering engine, and the number of clusters can be automatically optimized through a preset number of clusters or silhouette coefficient, and the historical weather data clustering clusters are obtained by division.

[0076] The mean or median feature of all data points within each cluster can be extracted to generate the representative weather data of the cluster, and the date set to which it belongs is marked.

[0077] According to the date association relationship, extract all historical wet-bulb temperature curves corresponding to each weather data clustering cluster from the weather database.

[0078] Through the wet-bulb temperature curve clustering engine, perform time series alignment and morphological similarity analysis on the curves within the same cluster, generate historical wet-bulb temperature curve clustering clusters through a curve clustering algorithm, and calculate its clustering center curve as the representative wet-bulb temperature curve.

[0079] Bind the weather data clustering cluster with its corresponding historical wet-bulb temperature curve clustering cluster to form an association relationship library of "weather pattern - wet-bulb curve".

[0080] When new historical data is added, the system automatically triggers the incremental clustering algorithm to adjust the boundaries of the original clustering clusters or divide new clusters, and synchronously updates the representative data and curves in the associated mapping library.

[0081] If it is detected that the mapping between a certain weather data clustering cluster and the historical wet-bulb temperature curve clustering cluster is mismatched (such as the correlation coefficient continuously being lower than the threshold), then the re-clustering process is started to repair the association relationship.

[0082] Through the clustering and association mapping mechanism, the present invention breaks through the limitations of the independent analysis of weather data and wet-bulb temperature curves in traditional methods. By double clustering of weather and wet-bulb temperature curves, the logical consistency of "weather pattern - wet-bulb curve" is ensured, and the distortion of the expected curve caused by the deviation of a single clustering dimension is avoided. At the same time, the incremental calibration mechanism enables the system to continuously adapt to the long-term changes of meteorological patterns, significantly improving the robustness of the environmental parameter prediction model.

[0083] Optionally, step S1 specifically includes: S11. Obtain the deviation value between the current wet-bulb temperature and the expected environmental temperature curve, and obtain the fluctuation range of the historical wet-bulb temperature in the same period.

[0084] S12. Based on the deviation value and the fluctuation range of the historical wet-bulb temperature in the same period, construct a compensation coefficient matrix.

[0085] S13. Based on the compensation coefficient matrix, perform compensation calculation on the expected environmental temperature curve to obtain the corrected environmental temperature curve corresponding to the expected environmental temperature curve.

[0086] In the embodiment of the present invention, the instantaneous deviation value and the cumulative deviation trend of the theoretical values at the corresponding time points of the current wet-bulb temperature and the expected environmental temperature curve can be calculated in real time through difference operation and a sliding time window.

[0087] Extract the fluctuation range of the historical wet-bulb temperature in the same period (including the mean, standard deviation, and extreme values) of the same season / weather type as the current date from the historical database, and establish an environmental parameter fluctuation benchmark model.

[0088] Based on the fuzzy logic or neural network algorithm, perform multi-dimensional matching analysis on the deviation value and the historical fluctuation range to generate a compensation coefficient matrix reflecting the temperature deviation direction and amplitude. The matrix dimension is associated with the time interval and the spatial distribution weight.

[0089] Through matrix multiplication and curve interpolation algorithm, apply the compensation coefficient matrix to the original data points of the expected environmental temperature curve, and output a smooth and continuous corrected environmental temperature curve.

[0090] The compensation coefficient matrix is as follows: Wherein, is the deviation value of the current wet-bulb temperature, , is the extreme value of the wet-bulb temperature in the same period of history, k1 is the linear compensation factor, and k2 is the exponential decay coefficient. is the exponential decay term, is the time decay constant.

[0091] Specifically, after the master device detects the deviation of the wet-bulb temperature, the deviation value calculation unit calculates the instantaneous deviation value of the current wet-bulb temperature and the expected curve within the same time window in a time-stamp alignment manner (e.g., ΔT = T_measured - T_expected).

[0092] Retrieve the wet-bulb temperature data in the same period of history (e.g., the same day of the same month in the past 5 years), statistically analyze the temperature fluctuation range in this time period (e.g., ±2.3°C), and extract its statistical distribution characteristics to obtain the historical fluctuation range.

