An HVAC multi-unit air conditioning system energy consumption optimization matching system and method

Abstract

The present application relates to air conditioning energy consumption optimization technical field, disclose a kind of HVAC multi-unit air conditioning system energy consumption optimization matching system and method, based on historical data and real-time operating parameter constructs polynomial function model, accurately describes the nonlinear relationship between each unit energy consumption and power distribution;Second, by principal component analysis and segmented modeling means such as enhancing model adaptability, and combining genetic algorithm realizes global optimization of model parameters, improves model prediction accuracy and robustness;In optimization control level, introduce multi-unit collaborative weight factor and adopt weighted particle swarm optimization algorithm, dynamically adjust power distribution strategy, realize the collaborative optimization control between multi-unit. Through complete acquisition-modeling-optimization-execution closed-loop control structure, realize the intelligent perception of the running state of multi-unit air conditioning system and energy consumption optimal scheduling.

Description

Technical Field

[0001] This invention relates to the field of air conditioning energy consumption optimization technology, and in particular to an energy consumption optimization and matching system and method for HVAC multi-unit air conditioning systems. Background Technology

[0002] In large buildings, industrial plants, and commercial complexes, HVAC (Heating, Ventilation, and Air Conditioning) systems are crucial for ensuring indoor comfort and air quality. To meet the diverse needs of large spaces and multiple areas, HVAC systems typically employ multiple air conditioning units operating in tandem. In such systems, different units handle temperature control for different areas, and their operating efficiency and energy consumption directly impact the overall building's energy consumption. However, current technologies generally employ static or empirical control methods to allocate unit power in multi-unit HVAC systems, leading to issues such as high overall energy consumption and uneven resource utilization.

[0003] Therefore, there is an urgent need for a method that combines operating parameter acquisition, dynamic modeling, and intelligent optimization control to construct an energy consumption optimization and matching system suitable for HVAC multi-unit systems, thereby improving system energy efficiency and operational economy. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide an energy consumption optimization and matching system and method for HVAC multi-unit air conditioning systems, so as to solve the technical problems of high overall energy consumption and uneven resource utilization in existing HVAC multi-unit air conditioning systems.

[0005] The first aspect of this invention discloses an energy consumption optimization and matching system for HVAC multi-unit air conditioning systems, which includes a data acquisition module, an energy consumption modeling module, an optimization control module, and an execution control module;

[0006] The data acquisition module is used to collect operating parameter data of each air conditioning unit;

[0007] The energy consumption modeling module is used to construct a polynomial function model of the relationship between energy consumption and power distribution of each air conditioning unit based on the collected operating parameter data; the polynomial function model is used to present the energy consumption variation characteristics of each unit under different power distribution conditions;

[0008] The optimization control module is used to execute a global power optimization algorithm based on a polynomial function model to determine the optimal power allocation scheme; the optimal power allocation scheme is the power allocation scheme with the lowest overall energy consumption of the multi-unit air conditioning system;

[0009] The execution control module is used to adjust the operating parameters of each air conditioning unit according to the optimal power allocation scheme.

[0010] Furthermore, the polynomial function model is constructed based on the historical operating parameters and real-time collected operating parameters of each air conditioning unit. The construction process includes the following steps:

[0011] S1. Normalize the historical and real-time operating parameters to obtain the first operating parameter;

[0012] S2. The first operating parameters are dimensionality reduced using the principal component analysis algorithm to obtain the second operating parameters;

[0013] S3. The least squares method is used to fit the polynomial function model based on the second running parameter, and the parameters of the polynomial function model are optimized by combining the fitting error evaluation.

[0014] Furthermore, the execution process of step S2 specifically includes:

[0015] S21. Construct a covariance matrix for the first operating parameters and extract the main component directions by eigenvalue decomposition;

[0016] S22. Introduce the COP (Coefficient of Performance) of the air conditioning system as a principal component screening constraint factor;

[0017] S23. Output the top K principal components whose cumulative contribution rate exceeds the contribution rate threshold, as the second running parameter.

[0018] Furthermore, the execution process of fitting the polynomial function model based on the second running parameters using the least squares method in step S3 specifically includes:

[0019] S31. For the second operating parameter under different working conditions, the least squares method is used to fit the polynomial function through piecewise modeling to construct a working condition model set containing multiple sub-models;

[0020] S32. Introduce weighting factors for the corresponding working conditions of each sub-model to form a working condition adaptive weighted polynomial function model;

[0021] S33. After the initial fitting of the polynomial function model is completed, a global search optimization of the model's parameter space is performed using a genetic algorithm.

[0022] Furthermore, step S3, which combines fitting error evaluation with parameter optimization of the polynomial function model, specifically includes:

[0023] During the deployment and operation phase of the polynomial function model, the corresponding predicted energy consumption is calculated using the constructed polynomial function model based on the real-time operating parameters of each air conditioning unit.

[0024] Obtain actual energy consumption data of multi-unit air conditioning systems and calculate the short-term residual between predicted energy consumption and actual energy consumption;

[0025] When the short-term residuals of multiple consecutive time steps exceed the set dynamic threshold, and the current polynomial function model is judged to have a deviation based on the changing trend of the operating environment parameters, the adaptive correction operation of the polynomial function model is triggered.

