Central air conditioning cooling water system global collaborative optimization control method and system

Abstract

The application discloses a kind of global collaborative optimization control method and system of central air conditioning cooling water system, the method obtains system current operating condition and the energy consumption model of cooling water pump group and cooling tower group of cooling water main engine;With cooling loop temperature difference and cooling tower approximation degree as control variable, candidate parameter combination is generated;Under the condition that cold load is unchanged, the cooling water pump group and cooling tower group power corresponding to each combination is sequentially predicted, the cooling side heat dissipation and cooling water flow are calculated, and the cooling water pump group and cooling tower group power are predicted;The sum of the power of the three types of equipment is used as the predicted total power of the system, and the target parameter combination with the lowest total power is selected to control the cooling water pump group and the cooling tower group.The energy consumption coupling relationship of the three types of equipment is included in the unified evaluation, which can avoid the increase of energy consumption of other equipment caused by local energy saving, realize collaborative control and reduce the overall operation energy consumption of the system.

Description

Technical Field

[0001] This invention relates to the field of central air conditioning cooling water system control technology, and in particular to a global collaborative optimization control method and system for a central air conditioning cooling water system. Background Technology

[0002] Central air conditioning cooling water systems typically include chillers, cooling water pump sets, and cooling tower sets. Their operating energy consumption directly impacts the overall energy consumption of the central air conditioning system. Existing cooling water systems mostly employ constant temperature difference and constant approximation degree control, meaning the cooling water pump sets are controlled according to a fixed cooling loop temperature difference, and the cooling tower sets are controlled according to a fixed cooling tower approximation degree. While this method is simple and reliable, the relevant setpoints are usually determined based on design conditions or operating experience, making it difficult to adapt to dynamic changes in operating conditions such as cooling load and outdoor wet-bulb temperature.

[0003] Some existing technologies divide operating conditions into levels based on outdoor temperature or load rate and configure corresponding control parameters for each level. However, the operating conditions and control parameters still rely on manual tuning, resulting in problems such as delayed level switching and difficulty in covering actual operating conditions. Other existing technologies mainly optimize the number of operating chillers or load distribution, while the cooling water pump sets and cooling tower sets still operate with fixed parameters.

[0004] However, there is an energy consumption coupling relationship between the chiller, cooling water pump set, and cooling tower set. Changes in the temperature difference of the cooling circuit and the proximity of the cooling tower not only affect the power of the cooling water pump set and cooling tower set, but also affect the condensing conditions and power of the chiller. Adjusting only one type of equipment independently may result in a decrease in the energy consumption of that type of equipment while the energy consumption of other equipment increases, making it difficult to achieve a low overall system energy consumption level. Therefore, it is necessary to perform global coordinated optimization of the chiller, cooling water pump set, and cooling tower set. Summary of the Invention

[0005] The purpose of this invention is to provide a global collaborative optimization control method and system for a central air conditioning cooling water system that comprehensively considers the energy consumption coupling relationship between the chiller, cooling water pump group and cooling tower group, and determines the control parameters with the goal of minimizing the total system power, in order to solve the above-mentioned technical problems.

[0006] To achieve the above objectives, the present invention provides a global collaborative optimization control method for a central air conditioning cooling water system, comprising: Obtain the current operating conditions of the cooling water system, as well as the first energy consumption model, the second energy consumption model, and the third energy consumption model corresponding to the chiller, the cooling water pump group, and the cooling tower group, respectively; Generate multiple candidate parameter combinations with cooling loop temperature difference and cooling tower approximation degree as control variables; Under the condition that the current cooling load remains unchanged, based on the current operating conditions and any of the candidate parameter combinations, the power of the chiller is predicted using the first energy consumption model. The cooling side heat dissipation is determined based on the current cooling load and the predicted power of the chiller, and the cooling water flow rate is determined based on the cooling side heat dissipation and the corresponding cooling circuit temperature difference. Based on the cooling water flow rate, the power of the cooling water pump set is predicted using the second energy consumption model; Based on the heat dissipation on the cooling side, the current operating conditions, and the corresponding candidate parameter combinations, the power of the cooling tower group is predicted using the third energy consumption model. The sum of the power of the chiller unit, the power of the cooling water pump group, and the power of the cooling tower group is determined as the predicted total power of the system for the corresponding candidate parameter combination; The system selects the target parameter combination with the lowest predicted total power from the multiple candidate parameter combinations, and controls the cooling water pump group and cooling tower group according to the target parameter combination.

[0007] Preferably, the first energy consumption model, the second energy consumption model, and the third energy consumption model are obtained in the following manner: Obtain the historical operating data of the chiller, cooling water pump set, and cooling tower set respectively; The historical operating data of the chiller, cooling water pump group and cooling tower group are cleaned and fuzzy sub-clustering are processed to obtain multiple typical operating condition cluster centers for each. Based on the cluster centers of the multiple typical operating conditions, fuzzy inference models corresponding to the chiller, cooling water pump group and cooling tower group are established respectively. The model parameters of each fuzzy inference model are determined by least squares parameter identification to obtain the first energy consumption model, the second energy consumption model and the third energy consumption model.

[0008] Preferably, it also includes: New operating samples of the chiller, cooling water pump group and cooling tower group are obtained, and the sample density value of the new operating samples and the degree of correlation between the new operating samples and the corresponding existing typical operating condition cluster centers are calculated. Based on the sample density value and correlation degree, the newly added operating samples are merged into the corresponding existing typical operating condition cluster centers, or new typical operating condition cluster centers are generated based on the newly added operating samples, and existing typical operating condition cluster centers that do not meet the preset retention conditions are eliminated. The first, second, or third energy consumption model is updated based on the updated typical operating condition cluster centers.

[0009] Preferably, the first energy consumption model takes the load rate of the chiller, the evaporator inlet water temperature, the evaporator outlet water temperature, the condenser inlet water temperature and the condenser outlet water temperature as inputs, and the chiller power as output. The second energy consumption model takes cooling water flow rate as input and cooling water pump set speed ratio and cooling water pump set power as output; The third energy consumption model takes the heat dissipation on the cooling side, the temperature difference in the cooling circuit, the inlet water temperature of the cooling tower, the outdoor wet-bulb temperature, and the cooling tower approximation degree as inputs, and the speed ratio of the cooling tower group and the power of the cooling tower group as outputs.