[0093] Compare the instantaneous deviation value with the historical fluctuation range. If the deviation value is within the historical fluctuation range, generate a linear compensation coefficient according to the deviation direction (positive / negative) and duration; if the deviation value exceeds the historical fluctuation extreme value, start the non-linear compensation model and generate an adaptive compensation coefficient in combination with the correction strategy of historical similar abnormal events.

[0094] The matrix dimension is divided into a time axis (hourly segmentation) and a space axis (cooling tower group distribution area) according to the system topology, and the weights of each matrix element are dynamically adjusted by the equipment energy consumption impact factor.

[0095] Perform a point-by-period superposition operation on the compensation coefficient matrix and the original data points of the expected ambient temperature curve: For the linear compensation interval, use the translation compensation formula: T_corrected = T_expected + k1·ΔT (k1 is the linear weight coefficient of the compensation coefficient matrix); For the non-linear compensation interval, call a curve fitting function (such as cubic spline interpolation) to reconstruct the curve shape to ensure that the corrected temperature curve is continuous and meets the thermodynamic constraint conditions.

[0096] Output the corrected ambient temperature curve to the master cooling equipment for subsequent control chart optimization.

[0097] After the master device applies the new curve, continuously monitor the matching degree between the actual wet-bulb temperature and the corrected curve. If the deviation value continuously decreases and stabilizes within the threshold, it is determined that the correction is effective; If there is still a significant deviation after correction, trigger the weight re-training process of the compensation coefficient matrix and update the parameter benchmark in the historical fluctuation analysis model.

[0098] Through the multi-dimensional compensation model and the historical data-driven correction mechanism, the present invention realizes the precision and self-adaptability of the environmental temperature curve correction. Compared with directly using a fixed offset for adjustment, the introduction of the compensation coefficient matrix enables the correction process to take into account the historical fluctuation law and the real-time topological state, avoiding problems such as over-compensation or response lag. At the same time, it reduces the number of oscillations of the control strategy caused by environmental mutations.

[0099] Optionally, step S13 specifically includes: S131. According to the predicted wet-bulb temperature curve, determine the predicted value of the wet-bulb temperature, and determine the first confidence level corresponding to the predicted value of the wet-bulb temperature according to the compensation coefficient matrix.

[0100] S132. Collect the measured average wet-bulb temperature at the location around the cooling tower within a preset radius range, and determine the second confidence level corresponding to the measured average wet-bulb temperature according to the health status of the sensor.

[0101] S133. Calculate the heat reflux influence amount according to the air outlet wind speed and exhaust gas temperature of the cooling tower, and determine the third confidence level corresponding to the heat reflux influence amount according to the compensation coefficient matrix.

[0102] S134. Based on the predicted value of the wet-bulb temperature and the corresponding first confidence level, the measured wet-bulb temperature and the corresponding second confidence level, and the heat reflux influence amount and the corresponding third confidence level, determine the corrected wet-bulb temperature.

[0103] S135. Based on the corrected wet-bulb temperature, process the expected environmental temperature curve, perform compensation calculation, and obtain the corrected environmental temperature curve corresponding to the expected environmental temperature curve.

[0104] In the embodiment of the present invention, the first confidence level (value range 0-1) of the predicted wet-bulb temperature value can be calculated based on the time decay characteristic of the compensation coefficient matrix and the error distribution law of the prediction model, which represents the credible weight of the predicted value.

[0105] By integrating a distributed temperature and humidity sensor network, collect multi-node wet-bulb temperature data within a preset radius range (such as 50 meters) around the cooling tower in real time, obtain the measured average wet-bulb temperature through outlier rejection and mean calculation, and generate the second confidence level in combination with the self-check status of the sensor (such as calibration period, fault code).

[0106] Obtain the real-time operating condition parameters through the air outlet wind speed sensor and the exhaust gas temperature probe, combine with the computational fluid dynamics model (CFD) simulation database, deduce the heat reflux influence amount (temperature superposition value ΔT_return), and generate the third confidence level according to the spatial distribution parameters in the compensation coefficient matrix.