[0026] The adaptive correction operation includes fine-tuning the operating condition weight factor and performing local weight optimization.

[0027] Furthermore, during the execution of the global power optimization algorithm by the optimization control module, a multi-unit collaborative weighting coefficient is introduced; the multi-unit collaborative weighting coefficient is used to characterize the response capability and collaborative effect among different air conditioning units in the energy consumption optimization process.

[0028] The collaborative weighting coefficient is dynamically updated based on the historical energy efficiency, real-time response speed, and load capacity of each air conditioning unit.

[0029] Furthermore, the global power optimization algorithm is a weighted particle swarm optimization algorithm (WPSO). The WPSO algorithm dynamically adjusts the search strategy based on cooperative weights during the power allocation process, specifically including:

[0030] During the initialization phase, the initial position and velocity of individual particles are adjusted according to the cooperative weight coefficients of each unit.

[0031] During the fitness assessment process, a dual-objective fusion assessment is performed based on the total energy consumption and response robustness of the multi-unit air conditioning system using a fitness function.

[0032] During the velocity and position update process, a cooperative guidance factor is introduced, and the particle search weight direction is dynamically adjusted based on the cooperative guidance factor; the cooperative guidance factor is used to guide highly cooperative units to preferentially obtain more power allocation space.

[0033] Furthermore, the collaborative guiding factor is dynamically updated by feeding back energy efficiency trends from the multi-unit air conditioning system within a sliding window.

[0034] The second aspect of this invention discloses an energy consumption optimization and matching method for a multi-unit HVAC air conditioning system, which is applied to the system disclosed in the first aspect. The method includes:

[0035] Collect operating parameter data for each air conditioning unit;

[0036] Based on the collected operating parameter data, a polynomial function model of the relationship between energy consumption and power distribution of each air conditioning unit is constructed; the polynomial function model is used to present the energy consumption variation characteristics of each unit under different power distribution conditions.

[0037] A global power optimization algorithm is executed based on a polynomial function model to determine the optimal power allocation scheme; the optimal power allocation scheme is the power allocation scheme with the lowest overall energy consumption of the multi-unit air conditioning system.

[0038] Adjust the operating parameters of each air conditioning unit according to the optimal power allocation scheme.

[0039] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0040] This invention achieves intelligent sensing and optimal energy consumption scheduling of multi-unit air conditioning systems by synergistically integrating four parts: data acquisition, energy consumption modeling, optimization control, and execution control. It establishes a complete closed-loop control structure encompassing data acquisition, modeling, optimization, and execution, enabling intelligent perception of the operating status of multi-unit air conditioning systems. First, a polynomial function model is constructed based on historical data and real-time operating parameters to accurately describe the nonlinear relationship between energy consumption and power distribution of each unit. Second, principal component analysis and piecewise modeling are used to enhance model adaptability, and a genetic algorithm is combined to achieve global optimization of model parameters, improving prediction accuracy and robustness. At the optimization control level, a multi-unit collaborative weighting factor is introduced, and a weighted particle swarm optimization algorithm is employed to dynamically adjust the power distribution strategy, achieving collaborative optimization control among multiple units. This improves overall system energy efficiency, significantly reducing the overall energy consumption of multi-unit HVAC systems under dynamic operating conditions, increasing energy utilization efficiency, and demonstrating a high degree of intelligence and engineering adaptability. Attached Figure Description

[0041] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0042] Figure 1 This is a schematic diagram of the structure of an energy consumption optimization and matching system for a multi-unit HVAC air conditioning system disclosed in Embodiment 1 of the present invention;

[0043] Figure 2 This is a flowchart illustrating an energy consumption optimization and matching method for a multi-unit HVAC air conditioning system, as disclosed in another embodiment of the present invention. Detailed Implementation

[0044] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0045] Example 1

[0046] The first aspect of this invention discloses an energy consumption optimization and matching system for HVAC multi-unit air conditioning systems. Please refer to [link / reference]. Figure 1 , Figure 1 This is a schematic diagram of the structure of an HVAC multi-unit air conditioning system energy consumption optimization and matching system disclosed in an embodiment of the present invention. The system includes a data acquisition module, an energy consumption modeling module, an optimization control module, and an execution control module.

[0047] The data acquisition module is used to collect operating parameter data of each air conditioning unit;

[0048] The energy consumption modeling module is used to construct a polynomial function model of the relationship between energy consumption and power distribution of each air conditioning unit based on the collected operating parameter data; the polynomial function model is used to present the energy consumption variation characteristics of each unit under different power distribution conditions;

[0049] The optimization control module is used to execute a global power optimization algorithm based on a polynomial function model to determine the optimal power allocation scheme; the optimal power allocation scheme is the power allocation scheme with the lowest overall energy consumption of the multi-unit air conditioning system;

[0050] The execution control module is used to adjust the operating parameters of each air conditioning unit according to the optimal power allocation scheme.

