A central air conditioning energy-saving control method and system based on multi-host cooperation

Abstract

This invention discloses a central air conditioning energy-saving control method and system based on multi-host collaboration, comprising the following steps: acquiring the host parameters of the central air conditioning system and establishing object mapping relationships between each host and the chilled water side, cooling water side, and terminal load side; collecting and preprocessing real-time operating data of the central air conditioning system; constructing a partial load energy efficiency characteristic model for each host, a host group collaborative energy consumption evaluation model, and a host operating capacity constraint set; constructing a load prediction sequence and determining the control update cycle; constructing a collaborative optimization decision model within the control update cycle; performing Big-M transformation on the collaborative optimization decision model to form a mixed integer optimization problem and solving it to obtain the optimal control strategy; outputting a set of control commands and issuing them to the host controller and system-side actuators. This invention employs a disjunctive optimization modeling method to achieve multi-host collaborative energy-saving control, possessing the advantages of global optimization and high operational stability.

Description

Technical Field

[0001] This invention relates to the field of intelligent control of central air conditioning, and in particular to a method and system for energy-saving control of central air conditioning based on multi-host collaboration. Background Technology

[0002] Existing central air conditioning systems mostly employ independent control of individual units or group control strategies based on empirical rules, achieving operational regulation by setting fixed start-stop sequences, load distribution ratios, and temperature setpoints. Some systems introduce simple load forecasting or time-of-use control strategies to adjust operating cycles based on historical load trends, but the overall approach remains primarily static or based on local optimization, lacking a detailed characterization of the collaborative energy consumption relationships among multiple units. Current control methods typically treat unit energy consumption and system-side transmission and distribution energy consumption separately, failing to establish a unified energy consumption evaluation framework and making it difficult to achieve global energy-saving optimization under complex operating conditions.

[0003] In multi-host collaborative operation scenarios, existing technologies struggle to simultaneously address the load characteristics of individual hosts, balance the total system cooling capacity, and constrain operational stability. Frequent start-ups and shutdowns, uneven load distribution, and fluctuations in supply and return water parameters become prominent issues. Traditional optimization methods often fail to embed start-up and shutdown logic into a solvable form within a unified optimization model, leading to a disconnect between the solution process and operational constraints. This makes it difficult to obtain a stable and executable globally optimal control strategy, thereby impacting system energy efficiency and operational reliability.

[0004] Therefore, how to provide a central air conditioning energy-saving control method and system based on multi-host collaboration is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a central air conditioning energy-saving control method and system based on multi-host collaboration. This invention adopts a disjunctive optimization modeling method to achieve multi-host collaborative energy-saving control, which has the advantages of global optimization and high operational stability.

[0006] According to an embodiment of the present invention, a central air conditioning energy-saving control method based on multi-host collaboration includes the following steps: Obtain the main unit parameters of the central air conditioning system, and establish object mapping relationships between each main unit and the chilled water side, cooling water side, and terminal load side respectively; Collect real-time operating data of the central air conditioning system and preprocess it to obtain a standardized operating dataset; Based on host parameters and standardized operating datasets, construct partial load energy efficiency characteristic models, host group collaborative energy consumption evaluation models, and host operating capacity constraint sets for each host. Construct a load forecast sequence and determine the control update cycle based on the load forecast sequence; Within the control update cycle, decision variables are set, and the generalized disjunctive programming method is used to establish start-up disjunctive branches and shutdown disjunctive branches. In addition, a collaborative optimization decision model is constructed by combining the partial load energy efficiency characteristic model, the host group collaborative energy consumption evaluation model and the host operating capacity constraint set. The Big-M transformation of the collaborative optimization decision model is performed to form a mixed integer optimization problem, and the optimal control strategy is obtained by solving the problem. Based on the optimal control strategy, a set of control commands is output and sent to the host controller and the system-side actuators.

[0007] Optionally, the host parameters include the number of hosts, host type parameters, and system topology parameters.

[0008] Optionally, the real-time operating data includes the start / stop status, load rate, input power, outlet water temperature, return water temperature, chilled water flow rate, cooling water flow rate, fault status indicator, and cumulative running time of each host, as well as the supply and return water temperature, supply and return water pressure difference, total flow rate, outdoor ambient temperature, and current load indicators on the system side. The preprocessing includes timestamp alignment, missing data processing, abnormal data processing, and dimensional normalization.

