Multi-brand central air conditioner authentication adjusting method and system based on mathematical hybrid model driving

Abstract

The invention relates to the field of adjustment and optimization of a central air-conditioning system, in particular to a multi-brand central air-conditioning authentication adjustment method based on mathematical hybrid model driving, which comprises the following steps of: constructing a comprehensive adjustment model of a multi-brand central air-conditioning system based on a mathematical hybrid model; performing parameter identification on the operation data of different brands of central air conditioners by utilizing a genetic algorithm, and optimizing a comprehensive adjustment model; an authentication adjustment expert table is predicted and generated through a comprehensive adjustment model of the multi-brand central air-conditioning system, and optimization adjustment is carried out by utilizing the characteristics of different brand equipment; and a closed-loop control system based on a mathematical hybrid model is constructed, a self-learning mechanism is combined, and an adjustment strategy is continuously optimized to realize automatic adjustment and authentication of the multi-brand central air conditioners. According to the method, the complex adjustment requirement of a multi-brand system can be met, the optimal energy-saving effect can be achieved under different load and environment conditions, and therefore the overall energy efficiency of a building is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of regulation and optimization of central air-conditioning systems, and in particular to a multi-brand central air-conditioning certification regulation method and system driven by a mathematical hybrid model. Background Art

[0002] With the intensification of the global energy crisis and the improvement of environmental protection requirements, building energy conservation has become one of the key topics of social concern. As the main energy-consuming equipment in modern buildings, central air-conditioning has a particularly prominent energy consumption problem. Statistics show that the energy consumption of central air-conditioning systems accounts for 40% to 60% of the total energy consumption of buildings, and most of the equipment is inefficient when working under partial load conditions, resulting in serious waste of energy. Therefore, improving the operating efficiency of central air-conditioning systems and reducing energy consumption have become important links in building energy-saving transformation. Most of the existing central air-conditioning control systems are adjusted according to the working characteristics of a single brand of equipment, which is difficult to cope with the complex operating environment of multi-brand and multi-model equipment. In addition, most traditional central air-conditioning systems use static control methods and cannot make real-time adjustments according to the actual environment and dynamic changes in load, resulting in poor adjustment accuracy and high energy consumption. For example, many air-conditioning equipment still operates in full-load mode when the load changes, resulting in energy waste. The central air-conditioning system itself has a complex structure, and the mutual influence between equipment further increases the difficulty of energy-saving optimization.

[0003] At present, there are some optimization and adjustment methods based on a single device on the market. These methods are usually based on physical models or simple linear models for control. However, since the equipment in the central air-conditioning system involves multiple brands and models, a single model is often difficult to apply to complex multi-brand systems. Data-driven models, such as artificial neural networks, have certain advantages in dealing with complex nonlinear problems, but their disadvantages are that they rely on a large amount of historical data and have poor prediction effects on unlearned areas. Therefore, how to combine physical models and data-driven models to propose a certification adjustment method that is suitable for multi-brand air-conditioning equipment is an important direction of current research. In addition, with the development of the Internet of Things and big data technology, the amount of operating data of air-conditioning systems continues to increase, which provides new possibilities for intelligent adjustment and optimization. By collecting and analyzing real-time data, the operating status of air-conditioning equipment can be understood more accurately, thereby realizing dynamic optimization control. However, the existing control system still lacks an effective mechanism to make full use of these data for real-time adjustment and optimization. Especially in multi-brand systems, the performance and operating mechanisms of different devices are different, and the existing single control method cannot meet complex needs. Summary of the invention

[0004] The purpose of the present invention is to provide a method for the certification and adjustment of central air conditioners of multiple brands driven by a mathematical hybrid model. The method optimizes and adjusts central air conditioner systems of different brands by combining a physical model with an artificial neural network model to achieve precise control and energy saving. Specifically, a method for the certification and adjustment of central air conditioners of multiple brands driven by a mathematical hybrid model is provided, which includes the following steps:

[0005] S1: Based on the mathematical hybrid model, a comprehensive regulation model of multi-brand central air-conditioning systems is constructed;

[0006] S2: Use genetic algorithms to identify parameters of central air conditioners of different brands and optimize the comprehensive adjustment model;

[0007] S3: Generate a certified adjustment expert table through the prediction of the comprehensive adjustment model of multi-brand central air-conditioning systems, and optimize the adjustment by using the characteristics of different brands of equipment;

[0008] S4: Build a closed-loop control system based on a mathematical hybrid model, combine it with a self-learning mechanism, and continuously optimize the adjustment strategy to achieve automatic adjustment and certification of central air conditioners of multiple brands.