[0051] Specifically, in this embodiment of the invention, operating parameters refer to key operating condition data reflecting the operating status and energy consumption characteristics of each HVAC air conditioning unit, including but not limited to: ambient temperature, ambient humidity, set temperature, cooling / heating load, compressor operating frequency, refrigerant pressure, fan speed, current, voltage, operating time, inlet and outlet air temperature difference, chilled water temperature difference, and Coefficient of Performance (COP). These parameters comprehensively cover the unit's input conditions, internal operating status, and output performance, accurately describing the nonlinear impact of power distribution changes on energy consumption, thereby improving the fitting accuracy and generalization ability of the subsequent polynomial function model.

[0052] The aforementioned operating parameters are primarily acquired through real-time readings from the existing embedded sensors and controller interfaces in the air conditioning system (such as temperature and humidity sensors, energy meters, differential pressure transmitters, PLCs, or BAS system gateways). Data reception, calibration, and storage are performed via a unified data acquisition module. Additionally, some parameters, such as the Coefficient of Performance (COP), are calculated from the collected energy consumption and cooling / heating output.

[0053] Furthermore, the polynomial function model is constructed based on the historical operating parameters and real-time collected operating parameters of each air conditioning unit. The construction process includes the following steps:

[0054] S1. Normalize the historical and real-time operating parameters to obtain the first operating parameter;

[0055] S2. The first operating parameters are dimensionality reduced using the principal component analysis algorithm to obtain the second operating parameters;

[0056] S3. The least squares method is used to fit the polynomial function model based on the second running parameter, and the parameters of the polynomial function model are optimized by combining the fitting error evaluation.

[0057] Furthermore, the execution process of step S2 specifically includes:

[0058] S21. Construct a covariance matrix for the first operating parameters and extract the main component directions by eigenvalue decomposition;

[0059] S22. Introduce the COP (Coefficient of Performance) of the air conditioning system as a principal component screening constraint factor;

[0060] S23. Output the top K principal components whose cumulative contribution rate exceeds the contribution rate threshold, as the second running parameter.

[0061] Specifically, this embodiment of the invention constructs a covariance matrix for the first operating parameter to characterize the correlation between the parameters, reflecting the degree of linear correlation between multidimensional operating parameters, which helps to identify parameter dimensions with high redundancy or low contribution. Furthermore, the covariance matrix is ​​decomposed based on the eigenvalue decomposition method to extract its corresponding eigenvectors and eigenvalues. The eigenvectors represent the directions of the principal components, and the eigenvalues ​​represent the weight of each principal component in the total variance, thus obtaining the basis for dimensionality reduction.

[0062] Building upon this foundation, this invention introduces the air conditioning system energy efficiency ratio (EER) as a constraint factor for principal component selection to enhance the energy efficiency explanatory power of the principal components. Specifically, when selecting principal components after eigenvalue decomposition, in addition to considering the traditional cumulative contribution rate threshold, the correlation between the principal component and the energy efficiency index COP is further considered. Only those principal components that have a strong correlation with the energy efficiency performance of multi-unit air conditioning systems are retained. This ensures that the dimensionality-reduced operating parameters not only retain the amount of data information but also enhance the correlation with energy consumption optimization objectives.

[0063] Furthermore, in step S23, the first K principal components whose cumulative contribution rates exceed a preset threshold and meet the COP constraint are used as the second operating parameter for the subsequent fitting process of the polynomial function model. This processing not only effectively reduces the model input dimension and computational complexity, but also significantly improves the model's ability to express energy consumption characteristics under multiple operating conditions, providing a more representative and energy-efficiency-oriented data foundation for subsequent global power optimization. In this embodiment, K is the number of principal components whose cumulative contribution rates exceed the set threshold and meet the COP constraint of the air conditioning system. Specifically, the value of K is dynamically determined based on the eigenvalue distribution and the degree of COP matching.

[0064] Furthermore, the execution process of fitting the polynomial function model based on the second running parameters using the least squares method in step S3 specifically includes:

[0065] S31. For the second operating parameter under different working conditions, the least squares method is used to fit the polynomial function through piecewise modeling to construct a working condition model set containing multiple sub-models;

[0066] S32. Introduce weighting factors for the corresponding working conditions of each sub-model to form a working condition adaptive weighted polynomial function model;

[0067] S33. After the initial fitting of the polynomial function model is completed, a global search optimization of the model's parameter space is performed using a genetic algorithm.

[0068] Specifically, for each air conditioning unit, it usually experiences a variety of typical operating conditions during its operating life cycle, such as high temperature and high humidity conditions, medium load conditions, and partial load rapid response conditions. In order to more realistically reflect the response characteristics of energy consumption with power change under different operating conditions, this invention adopts a piecewise modeling strategy to fit the polynomial function.