[0009] Optionally, the construction of the partial load energy efficiency characteristic model, the host cluster collaborative energy consumption evaluation model, and the host operating capacity constraint set specifically includes: Based on the host type parameter, a subset of data related to partial load operation is extracted from the standardized operation dataset. Samples with the start-up status are retained. Among the retained samples, samples with fault status are removed by using the fault status identifier as the filtering condition. The filtered samples are then segmented and aggregated according to the load rate interval to form a partial load segmented sample set. A partial load energy efficiency characteristic model for each host is constructed based on a partial load segmented sample set. The partial load energy efficiency characteristic model includes an energy consumption relationship expression with host load rate and inlet / outlet water temperature difference as independent variables and host input power as dependent variable, and an energy efficiency relationship expression with host load rate as independent variable and energy consumption level corresponding to unit cooling capacity as dependent variable. A host cluster collaborative energy consumption evaluation model is constructed based on system topology parameters and standardized operation dataset. The host cluster collaborative energy consumption evaluation model includes a host cluster total energy consumption expression, a system-side transmission and distribution energy consumption expression, and a collaborative coupling correction term expression. Based on the host parameters, extract the rated capacity range of each host and construct the host load upper and lower limit constraints in the host operating capacity constraint set; By combining the historical operating trajectories in the standardized operating dataset, the load adjustment rate boundary is determined and load change rate constraints for each host are formed in the host operating capacity constraint set. Construct a system energy balance relationship based on the current load index and generate a system total cooling capacity balance constraint in the host operating capacity constraint set; Based on the system topology parameters and the hydraulic operating states in the standardized operating dataset, the operating boundary conditions are established, and the supply and return water temperature constraints and supply and return water pressure difference constraints in the host operating capacity constraint set are generated.

[0010] Optionally, determining the control update cycle specifically includes: Extract current load metrics and historical load sequences from the standardized operational dataset, and generate corresponding timestamp indexes for each time point in the historical load sequence; The historical load sequence is timestamped and missing data is processed to obtain an aligned historical load sequence. Multiple load sequence fragment sets are extracted from the aligned historical load sequence based on a preset sliding time window. For each load sequence segment, a segment load statistical feature set is calculated, which includes the segment average load value, segment peak load value, segment load change amplitude value, and segment load change rate value. A load forecast sequence is generated based on the current load index and the set of statistical features of segment load. The load forecast sequence is obtained by arranging the predicted load indexes at multiple future prediction time points in chronological order. The load fluctuation adjustment is calculated based on the load forecast sequence, and then applied to the preset basic update cycle to obtain the control update cycle.

[0011] Optionally, the construction of the collaborative optimization decision model specifically includes: At the start of the control update cycle, acquire host parameters, standardized operation dataset, partial load energy efficiency characteristic model, host group collaborative energy consumption evaluation model, host operation capacity constraint set and operation stability constraint, and set host start and stop binary variables, host load distribution continuous variables and host outlet water temperature set value variables for each host, and set system side set value variables for the system side. The start / stop status of each host is determined based on the binary variable of host start / stop, and the load allocation value and allocation ratio of each host in the host group are determined based on the continuous variable of host load allocation. For each host, a host energy consumption term is constructed based on the partial load energy efficiency characteristic model, with the host load distribution continuous variable and the host outlet water temperature setpoint variable as inputs and the host input power as the output. An optimization objective function is established based on the host cluster collaborative energy consumption evaluation model, and the start-stop penalty term is incorporated into the optimization objective function. The optimization objective function includes host cluster energy consumption term, system-side transmission and distribution energy consumption term and collaborative coupling correction term. The host operating capacity constraint set and the operating stability constraint are incorporated into the constraint conditions. The operating stability constraint includes the minimum continuous operating time constraint, the minimum continuous downtime constraint, the water supply temperature setpoint change rate constraint and the differential pressure setpoint change rate constraint. A generalized disjunctive programming method is adopted to establish a power-on disjunctive branch and a power-off disjunctive branch for each host. The power-on disjunctive branch limits the host start-up / stop binary variable to the power-on state, limits the host load distribution continuous variable to the upper and lower limit range of the corresponding host load, and calls the host energy consumption item according to the partial load energy efficiency characteristic model. The power-off disjunctive branch limits the host start-up / stop binary variable to the power-off state, limits the host load distribution continuous variable to zero, and calculates the host energy consumption item according to the power-off energy consumption parameter. By combining the objective function, constraints, start-up disjunction, and stop-down disjunction, a collaborative optimization decision model is constructed.

[0012] Optionally, obtaining the optimal control strategy specifically includes: Obtain the collaborative optimization decision model, and use the Big-M transformation method for each host to transform the logical selection relationship between the power-on disjunction branch and the power-off disjunction branch into an algebraic constraint expression controlled by the host start-up and stop binary variables. The value range limit of the host load distribution continuous variable in the power-on state and the zero-taking limit in the power-off state are transformed into a set of linear inequality constraints associated with the preset large constant parameter. The optimization objective function, constraints, and linear inequality constraint set in the collaborative optimization decision model are merged to form an equivalent mixed integer optimization problem; The optimal control strategy is obtained by solving the mixed integer optimization problem. The optimal control strategy includes the values ​​of the start-stop control variables of each host, the load distribution control variables of each host, the outlet water temperature setting control variables of each host, and the system-side setting control variables.

[0013] Optionally, the set of control commands includes host start / stop control commands, host load setting commands, host outlet water temperature setting commands, supply water temperature setting commands, and differential pressure setting commands.