[0107] Using a Bayesian probability model or a weighted average algorithm, perform confidence weighted fusion on the predicted value, measured mean value, and heat reflux influence amount, and output the comprehensive corrected wet-bulb temperature.

[0108] After the master device obtains the generated compensation coefficient matrix, calculate the first confidence level (such as 0.85) of the predicted wet-bulb temperature value according to the stability parameter (such as variance) in the corresponding time interval of the compensation coefficient matrix.

[0109] If the health status of the sensor is normal (no alarm code), calculate the mean value and assign the second confidence level (such as 0.95); if there is a sensor anomaly, reduce the confidence weight (such as 0.6).

[0110] According to the air outlet wind speed and exhaust temperature, match the heat interference coefficient table in the CFD model, calculate the heat reflux influence amount under the current working condition, and assign the third confidence level (such as 0.75) according to the regional weight in the compensation coefficient matrix.

[0111] The multi-source data fusion engine receives three groups of input data and their confidence levels, the predicted wet-bulb temperature value (T_pred, confidence level P1), the measured mean wet-bulb temperature (T_meas, confidence level P2), and the heat reflux influence amount (ΔT_return, confidence level P3) Use the weighted fusion formula to calculate the corrected wet-bulb temperature: T_corrected = (T_pred×P1 + T_meas×P2 + (T_meas + ΔT_return)×P3) / (P1 + P2 + P3).

[0112] In a possible embodiment, when a certain confidence level is lower than the preset lower limit (such as 0.3), automatically eliminate this data source and only use the effective high-confidence data for calculation.

[0113] Taking the corrected wet-bulb temperature as the target value, perform segmented compensation on the expected ambient temperature curve. For the historical period that has deviated (such as the past 1 hour), use the reverse interpolation algorithm to correct the curve shape; for the current and future periods, use the corrected wet-bulb temperature as the reference point and adjust the curve trend through the exponential smoothing method to ensure time continuity.

[0114] In a possible embodiment, the refined weather forecast for a future period can be obtained through the WebService interface, and the predicted value of the wet-bulb temperature is parsed whose confidence weight w1 is negatively correlated with the linear compensation coefficient k1.

[0115] Calculate the measured mean wet-bulb temperature within the preset radius through the temperature and humidity sensor array deployed around the cooling tower whose weight w2 is dynamically adjusted according to the health status of the sensor; Calculate the influence of thermal reflux according to the air outlet velocity and exhaust temperature of the cooling tower using a CFD simulation model , where its weight w3 is associated with the exponential decay term in the compensation coefficient matrix ; Calculate the fused corrected wet-bulb temperature as: Using as the input, perform deformation processing on the expected ambient temperature curve using the cubic spline interpolation algorithm to obtain the corrected ambient temperature curve corresponding to the expected ambient temperature curve, specifically including: inserting dynamic control nodes at temperature mutation points (such as where the curvature exceeds 0.5 °C / min), and the node density is proportional to the standard deviation of the compensation coefficient matrix C(t). Dynamically scale the curve slope: where is the slope adjustment sensitivity parameter, and its value is dynamically calculated according to the weight distribution of w1, w2, w3, and an average value can be taken.

[0116] In a possible embodiment, the master control device records the correlation between the confidence of each data source and the actual deviation error in each fusion calculation, and updates the confidence evaluation model parameters through offline training (such as adjusting the influence weight of the sensor health status on the second confidence) to improve the accuracy of subsequent calculations.

[0117] The present invention solves the problem that the correction of a single data source is vulnerable to local interference or model errors through a multi-source data fusion mechanism and a confidence dynamic weighting strategy. By introducing a thermal reflux physical model to compensate for the environmental monitoring blind area and establishing a direct correlation between the sensor health status and data credibility, the calculation robustness of the corrected wet-bulb temperature is significantly improved.

[0118] Optionally, step S1 specifically includes: S14. Determine the master-slave control relationship between the master control cooling device and the controlled cooling device, and determine the current control diagram of the cooling tower according to the master-slave control relationship.