[0069] Specifically, the collected operating parameters, i.e., the second operating parameters, are classified according to predefined operating condition categories. The dataset under each operating condition category is used to fit a separate multinomial sub-model. Each air conditioning unit can correspond to multiple operating condition sub-models, forming the unit's "operating condition model set." Each sub-model expresses the functional relationship between energy consumption and power based on the operating characteristics under that operating condition, including:

[0070]

[0071] Among them, E (k) (P) represents the predicted energy consumption of the kth operating condition sub-model when the power input is P; k is the operating condition index, corresponding to different operating conditions; For the power term P in the k-th operating condition sub-model j The corresponding polynomial coefficients; n is the order of the polynomial function, and j is the index of the order.

[0072] Furthermore, to comprehensively assess the applicability of each sub-model in actual operation, this invention further introduces a weighting factor based on the characteristics of each operating condition and the current operating status of the unit. This factor is then used to weight and fuse all sub-models, constructing an adaptive weighted multinomial function model. This weighting factor is determined based on multiple criteria, including but not limited to the similarity between current operating environment parameters and historical characteristics of a certain operating condition, historical energy efficiency performance of the current operating condition, and unit response stability. Through weighted fusion, each air conditioning unit ultimately forms an overall energy consumption function model capable of adapting to changes in multiple operating conditions, including:

[0073]

[0074] Among them, E flused (P) represents the fused predicted energy consumption value, i.e., the output of the load-weighted polynomial function model; w k Let w be the weight factor of the k-th working condition sub-model, representing its importance in the current prediction, satisfying ∑w k =1; K is the total number of sub-models, corresponding to the number of identifiable operating conditions.

[0075] After initial fitting, to further improve the model's fitting accuracy for energy consumption trends under different inputs, this invention introduces a Genetic Algorithm (GA) to perform a global optimization search on the parameter space of the polynomial function model, thereby escaping local optima and optimizing the model's fitting error distribution. Preferably, the optimization objective function is achieved by combining multiple performance indicators such as minimum fitting residuals, weighted sum of squared errors, and multi-condition coverage. Specifically, firstly, high-precision modeling of different condition data is achieved by minimizing the weighted sum of squared residuals under multiple conditions. Then, the condition coverage indicator is introduced to measure the activation degree of the model under multiple conditions, avoiding excessive reliance on individual condition sub-models. Furthermore, a regularization term is added to the model parameters to constrain the size of the fitting coefficients and prevent overfitting. The constructed objective function is used for the design of the fitness function of the genetic algorithm; the smaller the objective function value, the higher the fitness value of the corresponding parameter combination. The GA algorithm evaluates the merits of different parameter combinations through the fitness function and finally outputs the globally optimal parameter set, thus completing the final construction of the unit's polynomial function model.

[0076] Preferably, let the parameters of the polynomial function model be θ, then the corresponding optimization objective function is:

[0077]

[0078] Where θ is the complete set of fitting parameters for the polynomial function model, i.e., under each working condition. Coefficient; l is the sample number under the k-th working condition, and there are a total of T k There are 1 data samples; represents the predicted energy consumption value of the k-th operating condition sub-model at the l-th sample point. λ1 is the measured actual energy consumption value, used as the target data for model fitting; λ1 is the weighting coefficient of the influence of the control condition coverage penalty term, with a value between 0 and 1; coverage is the proportion of the currently active condition; λ2 is the penalty coefficient of the regularization term, used to prevent overfitting, with a value between 0 and 1; ||θ|| 2 This is the sum of squares of all model parameters, used to avoid overfitting.

[0079] In the process of constructing the polynomial function model, this invention significantly improves the prediction accuracy and adaptability of the model under multiple operating conditions by introducing piecewise modeling, weighted fusion of operating conditions, and a global optimization mechanism. By employing the least squares method to piecewise fit the second operating parameter under different operating conditions, multiple targeted sub-models for each operating condition are constructed, which can fully reflect the energy consumption variation characteristics of the air conditioning unit under various loads and environmental conditions. Based on this, by introducing operating condition weight factors, the sub-models are weighted and fused to form an overall energy consumption prediction model with operating condition adaptability, enhancing the model's generalization ability and flexibility in complex operating scenarios. Furthermore, by combining a genetic algorithm to perform global search optimization on the model parameter space, the overall fitting effect and the ability to obtain the global optimal solution are further improved, effectively balancing model accuracy, adaptability, and stability, providing highly reliable predictive support for subsequent energy consumption optimization and control.

[0080] Furthermore, step S3, which combines fitting error evaluation with parameter optimization of the polynomial function model, specifically includes:

[0081] During the deployment and operation phase of the polynomial function model, the corresponding predicted energy consumption is calculated using the constructed polynomial function model based on the real-time operating parameters of each air conditioning unit.

[0082] Obtain actual energy consumption data of multi-unit air conditioning systems and calculate the short-term residual between predicted energy consumption and actual energy consumption;

[0083] When the short-term residuals of multiple consecutive time steps exceed the set dynamic threshold, and the current polynomial function model is judged to have a deviation based on the changing trend of the operating environment parameters, the adaptive correction operation of the polynomial function model is triggered.

[0084] The adaptive correction operation includes fine-tuning the operating condition weight factors and performing local weight optimization.