[0009] Furthermore, the mathematical hybrid model includes a gray box physical model of the central air conditioner and a cooling capacity prediction model of the BP neural network model, and the two are integrated to form a comprehensive regulation model for multi-brand central air conditioner systems.

[0010] Furthermore, actual operation data was used to determine the unknown parameters in the energy consumption model through model identification technology, and finally a gray box model suitable for predicting the energy consumption of the central air-conditioning cold source system was formed. The relevant formula is as follows:

[0011] Among them, the model of chiller energy consumption based on operating COP is as follows:

[0012]

[0013] Where: Q e is cooling load, kW; T chws is the chilled water supply temperature, °C; T cwi is the cooling water inlet temperature, °C; P chiller is the energy consumption of the chiller, kW; c1~c8 are model coefficients.

[0014] The energy consumption model of the chilled water pump can be modeled by a polynomial of the chilled water flow rate:

[0015]

[0016] Where: P chp is the energy consumption of the chilled water pump, kW; G chp is the chilled water pump flow, m3 / h; c is the specific heat capacity of water, kJ / (kg·K); ΔT chw is the temperature difference between supply and return water of chilled water; d1~d3 are model coefficients.

[0017] The cooling water pump energy consumption model can be modeled by a polynomial of the cooling water flow rate:

[0018]

[0019] Q con =Q e +P chiller (6)

[0020]

[0021] Where: P cp is the energy consumption of cooling water pump, kW; G cp is the cooling water pump flow, m 3 / h; ΔT cw Q is the temperature difference between cooling water supply and return water, ℃; con is the condensation heat, kW; ρ w is the density of water, kg / m 3 ; e1~e3 are model coefficients;

[0022] The cooling tower heat exchange model is as follows:

[0023]

[0024] Cooling tower fan energy consumption can be modeled by the fan air flow polynomial:

[0025]

[0026] Where: T wb is the outdoor air wet bulb temperature, ℃; m cw is the cooling water mass flow rate, kg / s; m air is the cooling tower air mass flow rate, kg / s; ρ air is the density of air, kg / m 3 ;P fan is the energy consumption of cooling tower fan, kW; G fan is the cooling tower fan flow rate, m 3 / h; g1~g7, f1~f3 are model coefficients;

[0027] Calculation of total energy consumption of cold source system operation:

[0028] P total =P chiller +P chp +P cp +P fan (11)

[0029] Where: P total is the total energy consumption of the cold source system, kW.

[0030] Furthermore, a cooling capacity prediction model of a BP neural network model is used to predict the equipment energy consumption of a central air conditioning cold source system. The BP neural network model uses historical input and output data to train the network, thereby determining the weights and bias values ​​between each layer; then, the new input data is sent to the trained network to predict the corresponding output results; specifically, when predicting the energy consumption of a chiller, the key factors affecting energy consumption (Q e ,T cwi ,T chws ,ΔT chw ,ΔT cw ), and define these factors as the input variables of the model (denoted as I), and define the total energy consumption of the chiller P chiller Set as the output variable of the model (denoted as O), the following are the specific calculation steps for energy consumption prediction using BP neural network:

[0031]

[0032] f2(x)=x (15)

[0033] In the formula: I is the input variable; H is the hidden layer variable; O is the output variable; w and b are the network layer weights and thresholds; f is the transfer function.

[0034] Furthermore, the gray box model and the BP neural network model are combined in the high-precision prediction interval of each model. The specific combination method is as follows:

[0035] Using K-means clustering technology, the cluster center representing the characteristics of the data set is identified, and the similarity between the input data and the training data is evaluated. Specifically, the membership function value between the input data and the cluster center is calculated to measure their closeness. The membership calculation formula is shown in (16) to (17):

[0036]

[0037] Where: x is the n-dimensional input data point; m ij is the cluster center point; d i To input data to C i The Euclidean distance of the cluster center; I is the membership of the input data to the training data.

[0038] Furthermore, a genetic algorithm is used to optimize the operating parameters of multi-brand equipment in the central air-conditioning system. Through multiple iterative calculations, the optimal parameter combination with the lowest energy consumption is found, as follows:

[0039] (1) Initialize the population. For central air conditioning equipment, randomly generate N individuals. Each individual represents a set of parameters. The population consists of these N parameter combinations.