[0119] S15. Based on the energy consumption component list of the central air-conditioning system, map each energy consumption component to a node in the graph structure and generate connection edges between upstream and downstream nodes to obtain the real-time power operation diagram of the central air-conditioning system.

[0120] In the embodiment of the present invention, the built-in device communication protocol parser and topology discovery algorithm can be used to identify the hierarchical control relationship between the master control cooling device and the controlled cooling device in the cooling tower group, and construct a tree-shaped current control diagram with the master control device as the root node.

[0121] By integrating the energy consumption acquisition interface of the device, the energy consumption component list of the central air-conditioning system (such as compressors, water pumps, and fans) is mapped to the nodes in the graph structure, and the connection edges between the nodes are generated through the energy consumption flow analysis model. The attributes of the connection edges include the real-time power transfer direction and efficiency parameters.

[0122] Through hierarchical topology analysis and real-time energy consumption flow modeling, the present invention solves the problem of the separate analysis of the device control relationship and the energy transfer path in the traditional method. The current control diagram ensures the certainty of the device group command transmission, while the real-time power operation diagram reveals the system energy efficiency bottleneck through dynamic weight parameters. The two cooperate to provide an accurate topology and energy consumption dual constraint model for the subsequent control strategy, which can reduce the control command transmission delay and improve the effectiveness of the energy consumption optimization strategy.

[0123] Optionally, the master-slave control relationship between the main control cooling device and the controlled cooling device of the cooling tower is a dynamically changing relationship. Step S14 specifically includes: S141. Poll the device nodes in the cooling tower control network through a preset communication protocol, and determine the physical address and communication port configuration of the main control cooling device based on the control permission identifier in the device response message.

[0124] S142. Analyze the device attribute parameters of each controlled cooling device.

[0125] Among them, the device attribute parameters include device ID, rated power threshold, and real-time operating frequency. S143. Map the main control cooling device and the controlled cooling device to the topology diagram to obtain the current control diagram of the cooling tower.

[0126] In the embodiment of the present invention, a multi-protocol communication adapter is integrated, which supports polling device nodes through preset communication protocols such as Modbus and BACnet, analyzes the control permission identifier (such as the master / slave status bit) in the device response message, and real-time identifies the physical address and port configuration of the master device.

[0127] Through the device parameter database, extract the device attribute parameters of the controlled cooling device, including key operating indicators such as device unique ID, rated power threshold, and real-time operating frequency.

[0128] According to the real-time identified master-slave control relationship and device attribute parameters, construct a hierarchical current control diagram with the master device as the control core and the controlled devices as branch nodes, which supports dynamic refreshing and storage of the topology structure.

[0129] Specifically, when the master device starts, it broadcasts a device status query command to the cooling tower control network, and all online devices return response messages containing control permission identifiers (such as the master device identifier is "Master" and the controlled device identifier is "Slave").

[0130] Analyze the device type field and IP / MAC address in the parsed response message to determine the physical address and communication port configuration (such as TCP port 502) of the main control cooling device in the current network, and mark it as the control instruction issuing node.

[0131] Write the parsed parameters into the device attribute database and synchronize the data with the topology mapping engine.

[0132] Take the identified main control device as the root node of the topology graph, and associate its physical address with the port configuration; Map the controlled devices as child nodes according to the device ID, and bind their rated power thresholds and real-time operating frequencies as node attributes; Based on the actual communication links between devices (such as wired / wireless connection paths), generate directed connection edges between nodes, and the edge weight parameters are dynamically calculated by communication delay and signal strength.

[0133] Output the current control graph with device attributes and real-time status for the main control device to perform control strategy distribution and energy efficiency optimization.

[0134] If it is detected that the main control device is offline or has communication anomalies, immediately trigger a re-polling process, elect a new main control device according to the preset fault switching strategy (such as priority weight sorting), and update the root node and connection relationship in the topology graph.

[0135] The present invention breaks through the limitations of the traditional fixed configuration of the master-slave relationship. Through response message parsing and device parameter fusion, it ensures that the control graph can accurately reflect the real-time operating status and network topology of the device group, providing underlying support for dynamic load balancing and fast fault switching.