[0014] According to an embodiment of the present invention, a central air conditioning energy-saving control system based on multi-host collaboration includes: The parameter mapping module is used to obtain the host parameters of the central air conditioning system and establish object mapping relationships between each host and the chilled water side, cooling water side and terminal load side. The data processing module is used to collect real-time operating data of the central air conditioning system and preprocess it to generate a standardized operating dataset. The model building module is used to construct a partial load energy efficiency characteristic model, a host cluster collaborative energy consumption evaluation model, and a host operating capacity constraint set based on host parameters and standardized operating datasets, respectively. The load forecasting module is used to construct load forecasting sequences and determine control update cycles; The optimization modeling module is used to set decision variables within the control update cycle, establish start-up and shutdown disjunction branches using the generalized disjunction programming method, and construct a collaborative optimization decision model by combining the partial load energy efficiency characteristic model, the host group collaborative energy consumption evaluation model, and the host operating capacity constraint set. The optimization and solution module is used to perform Big-M transformation on the collaborative optimization decision model to form a mixed integer optimization problem and solve it to obtain the optimal control strategy; The instruction issuance module is used to generate a set of control instructions based on the optimal control strategy and issue them to the host controller and the system-side actuators.

[0015] The beneficial effects of this invention are: This application establishes an object mapping relationship between the main unit and the chilled water side, cooling water side, and terminal load side. Based on a standardized operating dataset, it constructs a partial load energy efficiency characteristic model, a main unit cluster collaborative energy consumption evaluation model, and a set of main unit operating capacity constraints, achieving a unified modeling framework from physical structure to operating data to energy consumption expression. Compared to control methods relying on empirical rules or independent optimization of individual units, this application can characterize the changes in main unit partial load efficiency, system-side transmission and distribution energy consumption, and the collaborative coupling effects between main units within the same mathematical framework. This allows the total energy consumption of the main unit cluster, system-side energy consumption, and collaborative correction terms to form an overall optimization objective, improving the global energy efficiency optimization capability in multi-main unit scenarios from a structural perspective.

[0016] At the control level, this application dynamically determines the control update cycle by constructing a load forecast sequence, enabling adaptive matching between optimization decisions and load fluctuation levels, thus reducing the lag or frequent adjustment problems caused by fixed-cycle control. In the collaborative optimization process, a generalized disjunctive programming method is introduced to uniformly model the start-up and shutdown disjunctive branches, and a solvable mixed-integer optimization problem is formed through Big-M transformation. The start-up / shutdown logic of the main unit, upper and lower load limits, total cooling capacity balance constraints, and operational stability constraints are embedded in the same optimization model. This allows for the solution of a globally consistent optimal control strategy while satisfying operational safety and stability, reducing ineffective start-ups and shutdowns and drastic load fluctuations.

[0017] Furthermore, by translating the optimal control strategy into host start / stop control commands, load setpoint commands, and supply / return water parameter setpoint commands and distributing them uniformly, a closed-loop connection is achieved between the model solution results and the actual execution layer. Under multi-host collaborative operation conditions, the system can achieve reasonable load distribution, minimize energy consumption, and maintain consistent control of operational constraints, thereby improving overall energy-saving performance and enhancing the stability and reliability of system operation. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a central air conditioning energy-saving control method based on multi-host collaboration proposed in this invention; Figure 2 This is a flowchart illustrating the construction of a collaborative optimization decision-making model for a central air conditioning energy-saving control method based on multi-host collaboration proposed in this invention. Figure 3 This is a flowchart illustrating the optimal control strategy generation process for a central air conditioning energy-saving control method based on multi-host collaboration proposed in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figures 1-3 A central air conditioning energy-saving control method based on multi-host collaboration includes the following steps: Obtain the main unit parameters of the central air conditioning system, and establish object mapping relationships between each main unit and the chilled water side, cooling water side, and terminal load side respectively; Collect real-time operating data of the central air conditioning system and preprocess it to obtain a standardized operating dataset; Based on host parameters and standardized operating datasets, construct partial load energy efficiency characteristic models, host group collaborative energy consumption evaluation models, and host operating capacity constraint sets for each host. Construct a load forecast sequence and determine the control update cycle based on the load forecast sequence; Within the control update cycle, decision variables are set, and the generalized disjunctive programming method is used to establish start-up disjunctive branches and shutdown disjunctive branches. In addition, a collaborative optimization decision model is constructed by combining the partial load energy efficiency characteristic model, the host group collaborative energy consumption evaluation model and the host operating capacity constraint set. The Big-M transformation of the collaborative optimization decision model is performed to form a mixed integer optimization problem, and the optimal control strategy is obtained by solving the problem. Based on the optimal control strategy, a set of control commands is output and sent to the host controller and the system-side actuators.

[0021] In this embodiment, the host parameters include the number of hosts, host type parameters, and system topology parameters. The establishment of object mapping relationships includes: analyzing the pipe connection structure of the central air conditioning system based on the system topology parameters to determine the physical connection nodes and flow direction relationships of each host in the chilled water circuit, cooling water circuit, and terminal load circuit; extracting the equipment identifiers and control identifiers of the chilled water supply and return nodes, cooling water supply and return nodes, and terminal load heat exchange nodes corresponding to each host, and establishing association mapping rules between the host and the above nodes; generating a structured association data table of host—chilled water side—cooling water side—terminal load side with the host as the core, forming an object mapping relationship data structure that can be used for parameter calling and constraint expression.