[0040] (2) Design a fitness function to evaluate the performance of each individual (parameter combination) in the population. This fitness function needs to be able to accurately reflect the actual operation and performance indicators of the central air-conditioning equipment;

[0041] (3) According to the fitness function, individuals with higher fitness are selected from the population. These individuals will be copied to the next generation as representatives of excellent genes;

[0042] (4) Perform crossover and mutation operations on the selected individuals to generate N new individuals in the next generation population. The crossover operation simulates the reproduction process in biological genetics, while the mutation operation introduces new genetic variation to increase the diversity of the population;

[0043] (5) Check whether the preset optimization goals have been achieved, such as parameter accuracy, system performance improvement, etc. If the expected goals have been achieved, the algorithm terminates; otherwise, return to step 3 and continue the iterative optimization process until the optimal parameter combination is found.

[0044] Furthermore, the certified regulation expert table is dynamically adjusted according to the real-time environmental data, and the optimal control strategy is automatically selected in combination with the historical operating data and the current environmental conditions.

[0045] Furthermore, the self-learning mechanism automatically updates the weights and parameters in the mathematical hybrid model based on the historical operating data of the central air-conditioning system to improve the prediction accuracy and regulation ability of the model.

[0046] According to another aspect of the present invention, a multi-brand central air-conditioning certification adjustment system is provided, comprising:

[0047] The comprehensive regulation model building unit builds a comprehensive regulation model for multi-brand central air-conditioning systems based on a mathematical hybrid model;

[0048] The model optimization unit uses genetic algorithms to identify parameters of the operating data of central air conditioners of different brands and optimize the comprehensive adjustment model;

[0049] Equipment certification and adjustment unit: Generates a certification and adjustment expert table through the comprehensive adjustment model prediction of multi-brand central air-conditioning systems, and optimizes and adjusts equipment using the characteristics of different brands;

[0050] Adjustment strategy optimization unit: Build a closed-loop control system based on a mathematical hybrid model, combine it with a self-learning mechanism, and continuously optimize the adjustment strategy to achieve automatic adjustment and certification of central air conditioners of multiple brands.

[0051] Furthermore, a model integration unit is provided on the comprehensive regulation model construction unit, which is used to integrate the gray box physical model of the central air-conditioning and the cooling capacity prediction model of the BP neural network model to form a comprehensive regulation model of the multi-brand central air-conditioning system.

[0052] The present invention has significant advantages in the field of authentication and adjustment of central air-conditioning systems. It not only improves the energy efficiency and precise control ability of the system, but also effectively responds to the complex operation requirements of multi-brand and multi-model air-conditioning equipment. In order to realize the intelligent management of multi-brand central air-conditioning systems, the present invention proposes an authentication and adjustment method based on mathematical hybrid model driving. First, by combining the physical model with the artificial neural network model, a comprehensive model adapted to multi-brand air-conditioning equipment is established to ensure that the system can accurately simulate and optimize the operating characteristics of various types of equipment; secondly, in terms of parameter identification and optimization, a genetic algorithm is used to accurately identify and optimize multi-dimensional operating data, ensuring the stability and efficiency of the system in a complex environment; then, through the generation of expert tables and the formulation of adjustment rules, the characteristics of different brands of equipment are optimized and adjusted, achieving a balance between energy saving effect and user comfort; finally, through closed-loop control and self-learning mechanisms, the system can dynamically adjust the operating parameters according to real-time data, and continuously optimize the adjustment strategy, thereby greatly improving the energy efficiency and reliability of the system. In summary, the present invention not only realizes the precise control and energy-saving optimization of the central air-conditioning system, but also takes into account the diversity of equipment and the complexity of the operating environment, providing strong technical support for building energy saving and sustainable development. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 is a flow chart of the present invention;

[0054] Figure 2 Average relative error diagram of energy consumption prediction of cold source system;

[0055] Figure 3 Genetic algorithm flow chart;

[0056] Figure 4 Schematic diagram of test data structure;

[0057] Figure 5 Central air conditioning energy optimization control system block diagram;

[0058] Figure 6 Energy consumption change diagram of constant temperature control system;

[0059] Figure 7 Real-time energy consumption comparison chart of control systems. DETAILED DESCRIPTION

[0060] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0061] refer to Figure 1 The present invention provides a multi-brand central air conditioner authentication and adjustment method based on a mathematical hybrid model, comprising the following steps:

[0062] S1: Based on the mathematical hybrid model, a comprehensive regulation model of multi-brand central air-conditioning systems is constructed;