[0010] Preferably, the method for generating multiple candidate parameter combinations with cooling loop temperature difference and cooling tower approximation degree as control variables, and selecting the target parameter combination from the multiple candidate parameter combinations includes: A particle population is generated using the temperature difference in the cooling circuit and the proximity of the cooling tower as the position parameters of the particles. The total predicted power of the system for the candidate parameter combination corresponding to the position parameters of each particle is determined as the fitness value of the particle. The individual optimal position of each particle and the global optimal position of the particle population are updated based on the fitness value of each particle, and the velocity and position of each particle are updated based on the individual optimal position and the global optimal position. When the preset number of iterations is reached or the preset convergence condition is met, the candidate parameter combination corresponding to the global optimal position is determined as the target parameter combination.

[0011] Preferably, during the process of determining the target parameter combination using the particle population, each candidate parameter combination is subjected to equipment operation constraint verification, wherein the equipment operation constraints include at least one of the following: cooling water pump group operating frequency constraint, cooling tower group operating frequency constraint, cooling water flow rate constraint, and cooling tower inlet water temperature constraint; candidate parameter combinations that do not meet the equipment operation constraints are excluded, or a penalty is imposed on the fitness value of candidate parameter combinations that do not meet the equipment operation constraints.

[0012] Preferably, before generating the plurality of candidate parameter combinations, the method further includes: Ensure that the time interval between the current moment and the moment when the control command was last successfully issued is not less than the preset minimum control interval; It is determined that at least one of the chiller units is in cooling operation, the cooling water pump group is in synchronous frequency conversion temperature difference control, and the cooling tower group is in synchronous frequency conversion approximation control. The chilled water outlet temperature, cooling circuit temperature difference, and cooling tower proximity of the chiller are obtained and calculated within the preset statistical period. When the range of each parameter is not greater than the corresponding preset fluctuation threshold, the cooling water system is determined to be in a steady-state operation. The current operating conditions are input into the first energy consumption model, the second energy consumption model, and the third energy consumption model to obtain the corresponding predicted power of the equipment. Based on the predicted power of each equipment and the actual power of the corresponding equipment, it is determined that the power prediction error rates of the first energy consumption model, the second energy consumption model, and the third energy consumption model are all within the corresponding preset error range. After selecting the target parameter combination, the predicted energy saving rate is determined based on the difference between the current actual total power and the system predicted total power corresponding to the target parameter combination, as well as the current actual total power. When the system predicted total power is lower than the current actual total power, the predicted energy saving rate is not less than the preset lower limit of energy saving rate and not greater than the preset upper limit of energy saving rate, and at least one of the absolute values ​​of the difference between the target cooling circuit temperature difference and the current cooling circuit temperature difference and the absolute values ​​of the difference between the target cooling tower approximation degree and the current cooling tower approximation degree is not less than the corresponding preset adjustment threshold, the cooling water pump group and the cooling tower group are controlled according to the target parameter combination.

[0013] The present invention also provides a global collaborative optimization control system for a central air conditioning cooling water system, which includes a controller that performs control of the central air conditioning cooling water system based on the global collaborative optimization control method described above.

[0014] The present invention also provides a global collaborative optimization control system for a central air conditioning cooling water system, comprising: One or more processors; Memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including instructions for performing the global cooperative optimization control method as described above.

[0015] The present invention also provides a computer-readable storage medium comprising a computer program that can be executed by a processor to perform the global cooperative optimization control method as described above.

[0016] Compared with existing technologies, the global collaborative optimization control method provided by the above technical solution uses the cooling circuit temperature difference and cooling tower approximation degree as control variables. Under the condition that the current cooling load remains unchanged, it predicts the power of the chiller, the heat dissipation of the cooling side, the cooling water flow rate, the power of the cooling water pump group, and the power of the cooling tower group in sequence based on each candidate parameter combination. This incorporates the energy consumption coupling relationship between the three types of equipment into a unified evaluation, and determines the target parameter combination based on the lowest predicted total power of the system. This achieves collaborative control of the cooling water pump group and the cooling tower group, avoiding the increase in energy consumption of other equipment due to the reduction of energy consumption of a single type of equipment. It is conducive to the system adapting to the current operating conditions and reducing the overall operating energy consumption. Attached Figure Description

[0017] Figure 1 This is a flowchart of the global collaborative optimization control method in an embodiment of the present invention.

[0018] Figure 2 This is a flowchart illustrating the construction process of various energy consumption models in the embodiments of the present invention.

[0019] Figure 3 This is a multi-layered verification process for online optimization and control command issuance in this embodiment of the invention. Detailed Implementation

[0020] To illustrate the technical content, structural features, objectives, and effects of the present invention in detail, the following description is provided in conjunction with the embodiments and accompanying drawings.

[0021] This embodiment discloses a global collaborative optimization control method for a central air conditioning cooling water system, which is applicable to central air conditioning cooling water systems in commercial buildings, industrial plants, data centers, and other similar locations.

[0022] The cooling water system includes a chiller, a cooling water pump set, and a cooling tower set. The chiller includes an evaporator and a condenser. Chilled water circulates on the evaporator side and bears the cooling load, while cooling water circulates on the condenser side and removes condensation heat.

[0023] The cooling water pump set includes one or more variable frequency cooling water pumps. The outlet of the cooling water pump set is connected to the condenser inlet of the chiller through a cooling water pipeline. The condenser outlet is connected to the inlet of the cooling tower set through a cooling water pipeline. The outlet of the cooling tower set is connected back to the inlet of the cooling water pump set, forming a cooling water circulation loop.

[0024] A cooling tower group consists of one or more variable frequency cooling towers, which use fans to drive outdoor air to exchange heat and moisture with cooling water. The cooling water system is equipped with measuring points such as temperature sensors and power acquisition devices to collect operating parameters such as condenser inlet water temperature, condenser outlet water temperature, cooling tower outlet water temperature, outdoor wet-bulb temperature, and power of each piece of equipment.

[0025] Based on this, such as Figure 1 The global collaborative optimization control method in this embodiment includes the following steps.