[0022] In this embodiment, the real-time operating data includes the start / stop status, load rate, input power, outlet water temperature, return water temperature, chilled water flow rate, cooling water flow rate, fault status indicator, and cumulative running time of each host, as well as the supply and return water temperature, supply and return water pressure difference, total flow rate, outdoor ambient temperature and current load indicators on the system side. Preprocessing includes timestamp alignment, missing data processing, abnormal data processing and dimensional normalization.

[0023] In this embodiment, the construction of the partial load energy efficiency characteristic model, the host cluster collaborative energy consumption evaluation model, and the host operating capacity constraint set specifically includes: Based on the host type parameter, a subset of data related to partial load operation is extracted from the standardized operation dataset. Samples with the start / stop status as the start state are retained. Among the retained samples, samples with fault status are removed by using the fault status identifier as the filtering condition. The filtered samples are then segmented and aggregated according to the load rate interval to form a partial load segmented sample set. The data subset includes the load rate, input power, outlet water temperature, return water temperature, chilled water flow rate, cooling water flow rate, start / stop status, and fault status identifier of the corresponding host. A partial load energy efficiency characteristic model for each host is constructed based on a partial load segmented sample set. The partial load energy efficiency characteristic model includes an energy consumption relationship expression with host load rate and inlet / outlet water temperature difference as independent variables and host input power as dependent variable, and an energy efficiency relationship expression with host load rate as independent variable and energy consumption level corresponding to unit cooling capacity as dependent variable. The unit cooling capacity is calculated based on chilled water flow rate and inlet / outlet water temperature difference, and the host input power is calculated by the energy consumption relationship expression based on host load rate and inlet / outlet water temperature difference. A collaborative energy consumption evaluation model for the host cluster is constructed based on system topology parameters and standardized operation datasets. The collaborative energy consumption evaluation model for the host cluster includes the total energy consumption expression of the host cluster, the system-side transmission and distribution energy consumption expression, and the collaborative coupling correction term expression. The total energy consumption of the host cluster is determined by the sum of the input power of each host. The system-side transmission and distribution energy consumption is determined by the total flow rate of the system side and the supply and return water pressure difference. The collaborative coupling correction term is determined by the host start-stop combination state, the host load distribution state, and the system-side supply and return water temperature state. Based on the host parameters, the rated capacity range of each host is extracted and the upper and lower limit constraints of the host load in the host operation capacity constraint set are constructed. The upper and lower limit constraints of the host load are obtained by setting a minimum load threshold and a maximum load threshold for each host and limiting the host load distribution to be between the minimum load threshold and the maximum load threshold. By combining the historical operation trajectory in the standardized operation dataset, the load adjustment rate boundary is determined and the host load change rate constraint is formed in the host operation capacity constraint set. The host load change rate constraint is obtained by limiting the load distribution change of the same host within adjacent control update cycles to not exceed a preset change rate threshold. The system energy balance relationship is constructed around the current load index, and the total system cooling capacity balance constraint is generated in the host operating capacity constraint set. The total system cooling capacity balance constraint is obtained by limiting the sum of the cooling capacity corresponding to the load allocation of each host to be consistent with the current load index. The cooling capacity corresponding to the load allocation of each host is determined by the product of the rated cooling capacity of the host and the load allocation of the host. Based on the system topology parameters and the hydraulic operating status in the standardized operating dataset, the operating boundary conditions are established, and the supply and return water temperature constraints and supply and return water pressure difference constraints in the host operating capacity constraint set are generated. The supply and return water temperature constraints are obtained by limiting the system-side supply water temperature and the system-side return water temperature to between a preset upper temperature limit and a preset lower temperature limit, respectively. The supply and return water pressure difference constraints are obtained by limiting the system-side supply and return water pressure difference to between a preset upper pressure difference limit and a preset lower pressure difference limit.

[0024] In this embodiment, determining the control update cycle specifically includes: Extract the current load index and historical load sequence from the standardized operation dataset, and generate a corresponding timestamp index for each time point in the historical load sequence. The historical load sequence is obtained by arranging the historical load index with the same index caliber as the current load index in chronological order. The historical load sequence is timestamped and missing data is processed to obtain an aligned historical load sequence. Multiple load sequence fragment sets are extracted from the aligned historical load sequence based on a preset sliding time window. The sliding time window is determined by the time window length and the sliding step size. For each load sequence segment, a set of segment load statistical characteristics is calculated. The set of segment load statistical characteristics includes the segment average load value, segment peak load value, segment load change amplitude value, and segment load change rate value. The segment average load value is obtained by summing the load indicators at each time point within the segment and dividing by the number of time points within the segment. The segment peak load value is determined by the maximum value of the load indicators within the segment. The segment load change amplitude value is determined by the difference between the maximum and minimum values ​​of the load indicators within the segment. The segment load change rate value is obtained by dividing the difference between the load indicator at the end of the segment and the load indicator at the beginning of the segment by the segment time span. A load forecast sequence is generated based on the current load index and the statistical characteristics of the load segments. The load forecast sequence is obtained by arranging the predicted load indexes at multiple future forecast time points in chronological order. The predicted load indexes include the forecast time point identifier, the predicted cooling load value, and the load change trend identifier. The predicted cooling load value is based on the current load index and is obtained by incremental adjustment according to the average load value of the segments, the amplitude of the load change in the segments, and the rate of change of the load in the segments. The load change trend identifier is used to characterize whether the load at the forecast time point is rising, falling, or stable relative to the previous forecast time point. The load fluctuation adjustment is calculated based on the load forecast sequence and applied to the preset basic update cycle to obtain the control update cycle. The load fluctuation adjustment is determined by comparing the maximum value of the absolute value of the difference between the predicted cooling load values ​​at all adjacent prediction time points in the load forecast sequence with the preset fluctuation threshold. When the maximum value is greater than the preset fluctuation threshold, the preset basic update cycle is shortened; when the maximum value is not greater than the preset fluctuation threshold, the preset basic update cycle is extended.