[0085] Considering the dynamic nature of the operating environment of multi-unit air conditioning systems, traditional static fitting methods can lead to a decrease in prediction accuracy after the air conditioning system has been running for a period of time due to factors such as environmental changes, load fluctuations, or equipment aging. Therefore, in this embodiment of the invention, an adaptive correction mechanism based on real-time fitting error evaluation dynamically optimizes the parameters during the model operation phase, enhancing the long-term stability and adaptability of the prediction model.

[0086] Specifically, the dynamic threshold is automatically adjusted by combining the current environmental disturbance magnitude, the custom tolerance range, and the historical error fluctuation range. When the residual exceeds the threshold for multiple consecutive time steps (e.g., the past N times), and the trend analysis of the operating environment parameters (such as outdoor temperature and humidity, indoor load changes, ventilation frequency, etc.) indicates that the current model has generated a systematic deviation, the "adaptive correction operation" is automatically triggered.

[0087] Specifically, the trend of operating environment data refers to the continuous change trend of key environmental factors over a period of time, such as continuous increase in external temperature and abnormal increase in cooling water temperature. When it is determined that the operating environment parameters have "distribution drift" (such as the difference from the mean of the distribution during the training period exceeds the threshold), combined with the phenomenon of continuously high residuals, it is determined that the model currently has structural bias under external driving, and at this time the model parameter correction operation is triggered.

[0088] The adaptive correction operation preferably includes re-estimating the relative applicability of the sub-model based on the residual distribution, and applying sliding window statistical data to perform gradient fine-tuning or simple regression adjustment of the weights of each sub-model.

[0089] By introducing an adaptive parameter optimization mechanism based on short-term fitting residual evaluation during the deployment and operation phase of the polynomial function model, dynamic adjustment and continuous optimization of the model during actual operation are achieved. This mechanism can automatically trigger model correction operations when the system detects that the prediction error continuously exceeds a set threshold and judges that there is model bias based on the changing trend of operating environment parameters. By fine-tuning the operating condition weight factors and performing local weight optimization, not only is the model's adaptability to factors such as changes in operating conditions, environmental disturbances, and equipment aging effectively improved, but the computational burden of retraining the model is also avoided. This optimization process has online execution capability, which can significantly improve the energy consumption prediction accuracy and overall energy efficiency control level of multi-unit systems during long-term operation, thereby enhancing the stability and intelligence level of the entire energy consumption optimization and matching system.

[0090] Furthermore, during the execution of the global power optimization algorithm by the optimization control module, a multi-unit collaborative weighting coefficient is introduced; the multi-unit collaborative weighting coefficient is used to characterize the response capability and collaborative effect among different air conditioning units in the energy consumption optimization process.

[0091] The collaborative weighting coefficient is dynamically updated based on the historical energy efficiency, real-time response speed, and load capacity of each air conditioning unit.

[0092] In the optimization control module of this invention, in order to optimize the global power allocation strategy of HVAC multi-unit air conditioning system, a multi-unit collaborative weight coefficient mechanism is introduced to present the response capability and collaborative effect of different air conditioning units in the process of participating in energy consumption optimization scheduling. This enables the optimization algorithm to favor units with high responsiveness and high collaboration, thereby improving the overall scheduling efficiency and system stability.

[0093] Furthermore, the calculation method for the multi-unit collaborative weighting coefficient is as follows:

[0094] w i =α·η i +β·(1 / τi )+γ·λ i

[0095] Among them, w i η is the collaborative weighting coefficient for the i-th air conditioning unit; i Let τ be the unit energy efficiency of the i-th air conditioning unit; i λ is the average response time of the i-th air conditioning unit; i Let be the load adjustment flexibility index of the i-th air conditioning unit; α, β, and γ are empirical weighting coefficients, and their sum is 1.

[0096] By integrating multiple dimensions of indicators such as historical energy efficiency performance, real-time response speed, and load adjustment capability of the units, a comprehensive weight parameter is dynamically constructed to guide power optimization allocation. This effectively enhances the energy consumption optimization matching system's ability to perceive the differences in the operating capabilities of various air conditioning units. This allows the optimization algorithm to prioritize units with fast response, high energy efficiency, and strong adjustment capabilities when allocating power resources, thereby enhancing the accuracy and real-time performance of the overall energy efficiency scheduling of the energy consumption optimization matching system. The collaborative weight coefficient is automatically updated in each round of optimization, further improving the control strategy's adaptability to dynamic changes in the operating environment and providing an intelligent, flexible, and efficient optimization scheduling basis for energy-saving control of HVAC multi-unit systems.

[0097] Furthermore, the global power optimization algorithm is a weighted particle swarm optimization algorithm (WPSO). The WPSO algorithm dynamically adjusts the search strategy based on cooperative weights during the power allocation process, specifically including:

[0098] During the initialization phase, the initial position and velocity of individual particles are adjusted according to the cooperative weight coefficients of each unit.

[0099] During the fitness assessment process, a dual-objective fusion assessment is performed based on the total energy consumption and response robustness of the multi-unit air conditioning system using a fitness function.

[0100] During the velocity and position update process, a cooperative guidance factor is introduced, and the particle search weight direction is dynamically adjusted based on the cooperative guidance factor; the cooperative guidance factor is used to guide highly cooperative units to preferentially obtain more power allocation space.