[0026] S1. Obtain the current operating conditions of the cooling water system, as well as the first energy consumption model, second energy consumption model, and third energy consumption model corresponding to the chiller, cooling water pump group, and cooling tower group, respectively.

[0027] Current operating conditions include parameters such as current cooling load, chiller load rate, chilled water outlet temperature, chilled water temperature difference, condenser inlet water temperature, condenser outlet water temperature, outdoor wet-bulb temperature, and the current actual power of each device.

[0028] The first energy consumption model is used to characterize the mapping relationship between the operating parameters of the chiller and the power of the chiller; the second energy consumption model is used to characterize the mapping relationship between the cooling water flow rate and the power of the cooling water pump group; and the third energy consumption model is used to characterize the mapping relationship between the operating parameters of the cooling tower group and the power of the cooling tower group.

[0029] In some optional implementations, each energy consumption model can adopt a fuzzy inference model based on cluster centers of typical operating conditions, or other forms such as multinomial regression model, support vector regression model or neural network model, as long as it can predict the power of the corresponding equipment based on the input operating parameters.

[0030] S2. Generate multiple candidate parameter combinations with cooling circuit temperature difference and cooling tower approximation degree as control variables.

[0031] It should be noted that the cooling circuit temperature difference refers to the difference between the cooling water outlet temperature and the cooling water inlet temperature of the chiller condenser, while the cooling tower proximity refers to the difference between the cooling tower outlet temperature and the outdoor wet-bulb temperature. The cooling water pump set is controlled by synchronous frequency conversion based on the cooling circuit temperature difference setpoint, and the cooling tower set is controlled by synchronous frequency conversion based on the cooling tower proximity setpoint. Therefore, the cooling circuit temperature difference and the cooling tower proximity together determine the operating condition distribution of the cooling water system.

[0032] Each candidate parameter combination includes a candidate cooling loop temperature difference value and a candidate cooling tower approximation value.

[0033] Candidate parameter combinations are generated within their respective value ranges. For example, the cooling circuit temperature difference is within the range of 3℃ to 8℃, and the cooling tower proximity is within the range of 2℃ to 6℃. The above value ranges can be adjusted according to the equipment design parameters and on-site operating experience.

[0034] In some alternative implementations, candidate parameter combinations can be generated by gridded enumeration within a range of values ​​with a fixed step size, or by random sampling. Alternatively, they can be dynamically generated during the iteration process using heuristic optimization algorithms such as particle swarm optimization or genetic algorithms, as long as multiple parameter combinations to be evaluated can be formed within the range of values.

[0035] S3. Under the condition that the current cooling load remains unchanged, based on the current operating conditions and any candidate parameter combination, use the first energy consumption model to predict the power of the chiller.

[0036] Specifically, the optimization process is performed on the current operating section, while the cooling load and chiller load rate remain unchanged based on measured values. For each candidate parameter combination, the cooling tower outlet temperature is first determined based on the candidate cooling tower approximation and the current outdoor wet-bulb temperature, and this temperature is used as the chiller condenser inlet temperature. Then, the condenser outlet temperature is determined based on the candidate cooling loop temperature difference. The current load rate, evaporator inlet temperature, evaporator outlet temperature, and the aforementioned condenser inlet and outlet temperatures are input into the first energy consumption model to obtain the predicted chiller power under that candidate parameter combination.

[0037] When multiple chillers are running in the system, the above prediction is performed on each running chiller, and the predicted power values ​​of each chiller are summed up.

[0038] Changes in the condenser inlet water temperature and the temperature difference in the cooling circuit will alter the condensing temperature and condensing pressure of the chiller, thereby changing the compressor's pressure ratio and power consumption. Therefore, under the condition of constant cooling load, different combinations of candidate parameters correspond to different chiller power.

[0039] S4. Determine the heat dissipation on the cooling side based on the current cooling load and the predicted power of the chiller, and determine the cooling water flow rate based on the heat dissipation on the cooling side and the corresponding temperature difference in the cooling circuit.

[0040] Based on the law of conservation of energy, the heat discharged from the condenser side of the chiller is equal to the sum of the cooling load absorbed by the evaporator side and the compressor input power. Therefore, the heat dissipation on the cooling side can be determined as the heat dissipation on the cooling side equals the sum of the current cooling load and the chiller power. Based on this, the cooling water flow rate is determined according to the cooling water heat transfer formula Q=c·ρ·G·ΔT, where Q is the heat dissipation on the cooling side, c is the specific heat capacity of water, ρ is the density of water, G is the cooling water flow rate, and ΔT is the temperature difference in the cooling loop in this candidate parameter combination.

[0041] In this embodiment, the temperature difference between the inlet and outlet water of the cooling tower group is approximately the same as the temperature difference of the cooling circuit, and the flow rate of the cooling water circulating through the condenser, cooling water pump group, and cooling tower group is the same. When the system is equipped with a cooling water flow meter, the above calculations can also be checked or corrected by combining the actual flow rate values.

[0042] This step establishes a definite physical relationship between the candidate parameter combinations and the cooling water flow rate. That is, when the temperature difference in the cooling circuit increases, the cooling water flow rate required to deliver the same amount of heat dissipation decreases. At the same time, since the chiller power changes with the candidate parameter combinations, the heat dissipation on the cooling side also changes accordingly, and the two together determine the cooling water flow rate.

[0043] S5. Based on the cooling water flow rate, predict the power of the cooling water pump set using the second energy consumption model.

[0044] Input the cooling water flow rate determined in step S4 above into the second energy consumption model to obtain the predicted power value of the cooling water pump group under the candidate parameter combination.

[0045] When the cooling water pump set is in synchronous variable frequency operation mode, its speed and power change with the flow rate. When the flow rate decreases, the pump set speed decreases and the power decreases. Therefore, the cooling water flow rate can characterize the operating status of the cooling water pump set.

[0046] S6. Based on the heat dissipation on the cooling side, the current operating conditions, and the corresponding candidate parameter combinations, the power of the cooling tower group is predicted using the third energy consumption model.