[0025] In this embodiment, the construction of the collaborative optimization decision-making model specifically includes: At the start of the control update cycle, acquire host parameters, standardized operation dataset, partial load energy efficiency characteristic model, host group collaborative energy consumption evaluation model, host operation capacity constraint set and operation stability constraint, and set host start and stop binary variables, host load distribution continuous variables and host outlet water temperature set value variables for each host, and set system side set value variables for the system side, including water supply temperature set value and differential pressure set value. The start / stop status of each host is determined based on the binary variable of host start / stop, and the load allocation value and allocation ratio of each host in the host group are determined based on the continuous variable of host load allocation. For each host, a host energy consumption term is constructed based on the partial load energy efficiency characteristic model, with the host load distribution continuous variable and the host outlet water temperature setpoint variable as inputs and the host input power as output. The host energy consumption term is used to characterize the input power consumption of the host under the corresponding load distribution value and the corresponding outlet water temperature setpoint. The host group energy consumption term is obtained by summing the host energy consumption terms of each host in the host group. An optimization objective function is established based on the host cluster collaborative energy consumption evaluation model, and the start-stop penalty term is incorporated into the optimization objective function. The optimization objective function includes the host cluster energy consumption term, the system-side transmission and distribution energy consumption term, and the collaborative coupling correction term. The start-stop penalty term is determined by the number of host start-stop state changes and the start-stop interval index within the control update cycle. The start-stop interval index is determined by comparing the time interval between two adjacent start-stop state changes with the preset minimum start-stop interval threshold. The set of constraints on host operating capacity and operating stability are combined into the constraint conditions. The operating stability constraints include minimum continuous operating time constraint, minimum continuous downtime constraint, water supply temperature setpoint change rate constraint and differential pressure setpoint change rate constraint. The generalized disjunctive programming method is adopted to establish a power-on disjunctive branch and a shutdown disjunctive branch for each host. The power-on disjunctive branch limits the host start-up / stop binary variable to the power-on state, limits the host load distribution continuous variable to the upper and lower limit range of the corresponding host load, and calls the host energy consumption item according to the partial load energy efficiency characteristic model. The shutdown disjunctive branch limits the host start-up / stop binary variable to the shutdown state, limits the host load distribution continuous variable to zero, and calculates the host energy consumption item according to the shutdown energy consumption parameter. By combining the objective function, constraints, start-up disjunction, and stop-down disjunction, a collaborative optimization decision model is constructed.

[0026] In this embodiment, obtaining the optimal control strategy specifically includes: To obtain a collaborative optimization decision model, the Big-M transformation method is used for each host to transform the logical selection relationship between the power-on disjunction branch and the power-off disjunction branch into an algebraic constraint expression controlled by the host start-up and power-off binary variables. The value range limit of the host load distribution continuous variable in the power-on state and the zero-taking limit in the power-off state are transformed into a set of linear inequality constraints associated with a preset large constant parameter. The preset large constant parameter is used to map the logical selection conditions into a solvable algebraic constraint form. The objective function, constraints, and linear inequality constraint set in the collaborative optimization decision model are merged to form an equivalent mixed integer optimization problem. The objective function of the mixed integer optimization problem is the same as the objective function, and the constraints of the mixed integer optimization problem include both the constraints and the linear inequality constraint set. Solving the mixed integer optimization problem yields the optimal control strategy, which includes the values ​​of start / stop control variables for each host, the values ​​of load distribution control variables for each host, the values ​​of outlet water temperature setting control variables for each host, and the values ​​of system-side setting control variables. The solution process includes: standardizing the mixed integer optimization problem by unifying the objective function and constraints into a computable matrix expression; classifying the variables into integer and continuous variables based on the binary variables of host start / stop and the continuous variables of host load allocation; setting an initial solution for iteration and initializing the current optimal objective value; using a branch and bound method to divide the integer variables into branches; performing linear relaxation on the continuous variables at each branch node to obtain the current node objective value; comparing the current node objective value with the current optimal objective value and updating the current optimal objective value and control variable values; pruning the branch when the node objective value does not meet the preset pruning conditions; and outputting the corresponding host start / stop control variable values, host load allocation control variable values, host outlet water temperature setting control variable values, and system-side setting control variable values ​​after all branch nodes have been traversed.

[0027] In this embodiment, the control command set includes main unit start / stop control command, main unit load setting command, main unit outlet water temperature setting command, water supply temperature setting command, and differential pressure setting command.