[0136] Optionally, step S15 specifically includes: S151. Based on the energy consumption component list of the central air-conditioning system, obtain the operating parameters of each energy consumption component within the set sampling period.

[0137] S152. Based on the fluid network topology model, map each energy consumption component as a node in the graph structure, and establish a two-way connection edge when there is a physical pipeline connection between two energy consumption components to obtain an initial operating graph.

[0138] S153. Determine the weight value corresponding to the two-way connection edge according to the real-time pipeline flow rate, the maximum designed pipeline flow rate, the real-time temperature difference between the inlet and outlet water, the reference temperature difference, and the adjustment coefficient between the two energy consumption components corresponding to the two-way connection edge.

[0139] S154. In the initial operating graph, add the corresponding weight value to the two-way connection edge to obtain the real-time power operating graph of the central air-conditioning system.

[0140] In the embodiments of the present invention, sensors of each component of the central air-conditioning system are connected, and the operating parameters of the energy-consuming components are collected at a set sampling period (such as 10 seconds), including real-time data such as power, flow rate, and temperature.

[0141] A physical pipeline layout database of the built-in central air-conditioning system is used to convert the device connection relationships in the fluid network topology model into nodes and edges in a graph structure, supporting the dynamic generation and attribute binding of bidirectional connection edges.

[0142] Based on parameters such as pipeline flow rate and temperature difference, the weight calculation engine determines the weight values of the connection edges through a preset weight calculation rule (such as a composite function of flow saturation and temperature difference efficiency).

[0143] Integrate the node attributes and edge weights into the graph data structure to generate a real-time power operation graph with dynamic weights, and provide graph traversal and path analysis interfaces.

[0144] Specifically, according to the energy-consuming component list, the master device sends data collection instructions to all components in the list (such as chillers and circulation pumps) to obtain the operating parameters within the set sampling period, including: real-time pipeline flow rate (such as m³ / h); maximum designed pipeline flow rate (equipment nameplate parameter); real-time temperature difference between the inlet and outlet of the pipeline (ΔT_measured); reference temperature difference (theoretical ΔT_ref under the design condition).

[0145] Map each energy-consuming component to a node in the graph structure, and the node attributes include device type and real-time power value; if there is a direct physical pipeline connection between two components (such as the outlet of the cooling tower → the inlet of the water pump), then establish a bidirectional connection edge to form an initial operation graph describing the fluid flow direction and energy transfer path.

[0146] For each bidirectional connection edge, the weight calculation engine extracts the operating parameters of the two associated components and calculates the weight value according to the following formula: Weight = (real-time pipeline flow rate / maximum designed pipeline flow rate) × adjustment coefficient + (ΔT_measured / ΔT_ref) × (1 - adjustment coefficient) Among them, the adjustment coefficient is dynamically set according to the system operation mode (such as focusing on flow rate in the cooling mode, taking 0.7; focusing on temperature difference in the heating mode, taking 0.3).

[0147] The calculation result is normalized to the range of 0 - 1. The closer the weight value is to 1, the better the energy efficiency transfer state of the connection edge. Bind the calculated weight value to the corresponding bidirectional connection edge, update the initial operation graph, and generate a real-time power operation graph with dynamic weights.

[0148] The master device performs energy consumption path analysis based on this graph, identifies low-weight connection edges (such as weight < 0.4) and triggers optimization instructions (such as adjusting the valve opening or pump speed).

[0149] In a possible embodiment, based on the fluid network topology model, each energy-consuming component is mapped to a node in the graph structure, and connection edges are generated by the following rules: a) When there is a physical pipeline connection between two components, a bidirectional connection edge is established. b) The weight value of the connection edge is dynamically updated according to the real-time flow sensor data, and the weight calculation formula is: wherein, is the pipeline flow rate at time t, is the maximum design flow rate of the pipeline, is the temperature difference between the inlet and outlet water, is the reference temperature difference, and α and β are adjustment coefficients set according to experience.