[0047] By inputting the heat dissipation on the cooling side, the temperature difference in the cooling circuit, the inlet water temperature of the cooling tower, the outdoor wet-bulb temperature, and the cooling tower approximation degree into the third energy consumption model, the predicted power value of the cooling tower group under this candidate parameter combination is obtained. Among them, the cooling tower inlet water temperature is determined based on the temperature difference between the cooling tower outlet water temperature and the cooling circuit.

[0048] A reduced cooling tower proximity means the cooling tower outlet water temperature is closer to the outdoor wet-bulb temperature, requiring a larger fan airflow and power. Conversely, a reduced cooling tower proximity can decrease the cooling tower group power, but it will raise the chiller condenser inlet water temperature and increase the chiller power. The third energy consumption model incorporates these factors into a unified mapping relationship.

[0049] S7. Determine the sum of the chiller power, cooling water pump power, and cooling tower power as the predicted total system power for the corresponding candidate parameter combination. For the case of multiple chillers operating, the chiller power is the sum of the predicted power values ​​of each operating chiller. Repeat steps S3 to S7 for each candidate parameter combination to obtain the predicted total system power for each combination.

[0050] S8. Select the target parameter combination with the lowest predicted total power from multiple candidate parameter combinations, and control the cooling water pump group and cooling tower group according to the target parameter combination.

[0051] Specifically, the cooling circuit temperature difference in the target parameter combination is sent as the temperature difference setpoint of the cooling water pump group, and the cooling tower approximation in the target parameter combination is sent as the approximation setpoint of the cooling tower group. The cooling water pump group and the cooling tower group are then synchronously frequency-controlled according to the updated setpoints.

[0052] For the above technical solutions, the evaluation of each candidate parameter combination is carried out in a coupled calculation according to the chain relationship of chiller power, cooling side heat dissipation, cooling water flow rate, cooling water pump power and cooling tower power. The change of chiller power is transmitted to cooling water flow rate and cooling tower heat exchange demand through cooling side heat dissipation, so that the energy consumption coupling relationship between the three types of equipment is reflected in the same objective function.

[0053] Compared to adjusting individual devices independently, this embodiment selects target parameter combinations based solely on the system's predicted total power. This avoids localized optimization biases, such as increasing the cooling loop temperature difference to reduce pump power but increasing chiller power, or decreasing the cooling tower proximity to reduce chiller power but increasing cooling tower power. This ensures the cooling water system operates at a lower overall energy consumption level. Actual engineering verification shows that this method can reduce the total energy consumption of the cooling system by 5% to 10%.

[0054] On the other hand, such as Figure 2 In order to obtain an energy consumption model with reliable accuracy and clear physical meaning, the construction methods of the first energy consumption model, the second energy consumption model and the third energy consumption model are as follows.

[0055] First, historical operating data for the chiller, cooling water pump set, and cooling tower set are acquired. This historical operating data is loaded from the time-series database and the business database, aligned with a unified timestamp. It includes parameters such as chilled water outlet temperature, cooling water inlet temperature, chilled water temperature difference, cooling circuit temperature difference, compressor load rate and power on the chiller side; cooling water flow rate, frequency ratio, and power on the cooling water pump set side; and heat dissipation, inlet and outlet water temperature difference, inlet water temperature, outdoor wet-bulb temperature, approximation degree, and power on the cooling tower set side. The historical operating data can span one or more cooling seasons to include operating samples under different loads and climatic conditions.

[0056] Secondly, the historical operating data of each device were cleaned and subjected to fuzzy sub-clustering to obtain multiple typical operating condition cluster centers for each device.

[0057] Data cleaning includes removing samples with missing values, removing outliers that exceed physically reasonable ranges, and removing unsteady-state samples during equipment start-up and shutdown transitions, thereby ensuring that the samples participating in clustering reflect the steady-state operating characteristics of the equipment. Fuzzy sub-clustering uses each running sample as a potential cluster center, calculates the sample density value based on the sample distribution within its neighborhood radius, and selects the sample with the highest density value as the first cluster center.

[0058] Then, the density values ​​of the remaining samples are adjusted according to the amount of reduction in the density of samples in the neighborhood of the cluster center. The next cluster center is then selected from the adjusted density values. This process is repeated until the highest density value of the remaining samples is lower than the termination threshold.

[0059] Each cluster center obtained in this way is a representative point of a typical working condition, and its coordinates in the input parameter space reflect the typical values ​​of each operating parameter under that type of working condition.

[0060] It should be noted that the historical operating data of the chiller, cooling water pump set and cooling tower set are clustered independently. The number of cluster centers for typical operating conditions of each device is determined by its own data distribution, and there is no need to manually specify the number of clusters or divide the operating conditions in advance.

[0061] In some alternative implementations, the neighborhood radius of fuzzy sub-clustering can be taken in the normalization range of 0.3 to 0.7, for example, 0.5. The smaller the neighborhood radius, the more cluster centers of typical working conditions are obtained and the finer the division of working conditions.

[0062] Finally, based on cluster centers of multiple typical operating conditions, fuzzy inference models corresponding to chiller, cooling water pump group and cooling tower group are established respectively. The model parameters of each fuzzy inference model are determined by least squares parameter identification, and the first energy consumption model, the second energy consumption model and the third energy consumption model are obtained.

[0063] Specifically, a membership function, such as a Gaussian membership function, is constructed around the cluster center of each typical working condition. The degree to which the input parameters fall into the neighborhood of each typical working condition is represented by the membership value. The model output is obtained by weighted summation of the local outputs corresponding to each typical working condition according to their membership degrees. The local output parameters corresponding to each typical working condition are the model parameters to be identified, which are determined by least squares parameter identification using historical operating data. Since the model output is the result of weighted prediction by the cluster centers of each typical working condition, the typical working conditions and their weights on which the model output is based under any input working condition can be traced, and their physical meaning is clear, making it easy for engineers to understand and verify.

[0064] This embodiment adopts a method of automatically extracting cluster centers of typical operating conditions from historical operating data and modeling based on them. This ensures that the number and location of typical operating conditions are determined by the actual data distribution, avoiding the boundary rigidity problem caused by manually dividing operating conditions. At the same time, the fuzzy inference model performs continuous weighted transitions between each typical operating condition, enabling energy consumption prediction to continuously cover operating conditions near the boundary of the level, thereby improving the prediction accuracy and reliability of the optimization results of the energy consumption model within the actual operating condition range.