[0101] Furthermore, the collaborative guiding factor is dynamically updated by feeding back energy efficiency trends from multi-unit air conditioning systems within a sliding window.

[0102] Specifically, in this invention, in order to achieve efficient and intelligent optimization scheduling of power allocation for multi-unit air conditioning systems, a weighted particle swarm optimization algorithm is introduced, and a particle search and optimization process with collaborative driving capability is constructed by combining a dynamic adjustment mechanism of collaborative weight coefficients and collaborative guiding factors.

[0103] In the initialization phase of WPSO, instead of randomly distributing particle positions and velocities, the initial positions and search velocities of the particles are weighted and adjusted based on the cooperative weight coefficients of each air conditioning unit. For units with higher cooperative weights, their particles are given initial positions closer to the target power allocation region, and a higher search velocity is assigned, enhancing their dominance in the early search and thus improving the quality of the overall solution in the initial search phase, providing a good convergence foundation for subsequent iterations.

[0104] In the fitness calculation process of each iteration of the particle, a dual-objective fusion fitness function is constructed, which comprehensively considers the following two core evaluation dimensions:

[0105] Total system energy consumption target: Evaluate the total energy consumption generated by the power allocation scheme represented by the current particle in a multi-unit system, and the goal is to minimize this value;

[0106] Response robustness objective: Based on the collaborative weights and unit status (response speed, stability), assess the actual feasibility and robustness of the current power allocation scheme to avoid allocating excessive load to slow-responding or unadjustable units.

[0107] During the particle velocity and position update phase, a "cooperative guidance factor" is introduced to dynamically adjust the search weight direction of each particle during the update process. This factor guides the optimization algorithm to allocate more power space to highly cooperative units, thereby accelerating convergence and enhancing the feasibility of the scheme. The cooperative guidance factor is related to the cooperative weight coefficient but has real-time dynamics. During the operation phase, energy efficiency feedback data (such as the trend of unit power consumption change) of multiple units is recorded through a sliding window. Based on the trend of recent operating cycles, it is determined whether the cooperativeness of certain units has increased or decreased, and the guidance factor is adjusted in real time to enhance the influence of high-efficiency and fast-response units in the search process. Through the dynamic adjustment of the cooperative guidance factor, the particle swarm can be more inclined to explore more reasonable and feasible energy consumption allocation regions in the search space.

[0108] This invention introduces a weighted particle swarm optimization algorithm based on a collaborative weighting mechanism into the optimization control module, thereby achieving efficient search and intelligent scheduling of power allocation strategies for multi-unit air conditioning systems. This effectively strengthens the guiding role of highly collaborative units, improves search convergence speed and scheduling accuracy, and significantly enhances the energy efficiency matching capability and intelligent operation level of multi-unit systems under complex operating conditions without increasing the system's computational burden.

[0109] Example 2

[0110] The second aspect of this invention discloses an energy consumption optimization and matching method for HVAC multi-unit air conditioning systems. Please refer to [link / reference]. Figure 2 , Figure 2This is a flowchart illustrating an energy consumption optimization and matching method for a multi-unit HVAC air conditioning system, as disclosed in another embodiment of the present invention. The method includes:

[0111] Collect operating parameter data for each air conditioning unit;

[0112] Based on the collected operating parameter data, a polynomial function model of the relationship between energy consumption and power distribution of each air conditioning unit is constructed; the polynomial function model is used to present the energy consumption variation characteristics of each unit under different power distribution conditions.

[0113] A global power optimization algorithm is executed based on a polynomial function model to determine the optimal power allocation scheme; the optimal power allocation scheme is the power allocation scheme with the lowest overall energy consumption of the multi-unit air conditioning system.

[0114] Adjust the operating parameters of each air conditioning unit according to the optimal power allocation scheme.

[0115] Furthermore, the polynomial function model is constructed based on the historical operating parameters and real-time collected operating parameters of each air conditioning unit. The construction process includes the following steps:

[0116] S1. Normalize the historical and real-time operating parameters to obtain the first operating parameter;

[0117] S2. The first operating parameters are dimensionality reduced using the principal component analysis algorithm to obtain the second operating parameters;

[0118] S3. The least squares method is used to fit the polynomial function model based on the second running parameter, and the parameters of the polynomial function model are optimized by combining the fitting error evaluation.

[0119] Furthermore, the execution process of step S2 specifically includes:

[0120] S21. Construct a covariance matrix for the first operating parameters and extract the main component directions by eigenvalue decomposition;

[0121] S22. Introduce the COP (Coefficient of Performance) of the air conditioning system as a principal component screening constraint factor;

[0122] S23. Output the top K principal components whose cumulative contribution rate exceeds the contribution rate threshold, as the second running parameter.

[0123] Furthermore, the execution process of fitting the polynomial function model based on the second running parameters using the least squares method in step S3 specifically includes:

[0124] S31. For the second operating parameter under different working conditions, the least squares method is used to fit the polynomial function through piecewise modeling to construct a working condition model set containing multiple sub-models;

[0125] S32. Introduce weighting factors for the corresponding working conditions of each sub-model to form a working condition adaptive weighted polynomial function model;

[0126] S33. After the initial fitting of the polynomial function model is completed, a global search optimization of the model's parameter space is performed using a genetic algorithm.