[0063] In step S1, a comprehensive adjustment model of a multi-brand central air-conditioning system is constructed by combining the physical model and the neural network model. The specific modeling process is as follows:

[0064] (1) Gray box physical model construction

[0065] In the present invention, the wet bulb temperature of the outdoor environment and the required cooling load are regarded as uncontrolled parameters. Through in-depth analysis of the impact of equipment control on energy consumption, several controllable parameters are determined, including the supply water temperature of the chilled water, the supply and return water temperature difference, the cooling water inlet temperature of the chiller, and the inlet and outlet water temperature difference of the cooling water. Based on these input and output parameters, a gray box model for energy consumption analysis of the central air conditioning cold source system is constructed, and the relevant formulas are shown in equations (1) to (11).

[0066] Using actual operation data, model identification technology is used to determine the unknown parameters in the energy consumption model, and finally a gray box model suitable for predicting the energy consumption of the central air-conditioning cold source system is formed.

[0067] Among them, the model of chiller energy consumption based on operating COP is as follows:

[0068]

[0069] Where: Q e is cooling load, kW; T chws is the chilled water supply temperature, °C; T cwi is the cooling water inlet temperature, °C; P chiller is the energy consumption of the chiller, kW; c1~c8 are model coefficients.

[0070] The energy consumption model of the chilled water pump can be modeled by a polynomial of the chilled water flow rate:

[0071]

[0072] Where: P chp is the energy consumption of the chilled water pump, kW; G chp is the chilled water pump flow, m 3 / h; c is the specific heat capacity of water, kJ / (kg·K); ΔT chw is the temperature difference between supply and return water of chilled water; d1~d3 are model coefficients.

[0073] The cooling water pump energy consumption model can be modeled by a polynomial of the cooling water flow rate:

[0074]

[0075] Q con =Q e +P chiller (6)

[0076]

[0077] Where: P cp is the energy consumption of cooling water pump, kW; G cp is the cooling water pump flow, m 3 / h; ΔT cw Q is the temperature difference between cooling water supply and return water, ℃; con is the condensation heat, kW; ρ w is the density of water, kg / m 3 ; e1~e3 are model coefficients.

[0078] The cooling tower heat exchange model is as follows:

[0079]

[0080] Cooling tower fan energy consumption can be modeled by the fan air flow polynomial:

[0081]

[0082] Where: T wb is the outdoor air wet bulb temperature, ℃; m cw is the cooling water mass flow rate, kg / s; m air is the cooling tower air mass flow rate, kg / s; ρ air is the density of air, kg / m 3 ;P fan is the energy consumption of cooling tower fan, kW; G fan is the cooling tower fan flow rate, m 3 / h; g1~g7, f1~f3 are model coefficients.

[0083] Calculation of total energy consumption of cold source system operation:

[0084] P total =P chiller +P chp +P cp +P fan (11)

[0085] Where: P total is the total energy consumption of the cold source system, kW.

[0086] (2) Construction of BP neural network model

[0087] When predicting the energy consumption of central air conditioning cooling system, the BP neural network model relies on historical input and output data to train the network, thereby determining the weights and bias values ​​between each layer. Subsequently, new input data is fed into the trained network to predict the corresponding output results. For example, when predicting the energy consumption of a chiller, the key factors affecting energy consumption (Q e ,T cwi ,T chws ,ΔT chw ,ΔT cw ), and define these factors as the input variables of the model (denoted as I), and define the total energy consumption of the chiller P chiller Set as the output variable of the model (denoted as O), the following are the specific calculation steps for energy consumption prediction using BP neural network:

[0088]

[0089]

[0090] f2(x)=x (15)

[0091] In the formula: I is the input variable; H is the hidden layer variable; O is the output variable; w and b are the network layer weights and thresholds; f is the transfer function.

[0092] (3) Integration of mathematical hybrid models

[0093] Under different validation sets, the gray box model and the BP neural network model each have their own advantage intervals in energy consumption prediction. In order to further improve the accuracy of the prediction, the high-precision prediction intervals of the two models are combined, and the present invention proposes a method for combining the advantage intervals of the two models. In addition, using K-means clustering technology, the cluster center representing the characteristics of the data set can be identified, and the similarity between the model input data and the training data can be evaluated accordingly. Specifically, the present invention measures the degree of proximity between the input data and the cluster center by calculating the membership function value, and the membership calculation formula is shown in (16) to (17):

[0094]

[0095] Where: x is the n-dimensional input data point; m ij is the cluster center point; d i To input data to C i The Euclidean distance of the cluster center; I is the membership of the input data to the training data.