[0065] On the other hand, considering that the cooling water system will experience equipment aging, heat exchange efficiency decline and seasonal operating condition drift during long-term operation, causing the existing typical operating condition cluster centers to gradually deviate from the actual operating conditions, this embodiment further provides an incremental update method for the typical operating condition cluster centers.

[0066] Specifically, during system operation, new operating samples from the chiller, cooling water pump group, and cooling tower group are continuously acquired, and the sample density value of the new operating samples and the degree of correlation between the new operating samples and the corresponding existing typical operating condition cluster centers are calculated.

[0067] The sample density value is calculated according to the density definition consistent with fuzzy subtraction clustering, reflecting the degree of sample clustering in the neighborhood of the newly added operating sample; the degree of association can be represented by the distance or membership degree between the newly added operating sample and the nearest existing typical working condition cluster center. The closer the distance or the higher the membership degree, the higher the degree of association.

[0068] The cluster centers for typical operating conditions are conditionally updated based on sample density and correlation. When the correlation between a newly added operating sample and an existing typical operating condition cluster center is not lower than the merging threshold, the sample is determined to belong to an existing typical operating condition, merged into that existing typical operating condition cluster center, and the location of the cluster center is corrected based on the sample, for example, by updating the cluster center coordinates using a sample-weighted method. When the correlation between a newly added operating sample and all existing typical operating condition cluster centers is lower than the merging threshold, and its sample density value is not lower than the addition threshold, a new operating condition not covered by existing typical operating conditions is determined to have appeared, and a new typical operating condition cluster center is generated based on the newly added operating sample.

[0069] After generating new typical operating condition cluster centers, the existing typical operating condition cluster centers are further validated for retention conditions. Those that do not meet the preset retention conditions are eliminated. These preset retention conditions include, for example, that the number of times the cluster center is hit by newly added running samples within the most recent preset time window is not less than the minimum hit count, or that the distance between the cluster center and the new typical operating condition cluster center is not less than the deduplication distance threshold. In cases where overall operating conditions drift due to equipment aging, old typical operating condition cluster centers are eliminated because no new samples have been hit for a long period, and are directly replaced by new typical operating condition cluster centers. In cases where seasonal changes lead to the coexistence of old and new operating conditions, both old and new typical operating condition cluster centers are retained and participate in the construction of the energy consumption model.

[0070] After updating the cluster centers for typical operating conditions, least squares parameter identification is re-executed based on the updated cluster centers to update the corresponding first, second, or third energy consumption model. The cluster centers and energy consumption models of each device are updated independently, and the update of one device does not affect the use of models for other devices.

[0071] By employing a conditional merging, addition, and elimination mechanism based on sample density values ​​and correlation degrees, the cluster centers of typical operating conditions can automatically follow the operating condition drift as new operating data accumulates, eliminating the need for full re-clustering of historical data and reducing the computational overhead of model maintenance. At the same time, it avoids the continuous participation of failed cluster centers of typical operating conditions in energy consumption prediction, thereby maintaining the prediction accuracy of the energy consumption model under equipment aging and seasonal changes, and thus maintaining the effectiveness of the optimization results.

[0072] Furthermore, the first energy consumption model takes the chiller load rate, evaporator inlet water temperature, evaporator outlet water temperature, condenser inlet water temperature, and condenser outlet water temperature as inputs, and the chiller power as output. The load rate represents the proportion of cooling load borne by the chiller; the evaporator inlet water temperature and evaporator outlet water temperature together represent the evaporator-side operating conditions; and the condenser inlet water temperature and condenser outlet water temperature together represent the condenser-side operating conditions. These five input parameters comprehensively describe the heat transfer boundary conditions on both sides of the chiller, enabling the first energy consumption model to reflect the impact of changes in condenser-side operating conditions on the chiller power. This is precisely the transmission path of the candidate parameter combinations on chiller power during the optimization process: the candidate cooling tower proximity determines the condenser inlet water temperature, and the candidate cooling loop temperature difference determines the condenser outlet water temperature; changes in both cause changes in the predicted chiller power.

[0073] The second energy consumption model takes cooling water flow rate as input and cooling water pump set speed ratio and cooling water pump set power as output.

[0074] The third energy consumption model takes the heat dissipation on the cooling side, the temperature difference in the cooling circuit, the inlet water temperature of the cooling tower, the outdoor wet-bulb temperature, and the cooling tower approximation degree as inputs, and the speed ratio of the cooling tower group and the power of the cooling tower group as outputs.

[0075] The speed ratio refers to the ratio of the equipment's operating frequency to the power frequency. The power frequency is a fixed value, such as 50Hz. The operating frequencies of the cooling water pump group and the cooling tower group vary independently, so their speed ratios are output by their respective models.

[0076] In some optional implementations, the inputs of the first energy consumption model can also include derived parameters such as the temperature difference on the refrigeration side and the temperature difference in the cooling circuit. The cooling tower inlet water temperature in the input parameters of the third energy consumption model can also be calculated from the cooling tower outlet water temperature and the temperature difference in the cooling circuit. No separate measuring points are required; it is sufficient that all input parameters can collectively characterize the operating conditions of the corresponding equipment. Those skilled in the art should understand that the membership function is not limited to Gaussian type; triangular or bell-shaped membership functions can also be used.

[0077] Since the input parameters of the first energy consumption model cover the heat transfer boundary conditions of both the evaporation and condensation sides, the influence of candidate parameter combinations on the power of the chiller can be fully captured. Since the second and third energy consumption models output the speed ratio and power simultaneously, the optimization process can obtain both the power evaluation value and the operating frequency information for constraint verification, thus supporting both the calculation of the optimization target and the verification of equipment operation constraints without adding additional calculation steps.

[0078] On the other hand, in order to quickly obtain the target parameter combination that minimizes the total predicted power of the system within the online control cycle, this embodiment uses the particle swarm optimization algorithm to generate and optimize the candidate parameter combination.

[0079] Specifically, a particle population is generated using the cooling circuit temperature difference and the cooling tower proximity as particle position parameters. The position of each particle is a two-dimensional vector, with the first dimension being the cooling circuit temperature difference and the second dimension being the cooling tower proximity. The values ​​of each dimension are limited to their respective ranges.