[0028] A central air conditioning energy-saving control system based on multi-host collaboration includes: The parameter mapping module is used to obtain the host parameters of the central air conditioning system and establish object mapping relationships between each host and the chilled water side, cooling water side and terminal load side. The data processing module is used to collect real-time operating data of the central air conditioning system and preprocess it to generate a standardized operating dataset. The model building module is used to construct a partial load energy efficiency characteristic model, a host cluster collaborative energy consumption evaluation model, and a host operating capacity constraint set based on host parameters and standardized operating datasets, respectively. The load forecasting module is used to construct load forecasting sequences and determine control update cycles; The optimization modeling module is used to set decision variables within the control update cycle, establish start-up and shutdown disjunction branches using the generalized disjunction programming method, and construct a collaborative optimization decision model by combining the partial load energy efficiency characteristic model, the host group collaborative energy consumption evaluation model, and the host operating capacity constraint set. The optimization and solution module is used to perform Big-M transformation on the collaborative optimization decision model to form a mixed integer optimization problem and solve it to obtain the optimal control strategy; The instruction issuance module is used to generate a set of control instructions based on the optimal control strategy and issue them to the host controller and the system-side actuators.

[0029] Example 1: To verify the feasibility of this invention in practice, it was applied to the central air conditioning system of a large commercial complex. This complex is located in the core urban area of ​​East China, with a large building area, including office, commercial, and dining areas. The load fluctuates frequently and exhibits significant time-of-day variations. The system is equipped with multiple chiller units of different models, connected to the terminal load side via chilled water circuits, and the cooling water side operates in conjunction with the cooling tower system. Previously, the system primarily used a fixed rotation start-stop and experience-based load distribution method. During seasonal transitions and periods of high population density, some chiller units were prone to prolonged low-load operation, frequent start-stop cycles, and large fluctuations in supply and return water temperatures, resulting in high system energy consumption and insufficient equipment operational stability.

[0030] In this scenario, real-time operational data, including the start / stop status, load rate, input power, inlet and outlet water temperatures, system-side supply and return water pressure difference, total flow rate, and current load indicators of each host, are first collected and standardized operational datasets are generated through time alignment and anomaly handling. Based on this, partial load energy efficiency characteristic models for each host are constructed, and a collaborative energy consumption evaluation model for the host group is established in conjunction with the system topology. Simultaneously, a set of host operating capacity constraints and operational stability constraints are generated. The system constructs a load prediction sequence based on historical load change trends and dynamically determines the control update cycle, ensuring that the optimization frequency matches the degree of load fluctuation. Within each control update cycle, start-up and shutdown disjunctive branches are established through generalized disjunctive programming. The host start / stop logic and load allocation range are embedded into the collaborative optimization decision model, and a mixed-integer optimization problem is formed through Big-M transformation for solution, yielding the optimal control strategy. Finally, host start / stop control commands, load setpoint commands, and supply and return water setpoint commands are issued to the controller for execution.

[0031] After a period of continuous operation, comparison with historical operating records under similar load conditions in the same season revealed a significant reduction in the number of start-ups and shutdowns of the main units, a reduction in low-load operating periods, a downward trend in the total energy consumption of the main unit group and the system-side transmission and distribution energy consumption, a convergence in the temperature fluctuation range of the supply and return water, and a more stable operating curve. The cumulative operating time of the equipment was more evenly distributed, preventing individual main units from operating at high loads for extended periods. Operation and maintenance records indicate a decrease in system energy consumption and improved operational stability, verifying the effectiveness of this invention in achieving global optimization control and energy-saving operation in multi-main unit collaborative scenarios.

[0032] Table 1. Performance Comparison of the Invention and Traditional Central Air Conditioning Energy-Saving Control Methods

[0033] As can be clearly seen from Table 1, the method of the present invention is superior to the traditional method in many indicators.

[0034] Under the same seasonal and similar load conditions, the average daily total energy consumption of the traditional method is 48250 kWh, while that of the method of this invention is 46890 kWh, representing a reduction of approximately 2.8%. This improvement stems from the fact that the collaborative optimization decision model uniformly considers both the total energy consumption of the main unit cluster and the system-side transmission and distribution energy consumption in each control update cycle, making the main unit operating point closer to the high-efficiency range of the partial load energy efficiency characteristic model, thereby achieving a stable and non-excessive energy-saving effect. Correspondingly, the average COP of the main unit cluster increases from 4.62 to 4.75, an increase of approximately 2.8%, indicating that the input power is reduced under the same cooling output conditions, and the increase is consistent with the proportion of the decrease in total energy consumption.

[0035] The average number of daily start-stop cycles decreased from 18 to 15. This change is related to the incorporation of the start-stop penalty term into the optimization objective function. In solving the mixed-integer optimization problem, the binary variables of start-stop are constrained overall, suppressing frequent switching behavior. The proportion of low-load operation time decreased from 27.4% to 24.6%, a drop of 2.8 percentage points. This indicates that the continuous variable of host load distribution is more reasonable under the combined effect of the host operating capacity constraint set and the system total cooling capacity balance constraint, preventing some hosts from operating in a low-efficiency range for extended periods.