[0127] Furthermore, step S3, which combines fitting error evaluation with parameter optimization of the polynomial function model, specifically includes:

[0128] During the deployment and operation phase of the polynomial function model, the corresponding predicted energy consumption is calculated using the constructed polynomial function model based on the real-time operating parameters of each air conditioning unit.

[0129] Obtain actual energy consumption data of multi-unit air conditioning systems and calculate the short-term residual between predicted energy consumption and actual energy consumption;

[0130] When the short-term residuals of multiple consecutive time steps exceed the set dynamic threshold, and the current polynomial function model is judged to have a deviation based on the changing trend of the operating environment parameters, the adaptive correction operation of the polynomial function model is triggered.

[0131] The adaptive correction operation includes fine-tuning the operating condition weight factors and performing local weight optimization.

[0132] Furthermore, during the execution of the global power optimization algorithm by the optimization control module, a multi-unit collaborative weighting coefficient is introduced; the multi-unit collaborative weighting coefficient is used to characterize the response capability and collaborative effect among different air conditioning units in the energy consumption optimization process.

[0133] The collaborative weighting coefficient is dynamically updated based on the historical energy efficiency, real-time response speed, and load capacity of each air conditioning unit.

[0134] Furthermore, the calculation method for the multi-unit coordination weighting coefficient is as follows:

[0135] w i =α·η i +β·(1 / τ i )+γ·λ i

[0136] Among them, w i η is the collaborative weighting coefficient for the i-th air conditioning unit; i Let τ be the unit energy efficiency of the i-th air conditioning unit; i λ is the average response time of the i-th air conditioning unit; i Let be the load adjustment flexibility index of the i-th air conditioning unit; α, β, and γ are empirical weighting coefficients, and their sum is 1.

[0137] Furthermore, the global power optimization algorithm is the Weighted Particle Swarm Optimization (WPSO) algorithm. The WPSO algorithm dynamically adjusts the search strategy based on cooperative weights during the power allocation process, specifically including:

[0138] During the initialization phase, the initial position and velocity of individual particles are adjusted according to the cooperative weight coefficients of each unit.

[0139] During the fitness assessment process, a dual-objective fusion assessment is performed based on the total energy consumption and response robustness of the multi-unit air conditioning system using a fitness function.

[0140] During the velocity and position update process, a cooperative guidance factor is introduced, and the particle search weight direction is dynamically adjusted based on the cooperative guidance factor; the cooperative guidance factor is used to guide highly cooperative units to preferentially obtain more power allocation space.

[0141] Furthermore, the collaborative guiding factor is dynamically updated by feeding back energy efficiency trends from multi-unit air conditioning systems within a sliding window.

[0142] It should be noted that the specific implementation process of Embodiment 2 is similar to that of Embodiment 1, and will not be repeated in this embodiment.

[0143] Finally, it should be noted that the energy consumption optimization and matching system and method for HVAC multi-unit air conditioning systems disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An energy consumption optimization and matching system for a multi-unit HVAC air conditioning system, characterized in that, The energy consumption optimization and matching system includes a data acquisition module, an energy consumption modeling module, an optimization control module, and an execution control module; The data acquisition module is used to collect operating parameter data of each air conditioning unit; The energy consumption modeling module is used to construct a polynomial function model of the relationship between energy consumption and power distribution of each air conditioning unit based on the collected operating parameter data; the polynomial function model is used to present the energy consumption variation characteristics of each unit under different power distribution conditions; The optimization control module is used to execute a global power optimization algorithm based on a polynomial function model to determine the optimal power allocation scheme; the optimal power allocation scheme is the power allocation scheme with the lowest overall energy consumption of the multi-unit air conditioning system; The execution control module is used to adjust the operating parameters of each air conditioning unit according to the optimal power allocation scheme; The polynomial function model is constructed based on the historical operating parameters and real-time collected operating parameters of each air conditioning unit. The construction process includes the following steps: S1. Normalize the historical and real-time operating parameters to obtain the first operating parameter; S2. The first operating parameter is reduced in dimensionality using principal component analysis to obtain the second operating parameter; the principal component selection constraint factor in the principal component analysis algorithm is the COP of the air conditioning system. S3. The least squares method is used to fit a polynomial function model based on the second running parameters, and the parameters of the polynomial function model are optimized by combining the fitting error evaluation; During the execution of the global power optimization algorithm by the optimization control module, a multi-unit collaborative weighting coefficient is introduced; the multi-unit collaborative weighting coefficient is used to characterize the response capability and synergistic effect among different air conditioning units in the energy consumption optimization process; The collaborative weighting coefficient is dynamically updated based on the historical energy efficiency, real-time response speed, and load capacity of each air conditioning unit. The global power optimization algorithm is a weighted particle swarm optimization algorithm (WPSO). The WPSO algorithm dynamically adjusts the search strategy based on cooperative weights during the power allocation process, specifically including: During the initialization phase, the initial position and velocity of individual particles are adjusted according to the cooperative weight coefficients of each unit. During the fitness assessment process, a dual-objective fusion assessment is performed based on the total energy consumption and response robustness of the multi-unit air conditioning system using a fitness function. During the velocity and position update process, a cooperative guidance factor is introduced, and the particle search weight direction is dynamically adjusted based on the cooperative guidance factor; the cooperative guidance factor is used to guide highly cooperative units to preferentially obtain more power allocation space.