[0080] The initial position of the particle population can be randomly generated within the range of values, or the temperature difference of the cooling circuit and the proximity of the cooling tower in the current actual operation can be used as the initial position of one of the particles to ensure that the optimization result is not worse than the current operating conditions.

[0081] For each particle, the candidate parameter combination corresponding to its position parameter is used to obtain the system predicted total power according to the chain calculation process of steps S3 to S7 in the above embodiment, and the system predicted total power is determined as the fitness value of the particle. The lower the fitness value, the better the candidate parameter combination.

[0082] In each iteration, the individual optimal position of each particle and the global optimal position of the particle swarm are updated based on the fitness value of each particle. If the current fitness value of a particle is lower than its historical best fitness value, then the individual optimal position of that particle is updated to its current position. If the current fitness value of a particle is lower than the historical best fitness value of the swarm, then the global optimal position of that particle is updated to its current position. Subsequently, based on the individual optimal position and the global optimal position, the velocity and position of each particle are updated according to the velocity update formula and position update formula of the particle swarm optimization algorithm. Position components that exceed the range of values ​​after the update are truncated to the boundary value.

[0083] The iteration terminates when a preset number of iterations is reached or a preset convergence condition is met, and the candidate parameter combination corresponding to the global optimal position is determined as the target parameter combination. The preset convergence condition is, for example, that the improvement in the global optimal fitness value is less than a convergence threshold in several consecutive iterations.

[0084] Since the objective function for optimization is a nonlinear function that is chained together with multiple energy consumption models, its analytical gradient is difficult to obtain. Therefore, a group search method such as particle swarm optimization, which does not require gradient information, is adopted. Guided by both individual optimality and global optimality, the search is carried out in parallel in the two-dimensional decision space. This allows the optimization process to converge to the parameter combination with the lower total predicted power of the system within a limited number of evaluations, thus balancing the requirements of solution quality and real-time online computation.

[0085] In addition, considering that the candidate parameter combinations obtained through optimization must also ensure that the cooling water pump group and cooling tower group operate within the safe range of the equipment, this embodiment introduces equipment operation constraint verification during the optimization process.

[0086] Specifically, in the process of determining the target parameter combination using particle population, each candidate parameter combination is verified by equipment operation constraints. The equipment operation constraints include at least one of the following: cooling water pump group operating frequency constraints, cooling tower group operating frequency constraints, cooling water flow constraints, and cooling tower inlet water temperature constraints.

[0087] Among them, the operating frequency constraints of the cooling water pump group and the cooling tower group are verified after converting the speed ratio output by the second energy consumption model and the third energy consumption model into the operating frequency. For example, the operating frequency of the cooling water pump group is required to be in the range of 30Hz to 50Hz and the operating frequency of the cooling tower group is required to be in the range of 25Hz to 50Hz. The lower frequency limit is used to avoid the inverter from operating at low frequency for a long time and the insufficient head of the cooling water pump when operating at low frequency. The upper frequency limit corresponds to the rated operating conditions of the equipment.

[0088] The cooling water flow constraint requires that the cooling water flow calculated by the chain should not be lower than the minimum allowable flow of the chiller condenser and not higher than the maximum allowable flow of the pipeline, so as to avoid the condenser interruption protection action or the pipeline operating at excessive flow rate.

[0089] The cooling tower inlet water temperature constraint requires that the cooling tower inlet water temperature calculated based on the candidate parameter combination does not exceed the upper limit of the allowable inlet water temperature of the cooling tower packing.

[0090] For candidate parameter combinations that do not meet the equipment operation constraints, one of the following two processing methods can be adopted: First, they can be directly excluded, that is, the positions of the corresponding particles can be regenerated or pulled back to the feasible region before evaluation; Second, a penalty can be imposed on their fitness value, that is, a penalty term can be added on the basis of the system's predicted total power. The size of the penalty term can be proportional to the amount of constraint violation. For example, a fixed penalty value much larger than the normal total power can be added to each violated constraint, or the product of the constraint violation amount and the penalty coefficient can be added.

[0091] In a specific application scenario, the cooling loop temperature difference of a certain candidate parameter combination is 7.5℃. The cooling water flow rate obtained by chain calculation is lower than the minimum flow limit of the condenser. At this time, the fitness value of the candidate parameter combination is superimposed with a penalty term. In subsequent iterations, the particle gradually moves towards the direction with smaller temperature difference and larger flow rate. Finally, the globally optimal position that converges satisfies all equipment operation constraints.

[0092] In this embodiment, because each candidate parameter combination is verified for equipment operation constraints during the optimization iteration process, the output target parameter combination is naturally within the allowable operating range of the equipment. This avoids the situation where the optimization result is discarded as a whole due to exceeding the limit in the first optimization and then verification method. Thus, control parameters that can be directly issued and executed are obtained under the premise of ensuring the safety of water supply to the chiller condenser and the safe operation of the frequency converter.

[0093] In another preferred embodiment, considering that the results of online optimization directly affect the actual operating cooling water system, in order to ensure the safety and effectiveness of the control command issuance, this embodiment sets up a multi-layer verification mechanism before and after optimization, such as... Figure 3 Control is only executed when all layers of verification pass. That is, control interval verification, operational status verification, stability verification, and model accuracy verification are added before generating multiple candidate parameter combinations, and energy-saving space verification is added after selecting the target parameter combination.

[0094] The first layer is the control interval verification (Y1). It queries the last successfully issued control command and ensures that the time interval between the current time and that time is not less than the preset minimum control interval. For example, the preset minimum control interval is 15 to 30 minutes. If the time interval is shorter than this value, the current optimization process is terminated. Due to the large thermal inertia of the cooling water system, it takes a certain amount of time for the setpoint to reach a new steady state after an update. This layer of verification avoids excessively frequent control commands that could impact the equipment and prevent the superposition of new disturbances during the transition process.

[0095] The second layer is operational status verification (Y2). It ensures that at least one chiller is in cooling operation, the cooling water pump group is in synchronous variable frequency temperature difference control, and the cooling tower group is in synchronous variable frequency approximation control. If all chillers are shut down or in heating mode, or if there are power frequency operating devices in the cooling water pump group or cooling tower group, or if no corresponding control method is used, the target parameter combination cannot be effective through the existing control loop, and this optimization process terminates.