[0150] The present invention solves the problem in traditional energy consumption topology modeling that only static connection relationships are relied on while ignoring the dynamic energy efficiency state through the fluid network model and the multi-parameter weight calculation mechanism. By introducing the dual-factor weight model of flow saturation and temperature difference efficiency, the real-time power operation diagram can accurately reflect the actual efficiency of the system energy transfer path, providing a high-precision decision-making basis for the energy efficiency optimization strategy.

[0151] Optionally, the step of adjusting the current control diagram based on the corrected ambient temperature curve and the real-time power operation diagram to obtain the target control diagram specifically includes: Based on the trained graph convolutional model, with the goal of minimizing the overall energy consumption of the real-time power operation diagram, the corrected ambient temperature curve and the real-time power operation diagram are predicted to obtain the target control diagram.

[0152] In the embodiment of the present invention, a trained graph convolutional model (GCN) is preset in the system, and its network structure is adapted to the topological characteristics of the central air-conditioning system. The input layer receives the graph embedding vectors of the corrected ambient temperature curve and the real-time power operation diagram, and the output layer generates the optimization parameters of the target control diagram.

[0153] Define a multi-constraint optimization goal centered on minimizing the overall energy consumption of the real-time power operation diagram, including equipment power threshold constraint, temperature difference stability constraint, and communication delay constraint.

[0154] Convert the output parameters of the graph convolutional model into an executable device control instruction set and reconstruct them into a target control diagram that conforms to the communication protocol of the cooling tower group.

[0155] Specifically, the master control device discretizes the corrected ambient temperature curve into time series data, and performs spatio-temporal alignment with the node attributes (such as device power and connection edge weights) in the real-time power operation diagram to generate unified graph structure input data. The graph data is normalized, and node feature vectors (such as the real-time load rate of the device) and edge feature vectors (such as the energy consumption transfer efficiency) are extracted to construct the input tensor of the graph convolutional model.

[0156] Load the trained graph convolutional model, aggregate the local features of nodes and edges through multi-layer graph convolutional operations to generate high-dimensional graph embedding representations, and capture the energy consumption correlation relationships between devices; in the embedding space, with minimizing the overall energy consumption as the objective function, combined with constraint conditions to solve the optimal control parameters (such as device start-stop sequences and speed adjustment amounts). The model output is the topological adjustment scheme of the target control graph, including the master control device instruction set, the priority queue of controlled devices, and the connection edge weight update strategy.

[0157] Update the master-slave control relationship according to the device priority queue, dynamically switch the control link; write parameters such as the speed adjustment amount and power threshold into the corresponding device node attributes; update the connection edge weights according to the predicted values of the energy consumption transfer efficiency to generate the target control graph with optimized parameters.

[0158] The master control device sends the target control graph to the cooling tower group to trigger the synchronous execution of device control instructions.

[0159] In a possible embodiment, after the system runs, the actual energy consumption data is collected and the difference analysis is performed with the predicted values of the target control graph to calculate the model prediction error.

[0160] If the error exceeds the allowable range, trigger the incremental training process of the graph convolutional model, and use the real-time data to update the network weight parameters to achieve online optimization of the model.

[0161] The present invention breaks through the limitations of traditional control strategies through the graph convolutional model and the end-to-end optimization mechanism. The graph convolutional model can simultaneously model the multi-dimensional coupling relationships of environmental parameters, device topologies, and energy consumption transfers, and quickly search for the global energy consumption optimal solution in the embedding space.

[0162] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0163] The above-disclosed are only the preferred embodiments of the present invention. Certainly, the scope of the rights of the present invention cannot be limited thereby. Therefore, equivalent changes made according to the claims of the present invention still fall within the scope covered by the present invention.