[0096] This study uses the K-means clustering algorithm to analyze the data set. The specific implementation steps include: first, set the number of clusters k; second, randomly select k data points as the initial cluster centers; then, calculate the distance of each data point to each cluster center and classify it into the cluster represented by the nearest cluster center; finally, iteratively update the cluster center of each cluster until a specific convergence condition is met. In this invention, the square error criterion is selected as the evaluation criterion for clustering effect. The specific expression of this criterion is shown in formula (18).

[0097]

[0098] Where: E is the sum of the squared errors of all objects in the data set; p is the point in the data set; m i It is a class C i The cluster center of .

[0099] S2: Use genetic algorithms to identify parameters of central air conditioners of different brands and optimize the comprehensive adjustment model;

[0100] Using genetic algorithms to identify and optimize parameters of central air-conditioning equipment from multiple brands:

[0101] In the central air conditioning model, for simple functions that can be expanded by Taylor series, the least square method can be directly applied for fitting. However, for many functions that cannot be expanded by Taylor series or whose expansion process is too complicated, the identification of model parameters is not suitable for the least square method. In this case, the present invention realizes parameter identification by applying genetic algorithm.

[0102] Genetic algorithms draw on the core concepts of biological genetics, in which genes are the basic units that make up chromosomes and determine the genetic characteristics of organisms. Specific loci on chromosomes carry specific genetic information, which together constitute the genotype of the organism, and there is a close relationship between the genotype and the external characteristics of the organism, namely the phenotype. The phenotype is the result of the interaction between the genotype and the external environment. Within the framework of genetic algorithms, the genotype of an individual is usually represented by a one-dimensional binary code, where the locus corresponds to a specific position in the code and the allele corresponds to a specific value in the code. The goal of the genetic algorithm is to gradually improve the fitness of individuals in the population through an iterative optimization process, and ultimately approach the optimal solution to the problem.

[0103] The basic steps of genetic algorithm are as follows:

[0104] (1) Initialize the population. For central air conditioning equipment, randomly generate N individuals. Each individual represents a set of parameters. The population consists of these N parameter combinations.

[0105] (2) Design a fitness function to evaluate the performance of each individual (parameter combination) in the population. This fitness function needs to be able to accurately reflect the actual operation and performance indicators of the central air-conditioning equipment;

[0106] (3) According to the fitness function, individuals with higher fitness are selected from the population. These individuals will be copied to the next generation as representatives of excellent genes;

[0107] (4) Perform crossover and mutation operations on the selected individuals to generate N new individuals in the next generation population. The crossover operation simulates the reproduction process in biological genetics, while the mutation operation introduces new genetic variation to increase the diversity of the population;

[0108] (5) Check whether the preset optimization goals have been achieved, such as parameter accuracy, system performance improvement, etc. If the expected goals have been achieved, the algorithm terminates; otherwise, return to step 3 and continue the iterative optimization process until the optimal parameter combination is found.

[0109] S3: Generate a certified adjustment expert table through the prediction of the comprehensive adjustment model of multi-brand central air-conditioning systems, and optimize the adjustment by using the characteristics of different brands of equipment;

[0110] 8. According to a multi-brand central air-conditioning certification and adjustment method based on a mathematical hybrid model as described in claim 1, it is characterized in that the self-learning mechanism in step S4 automatically updates the weights and parameters in the mathematical hybrid model according to the historical operation data of the central air-conditioning system to improve the prediction accuracy and adjustment ability of the model.

[0111] Here, the main purpose is to determine the operating parameter configuration of the central air-conditioning system with the lowest energy consumption under specific cooling requirements based on genetic algorithms. Since the genetic algorithm needs to be iterated multiple times during program execution, the calculation time is long and it cannot meet the needs of real-time control. Therefore, a method of pre-calculating and establishing an expert database is designed. The certification adjustment expert table is dynamically adjusted according to real-time environmental data, combined with historical operating data and current environmental conditions, so as to quickly retrieve the system parameter combination with the best energy consumption under given cooling capacity conditions, so as to automatically select the optimal control strategy to meet the timeliness requirements of field control.

[0112] 1. Test data simplification

[0113] The ultimate goal of central air conditioning is to control the indoor temperature. The DOE-2 standard program developed by the US National Energy Administration can be used to calculate the relationship between the output cooling capacity and the indoor temperature. The simplified DOE-2 standard program needs to provide indoor temperature, outdoor temperature and other comprehensive factors such as Figure 4 .