[0096] The third layer is stability verification (Y3). The range of parameters for the chilled water outlet temperature, cooling circuit temperature difference, and cooling tower approach degree of the chiller is acquired and calculated within a preset statistical period (for a given parameter, the range is the difference between its maximum and minimum values). If the range of each parameter is no greater than the corresponding preset fluctuation threshold, the cooling water system is determined to be in steady-state operation. For example, the preset statistical period is the most recent 15 minutes, the fluctuation threshold for the chilled water outlet temperature is 0.5℃, and the fluctuation thresholds for the cooling circuit temperature difference and cooling tower approach degree are each 1.0℃. If the range of any parameter exceeds the corresponding fluctuation threshold, it indicates that the system is in a start-stop transition or a stage of drastic change in operating conditions. The operating data at this time cannot represent the steady-state condition, and optimization based on transient data may generate erroneous control commands. Therefore, the current process is terminated, and the system waits for the next trigger cycle.

[0097] The fourth layer is model accuracy verification (Y4). The current operating conditions are input into the first, second, and third energy consumption models to obtain the corresponding predicted power for each device. Based on the predicted power and the actual power of each device, the power prediction error rate of each energy consumption model is determined to be within a preset error range. The power prediction error rate can be calculated as the ratio of the absolute value of the difference between the predicted power and the actual power to the actual power. The preset error range is, for example, no more than 10%. If the power prediction error rate of any energy consumption model exceeds the preset error range, it indicates that the model is unreliable under the current operating conditions, the optimization results based on this model are untrustworthy, the current process is terminated, and a model update may be triggered.

[0098] The fifth layer is the energy-saving space verification (Y5). After selecting the target parameter combination, the predicted energy saving rate is determined based on the difference between the current actual total power and the system predicted total power corresponding to the target parameter combination, as well as the current actual total power. The predicted energy saving rate is equal to the difference divided by the current actual total power.

[0099] When the following conditions are met simultaneously, the cooling water pump group and cooling tower group are controlled according to the target parameter combination: the predicted total power of the system is lower than the current actual total power; the predicted energy saving rate is not less than the preset lower limit of energy saving rate and not greater than the preset upper limit of energy saving rate, for example, the preset lower limit of energy saving rate is 1% and the preset upper limit of energy saving rate is 30%; at least one of the absolute values ​​of the difference between the target cooling circuit temperature difference and the current cooling circuit temperature difference and the absolute values ​​of the difference between the target cooling tower approximation degree and the current cooling tower approximation degree is not less than the corresponding preset adjustment threshold, for example, each of the two preset adjustment thresholds is 0.3℃.

[0100] If the predicted energy saving rate is lower than the lower limit or the parameter adjustment amount is less than the adjustment threshold, it indicates that the energy saving benefit is limited and the adjustment action will still cause system disturbance, so this issuance will be skipped.

[0101] If the predicted energy saving rate is higher than the upper limit, it indicates that the prediction result deviates significantly from the normal energy saving level, which may be due to model error or data anomaly. In this case, no control command will be issued to prevent control actions based on distorted predictions.

[0102] In this embodiment, a multi-layered verification mechanism is set up, including control interval verification, operation status verification, stability verification, model accuracy verification, and energy-saving space verification. This ensures that the optimization process is executed only when the system meets the optimization conditions, the model is reliable, and there is effective energy-saving space. Risk factors such as model inaccuracy, transient data, and abnormal predictions are intercepted layer by layer before the control is issued. This ensures the safety and reliability of the control commands while achieving energy-saving optimization, and avoids invalid or harmful parameter adjustments.

[0103] In summary, the global collaborative optimization control method for central air conditioning cooling water systems provided by this invention uses the cooling loop temperature difference and cooling tower approximation degree as control variables. Through chain-coupled calculations of chiller power, cooling-side heat dissipation, cooling water flow rate, cooling water pump power, and cooling tower power, the energy consumption of the chiller, cooling water pump, and cooling tower is incorporated into a unified system predicted total power for evaluation. The method then selects the target parameter combination with the lowest predicted total power for execution control, thus solving the problem of local optimization deviations caused by independent adjustment of individual equipment. Practical engineering verification shows that the method of this invention can effectively reduce the total energy consumption of the cooling-side system.

[0104] In another preferred embodiment of the present invention, a global collaborative optimization control system for a central air conditioning cooling water system is also disclosed, which includes a controller that performs control of the central air conditioning cooling water system based on the global collaborative optimization control method in the above embodiments.

[0105] This invention also discloses another global collaborative optimization control system for a central air conditioning cooling water system, comprising one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors. The programs include instructions for executing the global collaborative optimization control method as described above. The processor may be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, used to execute relevant programs to achieve the functions required by the modules in the global collaborative optimization control system of this application embodiment, or to execute the global collaborative optimization control method of this application embodiment.

[0106] This invention also discloses a computer-readable storage medium comprising a computer program executable by a processor to perform the global cooperative optimization control method described above. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center integrating one or more available media. The available medium can be read-only memory (ROM), random access memory (RAM), or magnetic media, such as floppy disks, hard disks, magnetic tapes, magnetic disks, or optical media, such as digital versatile discs (DVDs), or semiconductor media, such as solid-state disks (SSDs).

[0107] This application also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. The processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the aforementioned global cooperative optimization control method.

[0108] The above-disclosed embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A global collaborative optimization control method for a central air conditioning cooling water system, characterized in that, include: Obtain the current operating conditions of the cooling water system, as well as the first energy consumption model, the second energy consumption model, and the third energy consumption model corresponding to the chiller, the cooling water pump group, and the cooling tower group, respectively; Generate multiple candidate parameter combinations with cooling loop temperature difference and cooling tower approximation degree as control variables; Under the condition that the current cooling load remains unchanged, based on the current operating conditions and any of the candidate parameter combinations, the power of the chiller is predicted using the first energy consumption model. The cooling side heat dissipation is determined based on the current cooling load and the predicted power of the chiller, and the cooling water flow rate is determined based on the cooling side heat dissipation and the corresponding cooling circuit temperature difference. Based on the cooling water flow rate, the power of the cooling water pump set is predicted using the second energy consumption model; Based on the heat dissipation on the cooling side, the current operating conditions, and the corresponding candidate parameter combinations, the power of the cooling tower group is predicted using the third energy consumption model. The sum of the power of the chiller unit, the power of the cooling water pump group, and the power of the cooling tower group is determined as the predicted total power of the system for the corresponding candidate parameter combination; The system selects the target parameter combination with the lowest predicted total power from the multiple candidate parameter combinations, and controls the cooling water pump group and cooling tower group according to the target parameter combination.