2. The HVAC multi-unit air conditioning system energy consumption optimization and matching system according to claim 1, characterized in that, The execution process of step S2 specifically includes: S21. Construct a covariance matrix for the first operating parameters and extract the main component directions by eigenvalue decomposition; S22. Introduce the COP (Coefficient of Performance) of the air conditioning system as a principal component screening constraint factor; S23. Output the top K principal components whose cumulative contribution rate exceeds the contribution rate threshold, as the second running parameter.

3. The HVAC multi-unit air conditioning system energy consumption optimization and matching system according to claim 1, characterized in that, The execution process of fitting a polynomial function model based on the second running parameter using the least squares method in step S3 specifically includes: S31. For the second operating parameter under different working conditions, the least squares method is used to fit the polynomial function through piecewise modeling to construct a working condition model set containing multiple sub-models; S32. Introduce weighting factors for the corresponding working conditions of each sub-model to form a working condition adaptive weighted polynomial function model; S33. After the initial fitting of the polynomial function model is completed, a global search optimization of the model's parameter space is performed using a genetic algorithm.

4. The HVAC multi-unit air conditioning system energy consumption optimization and matching system according to claim 1 or 3, characterized in that, Step S3, which combines fitting error evaluation to optimize the parameters of the multinomial function model, specifically includes: During the deployment and operation phase of the polynomial function model, the corresponding predicted energy consumption is calculated using the constructed polynomial function model based on the real-time operating parameters of each air conditioning unit. Obtain actual energy consumption data of multi-unit air conditioning systems and calculate the short-term residual between predicted energy consumption and actual energy consumption; When the short-term residuals of multiple consecutive time steps exceed the set dynamic threshold, and the current polynomial function model is judged to have a deviation based on the changing trend of the operating environment parameters, the adaptive correction operation of the polynomial function model is triggered. The adaptive correction operation includes fine-tuning the operating condition weight factor and performing local weight optimization.

5. The HVAC multi-unit air conditioning system energy consumption optimization and matching system according to claim 1, characterized in that, The collaborative guiding factor is dynamically updated based on the energy efficiency trend feedback from the multi-unit air conditioning system within the sliding window.

6. A method for optimizing and matching energy consumption in a multi-unit HVAC air conditioning system, wherein the method is applied to the energy consumption optimization and matching system according to any one of claims 1-5, characterized in that, The method includes: Collect operating parameter data for each air conditioning unit; Based on the collected operating parameter data, a polynomial function model of the relationship between energy consumption and power distribution of each air conditioning unit is constructed; the polynomial function model is used to present the energy consumption variation characteristics of each unit under different power distribution conditions. A global power optimization algorithm is executed based on a polynomial function model to determine the optimal power allocation scheme; the optimal power allocation scheme is the power allocation scheme with the lowest overall energy consumption of the multi-unit air conditioning system. Adjust the operating parameters of each air conditioning unit according to the optimal power allocation scheme; The polynomial function model is constructed based on the historical operating parameters and real-time collected operating parameters of each air conditioning unit. The construction process includes the following steps: S1. Normalize the historical and real-time operating parameters to obtain the first operating parameter; S2. The first operating parameter is reduced in dimensionality using principal component analysis to obtain the second operating parameter; the principal component selection constraint factor in the principal component analysis algorithm is the COP of the air conditioning system. S3. The least squares method is used to fit a polynomial function model based on the second running parameters, and the parameters of the polynomial function model are optimized by combining the fitting error evaluation; In the process of executing the global power optimization algorithm, a multi-unit collaborative weight coefficient is introduced; the multi-unit collaborative weight coefficient is used to characterize the response capability and synergistic effect between different air conditioning units in the energy consumption optimization process; The collaborative weighting coefficient is dynamically updated based on the historical energy efficiency, real-time response speed, and load capacity of each air conditioning unit. The global power optimization algorithm is a weighted particle swarm optimization algorithm (WPSO). The WPSO algorithm dynamically adjusts the search strategy based on cooperative weights during the power allocation process, specifically including: During the initialization phase, the initial position and velocity of individual particles are adjusted according to the cooperative weight coefficients of each unit. During the fitness assessment process, a dual-objective fusion assessment is performed based on the total energy consumption and response robustness of the multi-unit air conditioning system using a fitness function. During the velocity and position update process, a cooperative guidance factor is introduced, and the particle search weight direction is dynamically adjusted based on the cooperative guidance factor; the cooperative guidance factor is used to guide highly cooperative units to preferentially obtain more power allocation space.

Images