[0114] (1) Indoor temperature

[0115] The indoor temperature is divided into two categories: high temperature and low temperature. The temperature below 23.5℃ is defined as low temperature and is represented by "20", and the temperature above 23.5℃ is defined as high temperature and is represented by "27".

[0116] (2) Outdoor temperature

[0117] Due to the large range of outdoor temperature fluctuations, the outdoor temperature is divided into three categories: low temperature, high temperature, and very high temperature. "23" is used to indicate below 26℃, "29" is used to indicate between 26℃ and 32℃, and "35" is used to indicate above 32℃.

[0118] (3) Other comprehensive factors

[0119] Other comprehensive factors include:

[0120] (1) Heat transmitted from external walls, roofs, floors, etc.

[0121] (2) Heat that enters the room through glass windows. It can be divided into heat conducted by glass windows and heat absorbed and transmitted by solar radiation through windows;

[0122] (3) Heat from outdoor air penetrating through window gaps;

[0123] (4) heat dissipation from the human body;

[0124] (5) Heat dissipation of indoor equipment;

[0125] The test table of the control expert table tries to cover the test data combinations as shown in Table 1.

[0126] Table 1 Expert table test data table

[0127] Heat exchange rate Outdoor temperature Indoor temperature Output temperature difference 1 35 27 From -10 to 1, interval 0.1 1 35 27 10 2 35 27 From -10 to 1, interval 0.1 2 35 27 10 1 29 27 From 0 to 1, with intervals of 0.1 1 29 27 10 2 29 27 From -10 to -0.9, interval 0.1 ... ... ... ...

[0128] 2. Simplify control variables

[0129] According to the system hybrid model, the control variable of the system is the energy consumption of each part of the system, but in actual control, the energy consumption is not directly controlled. According to the physical model and actual engineering situation, the corresponding relationship between energy consumption and control variables is established as shown in Table 2:

[0130] Table 2 Simplified table of control variables

[0131]

[0132]

[0133] This part is mainly about simplifying the test data, involving the classification of indoor and outdoor temperatures of the central air-conditioning system and the influence of other comprehensive factors. The factors such as indoor and outdoor temperature and heat conduction are simplified through the DOE-2 standard procedure, and an "expert table" is generated to cover different test data combinations. This is closely related to the generation of the certification adjustment expert table through the comprehensive adjustment model. The role of the expert table is to quickly find the optimal combination of central air-conditioning operating parameters according to the actual environmental conditions, so as to achieve the best energy-saving effect under different conditions. The core task of step 3 is to predict and generate the expert table through the adjustment model, and the simplified data part is to ensure that the test data covers a variety of environmental change scenarios, so that the expert table can quickly respond to the actual situation and provide a suitable adjustment plan.

[0134] S4: Build a closed-loop control system based on a mathematical hybrid model, combine it with a self-learning mechanism, and continuously optimize the adjustment strategy to achieve automatic adjustment and certification of central air conditioners of multiple brands.

[0135] Here, a closed-loop control strategy is proposed to optimize the energy consumption of a central air-conditioning system. Figure 5 As shown in Figure 1, the system dynamically calculates the deviation between the target temperature and the actual temperature by monitoring the external environmental parameters and indoor temperature in real time. The system's built-in expert database provides the optimal control parameters based on these deviations, thereby achieving precise adjustment of the central air-conditioning system. To ensure the continued effectiveness of the control strategy, the system will use the latest operating data to update its knowledge base and combine mathematical models with actual experience to generate a more accurate control parameter table to optimize future control decisions.

[0136] In actual operation, the system determines the optimal control parameters by querying the energy-saving control table based on the real-time monitored indoor and outdoor temperature difference, the equipment operation time, and the deviation between the preset target temperature and the current actual temperature, and then implements the adjustment. In the constant temperature control mode, the system sets a constant target temperature and uses a hybrid model to predict energy consumption, such as Figure 6 Compared with conventional control methods, this control strategy shows significant advantages in energy saving, such as Figure 7 As shown, the energy saving effect is as high as 52.89%.

[0137] Traditional thermostatic control strategies often do not consider the actual use of the building when adjusting the indoor temperature, but set the temperature uniformly to a fixed value. This practice leads to unnecessary energy consumption in many scenarios. To overcome this limitation, this study proposes an energy consumption control method based on a hybrid model of a central air-conditioning system. This method considers the indoor temperature requirements in different time periods, weights these requirements, and reasonably allocates energy within the preset total energy consumption framework to achieve energy saving goals.