2. The global collaborative optimization control method according to claim 1, characterized in that, The first energy consumption model, the second energy consumption model, and the third energy consumption model are obtained in the following ways: Obtain the historical operating data of the chiller, cooling water pump set, and cooling tower set respectively; The historical operating data of the chiller, cooling water pump group and cooling tower group are cleaned and fuzzy sub-clustering are processed to obtain multiple typical operating condition cluster centers for each. Based on the cluster centers of the multiple typical operating conditions, fuzzy inference models corresponding to the chiller, cooling water pump group and cooling tower group are established respectively. The model parameters of each fuzzy inference model are determined by least squares parameter identification to obtain the first energy consumption model, the second energy consumption model and the third energy consumption model.

3. The global collaborative optimization control method according to claim 2, characterized in that, Also includes: New operating samples of the chiller, cooling water pump group and cooling tower group are obtained, and the sample density value of the new operating samples and the degree of correlation between the new operating samples and the corresponding existing typical operating condition cluster centers are calculated. Based on the sample density value and correlation degree, the newly added operating samples are merged into the corresponding existing typical operating condition cluster centers, or new typical operating condition cluster centers are generated based on the newly added operating samples, and existing typical operating condition cluster centers that do not meet the preset retention conditions are eliminated. The first, second, or third energy consumption model is updated based on the updated typical operating condition cluster centers.

4. The global collaborative optimization control method according to claim 2, characterized in that, The first energy consumption model takes the load rate of the chiller, the evaporator inlet water temperature, the evaporator outlet water temperature, the condenser inlet water temperature, and the condenser outlet water temperature as inputs, and the chiller power as output. The second energy consumption model takes cooling water flow rate as input and cooling water pump set speed ratio and cooling water pump set power as output; The third energy consumption model takes the heat dissipation on the cooling side, the temperature difference in the cooling circuit, the inlet water temperature of the cooling tower, the outdoor wet-bulb temperature, and the cooling tower approximation degree as inputs, and the speed ratio of the cooling tower group and the power of the cooling tower group as outputs.

5. The global collaborative optimization control method according to claim 1, characterized in that, A method for generating multiple candidate parameter combinations with cooling loop temperature difference and cooling tower approximation degree as control variables, and selecting the target parameter combination from the multiple candidate parameter combinations, includes: A particle population is generated using the temperature difference in the cooling circuit and the proximity of the cooling tower as the position parameters of the particles. The total predicted power of the system for the candidate parameter combination corresponding to the position parameters of each particle is determined as the fitness value of the particle. The individual optimal position of each particle and the global optimal position of the particle population are updated based on the fitness value of each particle, and the velocity and position of each particle are updated based on the individual optimal position and the global optimal position. When the preset number of iterations is reached or the preset convergence condition is met, the candidate parameter combination corresponding to the global optimal position is determined as the target parameter combination.

6. The global collaborative optimization control method according to claim 5, characterized in that, In the process of determining the target parameter combination using the particle population, each candidate parameter combination is verified by equipment operation constraints, which include at least one of the following: cooling water pump group operating frequency constraint, cooling tower group operating frequency constraint, cooling water flow rate constraint, and cooling tower inlet water temperature constraint; candidate parameter combinations that do not meet the equipment operation constraints are excluded, or a penalty is imposed on the fitness value of candidate parameter combinations that do not meet the equipment operation constraints.

7. The global collaborative optimization control method according to claim 1, characterized in that, Before generating the multiple candidate parameter combinations, the process also includes: Ensure that the time interval between the current moment and the moment when the control command was last successfully issued is not less than the preset minimum control interval; It is determined that at least one of the chiller units is in cooling operation, the cooling water pump group is in synchronous frequency conversion temperature difference control, and the cooling tower group is in synchronous frequency conversion approximation control. The chilled water outlet temperature, cooling circuit temperature difference, and cooling tower proximity of the chiller are obtained and calculated within the preset statistical period. When the range of each parameter is not greater than the corresponding preset fluctuation threshold, the cooling water system is determined to be in a steady-state operation. The current operating conditions are input into the first energy consumption model, the second energy consumption model, and the third energy consumption model to obtain the corresponding predicted power of the equipment. Based on the predicted power of each equipment and the actual power of the corresponding equipment, it is determined that the power prediction error rates of the first energy consumption model, the second energy consumption model, and the third energy consumption model are all within the corresponding preset error range. After selecting the target parameter combination, the predicted energy saving rate is determined based on the difference between the current actual total power and the system predicted total power corresponding to the target parameter combination, as well as the current actual total power. When the system predicted total power is lower than the current actual total power, the predicted energy saving rate is not less than the preset lower limit of energy saving rate and not greater than the preset upper limit of energy saving rate, and at least one of the absolute values ​​of the difference between the target cooling circuit temperature difference and the current cooling circuit temperature difference and the absolute values ​​of the difference between the target cooling tower approximation degree and the current cooling tower approximation degree is not less than the corresponding preset adjustment threshold, the cooling water pump group and the cooling tower group are controlled according to the target parameter combination.

8. A global collaborative optimization control system for a central air conditioning cooling water system, characterized in that, The system includes a controller that performs control of the central air conditioning cooling water system based on the global collaborative optimization control method according to any one of claims 1 to 7.

9. A global collaborative optimization control system for a central air conditioning cooling water system, characterized in that, include: One or more processors; Memory; And one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the programs including instructions for performing the global cooperative optimization control method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It includes a computer program that can be executed by a processor to perform the global collaborative optimization control method as described in any one of claims 1 to 7.

Images