[0036] The temperature difference fluctuation range of the supply and return water decreased from 1.6℃ to 1.3℃, and the proportion of energy consumption in system-side transmission and distribution decreased from 21.3% to 20.6%. The reduced temperature difference fluctuation indicates that after the control update cycle was dynamically adjusted according to the load forecast sequence, the constraints on the rate of change of the supply water temperature setpoint and the rate of change of the differential pressure setpoint were effectively executed, making the system operation more stable. The slight decrease in the proportion of transmission and distribution energy consumption is related to the participation of the collaborative coupling correction term in the integrated optimization of the main unit group collaborative energy consumption evaluation model, making the energy consumption relationship between the main unit side and the system side more coordinated.

[0037] The load distribution standard deviation decreased from 8.4% to 7.2%, indicating a more balanced load distribution among the hosts. The average control update cycle decreased from 20 minutes to 16 minutes, demonstrating that appropriately increasing the control frequency during load fluctuations and maintaining a reasonable cycle length during load stability achieves continuous energy savings while ensuring operational stability. The overall data variation was moderate, consistent with actual engineering operation levels, verifying the effectiveness of this invention in achieving steady-state optimization and energy efficiency improvement in multi-host collaborative scenarios.

[0038] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A central air conditioning energy-saving control method based on multi-host collaboration, characterized in that, Includes the following steps: Obtain the main unit parameters of the central air conditioning system, and establish object mapping relationships between each main unit and the chilled water side, cooling water side, and terminal load side respectively; Collect real-time operating data of the central air conditioning system and preprocess it to obtain a standardized operating dataset; Based on host parameters and standardized operating datasets, construct partial load energy efficiency characteristic models, host group collaborative energy consumption evaluation models, and host operating capacity constraint sets for each host. Construct a load forecast sequence and determine the control update cycle based on the load forecast sequence; Within the control update cycle, decision variables are set, and the generalized disjunctive programming method is used to establish start-up disjunctive branches and shutdown disjunctive branches. In addition, a collaborative optimization decision model is constructed by combining the partial load energy efficiency characteristic model, the host group collaborative energy consumption evaluation model and the host operating capacity constraint set. The Big-M transformation of the collaborative optimization decision model is performed to form a mixed integer optimization problem, and the optimal control strategy is obtained by solving the problem. Based on the optimal control strategy, a set of control commands is output and sent to the host controller and the system-side actuators.

2. The energy-saving control method for central air conditioning based on multi-host collaboration according to claim 1, characterized in that, The host parameters include the number of hosts, host type parameters, and system topology parameters.

3. The energy-saving control method for central air conditioning based on multi-host collaboration according to claim 1, characterized in that, The real-time operating data includes the start / stop status, load rate, input power, outlet water temperature, return water temperature, chilled water flow rate, cooling water flow rate, fault status indicator, and cumulative running time of each host, as well as the supply and return water temperature, supply and return water pressure difference, total flow rate, outdoor ambient temperature, and current load indicators on the system side. The preprocessing includes timestamp alignment, missing data processing, abnormal data processing, and dimensional normalization.

4. The energy-saving control method for central air conditioning based on multi-host collaboration according to claim 1, characterized in that, The construction of the partial load energy efficiency characteristic model, the host cluster collaborative energy consumption evaluation model, and the host operating capacity constraint set specifically includes: Based on the host type parameter, a subset of data related to partial load operation is extracted from the standardized operation dataset. Samples with the start-up status are retained. Among the retained samples, samples with fault status are removed by using the fault status identifier as the filtering condition. The filtered samples are then segmented and aggregated according to the load rate interval to form a partial load segmented sample set. A partial load energy efficiency characteristic model for each host is constructed based on a partial load segmented sample set. The partial load energy efficiency characteristic model includes an energy consumption relationship expression with host load rate and inlet / outlet water temperature difference as independent variables and host input power as dependent variable, and an energy efficiency relationship expression with host load rate as independent variable and energy consumption level corresponding to unit cooling capacity as dependent variable. A host cluster collaborative energy consumption evaluation model is constructed based on system topology parameters and standardized operation dataset. The host cluster collaborative energy consumption evaluation model includes a host cluster total energy consumption expression, a system-side transmission and distribution energy consumption expression, and a collaborative coupling correction term expression. Based on the host parameters, extract the rated capacity range of each host and construct the host load upper and lower limit constraints in the host operating capacity constraint set; By combining the historical operating trajectories in the standardized operating dataset, the load adjustment rate boundary is determined and load change rate constraints for each host are formed in the host operating capacity constraint set. Construct a system energy balance relationship based on the current load index and generate a system total cooling capacity balance constraint in the host operating capacity constraint set; Based on the system topology parameters and the hydraulic operating states in the standardized operating dataset, the operating boundary conditions are established, and the supply and return water temperature constraints and supply and return water pressure difference constraints in the host operating capacity constraint set are generated.