[0138] This energy consumption-based control strategy not only achieves significant results in energy saving, but also provides an innovative perspective for the regulation of air conditioning systems. It enables central air conditioning systems to operate more efficiently and better adapt to the changing usage environment and user needs.

[0139] According to another aspect of the present invention, a multi-brand central air-conditioning certification adjustment system is provided, comprising:

[0140] The comprehensive regulation model building unit builds a comprehensive regulation model for multi-brand central air-conditioning systems based on a mathematical hybrid model;

[0141] The model optimization unit uses genetic algorithms to identify parameters of the operating data of central air conditioners of different brands and optimize the comprehensive adjustment model;

[0142] Equipment certification and adjustment unit: Generates a certification and adjustment expert table through the comprehensive adjustment model prediction of multi-brand central air-conditioning systems, and optimizes and adjusts equipment using the characteristics of different brands;

[0143] Adjustment strategy optimization unit: Build a closed-loop control system based on a mathematical hybrid model, combine it with a self-learning mechanism, and continuously optimize the adjustment strategy to achieve automatic adjustment and certification of central air conditioners of multiple brands.

[0144] As an improvement to the solution, a model integration unit is provided on the comprehensive regulation model construction unit, which is used to integrate the gray box physical model of the central air-conditioning and the cooling capacity prediction model of the BP neural network model to form a comprehensive regulation model for multi-brand central air-conditioning systems.

[0145] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-brand central air conditioning certification adjustment method based on a mathematical hybrid model, characterized in that: The following steps are included: Based on the mathematical hybrid model, a comprehensive regulation model for multi-brand central air-conditioning systems is constructed; Genetic algorithms are used to identify parameters of central air conditioners of different brands and optimize the comprehensive adjustment model; Generate a certified adjustment expert table through the prediction of the comprehensive adjustment model of multi-brand central air-conditioning systems, and optimize the adjustment by using the characteristics of different brands of equipment; Build a closed-loop control system based on a mathematical hybrid model, combine it with a self-learning mechanism, and continuously optimize the adjustment strategy to achieve automatic adjustment and certification of central air conditioners of multiple brands.

2. According to claim 1, a method for multi-brand central air conditioning authentication and adjustment based on mathematical hybrid model driving is characterized in that: The mathematical hybrid model includes a gray box physical model of the central air conditioner and a cooling capacity prediction model of the BP neural network model, and the two are integrated to form a comprehensive regulation model for a multi-brand central air conditioner system.

3. According to claim 2, a method for multi-brand central air conditioning authentication and adjustment based on mathematical hybrid model driving is characterized in that: Using actual operation data, the unknown parameters in the energy consumption model are determined through model identification technology, and finally a gray box model suitable for predicting the energy consumption of the central air-conditioning cold source system is formed. The relevant formula is as follows: Among them, the model of chiller energy consumption based on operating COP is as follows: Where: Q e is the cooling load, in kW; T chws T is the chilled water supply temperature, unit: °C; cwi For cooling water Inlet water temperature, unit: °C; P chiller is the energy consumption of the chiller, in kW; c1~c8 are model coefficients; The energy consumption model of the chilled water pump can be modeled by a polynomial of the chilled water flow rate: Where: P chp is the energy consumption of the chilled water pump, in kW; G chp is the chilled water pump flow rate, unit: m 3 / h; c is the specific heat capacity of water, unit is kJ / (kg·K); ΔT chw is the temperature difference between supply and return water of chilled water; d1~d3 are model coefficients; The energy consumption model of the cooling water pump can be modeled by a polynomial of the cooling water flow rate: Q con =Q e +P chiller (6) Where: P cp is the energy consumption of cooling water pump, unit is kW; G cp is the cooling water pump flow rate, unit: m 3 / h; ΔT cw Q is the temperature difference between cooling water supply and return water, unit: °C; con is the condensation heat, unit kW; ρ w is the density of water, in kg / m 3 ; e1~e3 are model coefficients; The cooling tower heat exchange model is as follows: Cooling tower fan energy consumption can be modeled by the fan air flow polynomial: Where: T wb is the outdoor air wet bulb temperature, unit: °C; m cw is the cooling water mass flow rate, unit: kg / s; m air is the cooling tower air mass flow rate, unit: kg / s; ρ air is the density of air, in kg / m 3 ;P fan is the energy consumption of cooling tower fan, unit is kW; G fan is the cooling tower fan flow rate, unit: m 3 / h; g1~g7, f1~f3 are model coefficients; Calculation of total energy consumption of cold source system operation: P total =P chiller +P chp +P cp +P fan (11) Where: P total It is the total energy consumption of the cold source system, in kW.