Claims

1. A method for energy saving control of a central air conditioning cooling tower, characterized in that: The method comprises the following steps: When it is detected that the current wet-bulb temperature deviates from the expected ambient temperature curve, a corrected ambient temperature curve corresponding to the expected ambient temperature curve is obtained; and, a current control diagram of the cooling tower and a real-time power operation diagram of the central air-conditioning system are obtained, wherein the current control diagram includes a master-controlled cooling device and a plurality of controlled cooling devices in a topological setting, the master-controlled cooling device is in communication connection with the controlled cooling devices, and the real-time power operation diagram includes a plurality of energy consumption components and connection edges between the energy consumption components, wherein the connection edges are energy consumption transfer relations between the energy consumption components; Based on the modified ambient temperature curve and the real-time power operation diagram, the current control diagram is adjusted to obtain a target control diagram; The controlled cooling equipment of the cooling tower is controlled through the target control diagram.

2. The energy-saving control method for a central air-conditioning cooling tower according to claim 1, characterized in that: Before the step of acquiring a corrected ambient temperature curve corresponding to the expected ambient temperature curve when the current wet-bulb temperature is detected to deviate from the expected ambient temperature curve, the method further includes: Determine the expected ambient temperature curve at the location of the cooling tower based on the weather data of the day; Collect current wet-bulb temperature; When the current wet-bulb temperature deviates from the expected ambient temperature curve by more than a preset deviation value, a temperature curve is predicted based on the historical wet-bulb temperature of the day to obtain a predicted wet-bulb temperature curve; A correlation between the predicted wet-bulb temperature curve and the expected ambient temperature curve is calculated, and when the correlation is less than a correlation threshold, it is determined that the wet-bulb temperature deviates from the expected ambient temperature curve.

3. The energy-saving control method for a central air-conditioning cooling tower according to claim 2, characterized in that: The step of determining the expected ambient temperature curve at the location of the cooling tower according to the weather data of the day specifically includes: Obtain a multi-day historical wet-bulb temperature curve of the location of the cooling tower, and obtain a multi-day historical weather data corresponding to the multi-day historical wet-bulb temperature curve in terms of date; Determine a plurality of representative weather data based on the plurality of days of historical weather data, and determine a plurality of representative wet-bulb temperature curves based on the plurality of days of historical wet-bulb temperature curves, wherein the representative weather data correspond to the representative wet-bulb temperature curves one by one; Associating the representative weather data with the representative wet-bulb temperature curve; According to the similarity between the weather data of the day and the representative weather data, the representative wet-bulb temperature curve corresponding to the representative weather data with the highest similarity is determined as the expected ambient temperature curve at the location of the cooling tower.

4. The energy-saving control method for a central air-conditioning cooling tower according to claim 3, characterized in that: The step of determining a plurality of representative weather data based on the multi-day historical weather data, and determining a plurality of representative wet-bulb temperature curves according to the multi-day historical wet-bulb temperature curves specifically includes: Clustering multiple days of historical weather data to obtain historical weather data clusters, and determining the cluster center of the weather data cluster as the representative weather data corresponding to the weather data cluster, each of the historical weather data clusters includes at least one day of historical weather data; Based on the correspondence between the multi-day historical wet-bulb temperature curve and the multi-day historical weather data on the date, a historical wet-bulb temperature curve cluster is determined in the multi-day historical wet-bulb temperature curve, and the cluster center of the historical wet-bulb temperature curve cluster is determined as the representative wet-bulb temperature curve corresponding to the historical wet-bulb temperature curve cluster, each of the historical wet-bulb temperature curve clusters includes at least one day's historical wet-bulb temperature curve, and the historical wet-bulb temperature curve clusters correspond one-to-one to the weather data clusters.

5. The energy-saving control method for a central air-conditioning cooling tower according to claim 2, characterized in that: The steps of obtaining a corrected ambient temperature curve corresponding to the expected ambient temperature curve specifically include: Obtaining the deviation value between the current wet-bulb temperature and the expected ambient temperature curve, and obtaining the fluctuation range of the wet-bulb temperature during the same period in history; Based on the deviation value and the historical wet-bulb temperature fluctuation range during the same period, a compensation coefficient matrix is ​​constructed; Based on the compensation coefficient matrix, a compensation calculation is performed on the expected ambient temperature curve to obtain a corrected ambient temperature curve corresponding to the expected ambient temperature curve.