5. The energy-saving control method for central air conditioning based on multi-host collaboration according to claim 1, characterized in that, The determination of the control update cycle specifically includes: Extract current load metrics and historical load sequences from the standardized operational dataset, and generate corresponding timestamp indexes for each time point in the historical load sequence; The historical load sequence is timestamped and missing data is processed to obtain an aligned historical load sequence. Multiple load sequence fragment sets are extracted from the aligned historical load sequence based on a preset sliding time window. For each load sequence segment, a segment load statistical feature set is calculated, which includes the segment average load value, segment peak load value, segment load change amplitude value, and segment load change rate value. A load forecast sequence is generated based on the current load index and the set of statistical features of segment load. The load forecast sequence is obtained by arranging the predicted load indexes at multiple future prediction time points in chronological order. The load fluctuation adjustment is calculated based on the load forecast sequence, and then applied to the preset basic update cycle to obtain the control update cycle.

6. The energy-saving control method for central air conditioning based on multi-host collaboration according to claim 1, characterized in that, The construction of the collaborative optimization decision-making model specifically includes: At the start of the control update cycle, acquire host parameters, standardized operation dataset, partial load energy efficiency characteristic model, host group collaborative energy consumption evaluation model, host operation capacity constraint set and operation stability constraint, and set host start and stop binary variables, host load distribution continuous variables and host outlet water temperature set value variables for each host, and set system side set value variables for the system side. The start / stop status of each host is determined based on the binary variable of host start / stop, and the load allocation value and allocation ratio of each host in the host group are determined based on the continuous variable of host load allocation. For each host, a host energy consumption term is constructed based on the partial load energy efficiency characteristic model, with the host load distribution continuous variable and the host outlet water temperature setpoint variable as inputs and the host input power as the output. An optimization objective function is established based on the host cluster collaborative energy consumption evaluation model, and the start-stop penalty term is incorporated into the optimization objective function. The optimization objective function includes host cluster energy consumption term, system-side transmission and distribution energy consumption term and collaborative coupling correction term. The host operating capacity constraint set and the operating stability constraint are incorporated into the constraint conditions. The operating stability constraint includes the minimum continuous operating time constraint, the minimum continuous downtime constraint, the water supply temperature setpoint change rate constraint and the differential pressure setpoint change rate constraint. A generalized disjunctive programming method is adopted to establish a power-on disjunctive branch and a power-off disjunctive branch for each host. The power-on disjunctive branch limits the host start-up / stop binary variable to the power-on state, limits the host load distribution continuous variable to the upper and lower limit range of the corresponding host load, and calls the host energy consumption item according to the partial load energy efficiency characteristic model. The power-off disjunctive branch limits the host start-up / stop binary variable to the power-off state, limits the host load distribution continuous variable to zero, and calculates the host energy consumption item according to the power-off energy consumption parameter. By combining the objective function, constraints, start-up disjunction, and stop-down disjunction, a collaborative optimization decision model is constructed.

7. The energy-saving control method for central air conditioning based on multi-host collaboration according to claim 1, characterized in that, The optimal control strategy is obtained specifically through: Obtain the collaborative optimization decision model, and use the Big-M transformation method for each host to transform the logical selection relationship between the power-on disjunction branch and the power-off disjunction branch into an algebraic constraint expression controlled by the host start-up and stop binary variables. The value range limit of the host load distribution continuous variable in the power-on state and the zero-taking limit in the power-off state are transformed into a set of linear inequality constraints associated with the preset large constant parameter. The optimization objective function, constraints, and linear inequality constraint set in the collaborative optimization decision model are merged to form an equivalent mixed integer optimization problem; The optimal control strategy is obtained by solving the mixed integer optimization problem. The optimal control strategy includes the values ​​of the start-up and shutdown control variables of each host, the load distribution control variables of each host, the outlet water temperature setting control variables of each host, and the system-side setting control variables.

8. The energy-saving control method for central air conditioning based on multi-host collaboration according to claim 1, characterized in that, The set of control commands includes main unit start / stop control commands, main unit load setting commands, main unit outlet water temperature setting commands, water supply temperature setting commands, and differential pressure setting commands.

9. A central air conditioning energy-saving control system based on multi-host collaboration, comprising the central air conditioning energy-saving control method based on multi-host collaboration as described in any one of claims 1 to 8, characterized in that, include: The parameter mapping module is used to obtain the host parameters of the central air conditioning system and establish object mapping relationships between each host and the chilled water side, cooling water side and terminal load side. The data processing module is used to collect real-time operating data of the central air conditioning system and preprocess it to generate a standardized operating dataset. The model building module is used to construct a partial load energy efficiency characteristic model, a host cluster collaborative energy consumption evaluation model, and a host operating capacity constraint set based on host parameters and standardized operating datasets, respectively. The load forecasting module is used to construct load forecasting sequences and determine control update cycles; The optimization modeling module is used to set decision variables within the control update cycle, establish start-up and shutdown disjunction branches using the generalized disjunction programming method, and construct a collaborative optimization decision model by combining the partial load energy efficiency characteristic model, the host group collaborative energy consumption evaluation model, and the host operating capacity constraint set. The optimization and solution module is used to perform Big-M transformation on the collaborative optimization decision model to form a mixed integer optimization problem and solve it to obtain the optimal control strategy; The instruction issuance module is used to generate a set of control instructions based on the optimal control strategy and issue them to the host controller and the system-side actuators.

Images