4. According to claim 2, a method for multi-brand central air conditioning authentication and adjustment based on mathematical hybrid model driving is characterized in that: The cooling capacity prediction model of the BP neural network model is used to predict the equipment energy consumption of the central air conditioning cold source system. The BP neural network model uses historical input and output data to train the network to determine the weights and bias values ​​between each layer; then, the new input data is sent to the trained network to predict the corresponding output results; specifically, when predicting the energy consumption of the chiller, the key factors affecting the energy consumption (Q e ,T cwi ,T chws ,ΔT chw ,ΔT cw ), and define these factors as the input variables of the model as I, and the total energy consumption of the chiller as P chiller The output variable of the model is set as O. The following are the specific calculation steps for energy consumption prediction using BP neural network: f2(x)=x (15) In the formula: I is the input variable; H is the hidden layer variable; O is the output variable; w and b are the network layer weights and thresholds; f is the transfer function.

5. According to claim 2, a method for multi-brand central air conditioning authentication and adjustment based on mathematical hybrid model driving is characterized in that: The gray box model and the BP neural network model are combined in the high-precision prediction interval of each model. The specific combination method is as follows: Using K-means clustering technology, the cluster center representing the characteristics of the data set is identified, and the similarity between the model input data and the training data is evaluated. Specifically, the membership function value between the input data and the cluster center is calculated to measure their closeness. The membership calculation formula is shown in (16) to (17): Where: x is the n-dimensional input data point; m ij is the cluster center point; d i To input data to C i The Euclidean distance of the cluster center; I is the membership of the input data to the training data.

6. According to claim 1, a method for multi-brand central air conditioning authentication and adjustment based on mathematical hybrid model driving is characterized in that: Genetic algorithm is used to optimize the operating parameters of multi-brand equipment in the central air-conditioning system. Through multiple iterative calculations, the optimal parameter combination with the lowest energy consumption is found. The details are as follows: Initialize the population. For central air conditioning equipment, randomly generate N individuals. Each individual represents a set of parameters. The population consists of these N parameter combinations. Design a fitness function to evaluate the performance of each individual (parameter combination) in the population; According to the fitness function, individuals with higher fitness are selected from the population. These individuals will be copied to the next generation as representatives of excellent genes. Perform crossover and mutation operations on the selected individuals to generate N new individuals in the next generation population; Check whether the parameter accuracy and system performance improvement have reached the preset optimization goals, and the algorithm terminates; otherwise, return to continue the iterative optimization process until the optimal parameter combination is found.

7. According to claim 1, a method for multi-brand central air conditioning authentication and adjustment based on mathematical hybrid model driving is characterized in that: The certified regulation expert table is dynamically adjusted according to real-time environmental data, and automatically selects the optimal control strategy based on historical operating data and current environmental conditions.

8. According to claim 1, a method for multi-brand central air conditioning authentication and adjustment based on mathematical hybrid model driving is characterized in that: The self-learning mechanism in step S4 automatically updates the weights and parameters in the mathematical hybrid model based on the historical operating data of the central air-conditioning system to improve the prediction accuracy and regulation capability of the model.

9. A multi-brand central air conditioning certification and adjustment system, characterized in that: include, The comprehensive regulation model building unit builds a comprehensive regulation model for multi-brand central air-conditioning systems based on a mathematical hybrid model; The model optimization unit uses genetic algorithms to identify parameters of the operating data of central air conditioners of different brands and optimize the comprehensive adjustment model; Equipment certification and adjustment unit: Generates a certification and adjustment expert table through the comprehensive adjustment model prediction of multi-brand central air-conditioning systems, and optimizes and adjusts equipment using the characteristics of different brands; Adjustment strategy optimization unit: Build a closed-loop control system based on a mathematical hybrid model, combine it with a self-learning mechanism, and continuously optimize the adjustment strategy to achieve automatic adjustment and certification of central air conditioners of multiple brands.

10. A multi-brand central air conditioning authentication and adjustment system as claimed in claim 9, characterized in that: The comprehensive regulation model construction unit is provided with a model integration unit for integrating the gray box physical model of the central air conditioner and the cooling capacity prediction model of the BP neural network model to form a comprehensive regulation model of a multi-brand central air conditioner system.

Images