6. The energy-saving control method for a central air-conditioning cooling tower according to claim 5, characterized in that: The step of performing compensation calculation on the expected ambient temperature curve based on the compensation coefficient matrix to obtain a corrected ambient temperature curve corresponding to the expected ambient temperature curve specifically includes: Determining a predicted wet-bulb temperature value according to the predicted wet-bulb temperature curve, and determining a first confidence level corresponding to the predicted wet-bulb temperature value according to the compensation coefficient matrix; Collecting the average value of the measured wet-bulb temperature within a preset radius around the cooling tower, and determining a second confidence level corresponding to the average value of the measured wet-bulb temperature according to the health status of the sensor; Calculating the heat reflux influence amount according to the wind speed and exhaust temperature at the cooling tower outlet, and determining the third confidence level corresponding to the heat reflux influence amount according to the compensation coefficient matrix; Determine a corrected wet-bulb temperature based on the wet-bulb temperature prediction value and the corresponding first confidence level, the measured wet-bulb temperature and the corresponding second confidence level, the heat reflow influence amount and the corresponding third confidence level; Based on the corrected wet-bulb temperature, the expected ambient temperature curve is processed and a compensation calculation is performed to obtain a corrected ambient temperature curve corresponding to the expected ambient temperature curve.

7. The energy-saving control method for a central air-conditioning cooling tower according to claim 1, characterized in that: The step of obtaining the current control diagram of the cooling tower and the real-time power operation diagram of the central air conditioning system specifically includes: Determine the master-controlled relationship between the master-controlled cooling device and the controlled cooling device, and determine the current control diagram of the cooling tower according to the master-controlled relationship; Based on the list of energy-consuming components of the central air-conditioning system, each energy-consuming component is mapped as a node in the graph structure, and connecting edges between upstream and downstream nodes are generated to obtain the real-time power operation graph of the central air-conditioning system.

8. The energy-saving control method for a central air-conditioning cooling tower according to claim 7, characterized in that: The master-controlled relationship between the master-controlled cooling device and the controlled cooling device of the cooling tower is a dynamically changing relationship. The step of determining the master-controlled relationship between the master-controlled cooling device and the controlled cooling device and determining the current control diagram of the cooling tower according to the master-controlled relationship specifically includes: Poll the device nodes in the cooling tower control network through a preset communication protocol, and determine the physical address and communication port configuration of the main control cooling device based on the control authority identifier in the device response message; Parsing device attribute parameters of each controlled cooling device, wherein the device attribute parameters include device ID, rated power threshold, and real-time operating frequency; The main control cooling device and the controlled cooling device are mapped into a topology diagram to obtain a current control diagram of the cooling tower.

9. The energy-saving control method for a central air-conditioning cooling tower according to claim 7, characterized in that: The steps of mapping each energy-consuming component into a node in a graph structure based on the energy-consuming component list of the central air-conditioning system and generating connecting edges between upstream and downstream nodes to obtain a real-time power operation graph of the central air-conditioning system specifically include: Based on the list of energy-consuming components of the central air-conditioning system, obtain the operating parameters of each of the energy-consuming components within a set sampling period; Based on the fluid network topology model, each of the energy-consuming components is mapped as a node in the graph structure, and when there is a physical pipeline connection between two of the energy-consuming components, a bidirectional connection edge is established to obtain an initial operation graph; Determine a weight value corresponding to the bidirectional connection edge according to the real-time pipeline flow, the maximum design pipeline flow, the real-time inlet and outlet water temperature difference, the reference temperature difference and the adjustment coefficient between the two energy-consuming components corresponding to the bidirectional connection edge; In the initial operation diagram, the corresponding weight values ​​are added to the bidirectional connection edges to obtain a real-time power operation diagram of the central air-conditioning system.

10. The energy-saving control method for a central air-conditioning cooling tower according to any one of claims 1 to 9, characterized in that: The step of adjusting the current control diagram based on the corrected ambient temperature curve and the real-time power operation diagram to obtain a target control diagram specifically includes: Based on the trained graph convolution model, with the goal of minimizing the overall energy consumption of the real-time power operation diagram, the corrected ambient temperature curve and the real-time power operation diagram are predicted and processed to obtain a target control diagram.

Images