Multi-scene heterogeneous fusion and multi-system collaborative optimization method and system for air conditioning system of large hydropower station

Abstract

The invention provides a multi-scene heterogeneous fusion and multi-system collaborative optimization method and system for an air conditioning system of a large hydropower station. The method comprises the following steps: firstly, acquiring equipment operation data and weather environment parameters related to the air conditioning system; establishing a multi-scene isoproton model and an operation cycle adaptive fusion building cold load prediction model; an underground building load prediction model based on mechanism-data fusion is constructed based on the equipment operation parameters and unsteady state heat transfer of underground building rock wall heat storage; capturing a long-term interaction relationship between meteorological factors and personnel behavior factors based on a self-attention mechanism of Transform, and establishing a meteorological-personnel behavior multi-factor coupled building load prediction model; and in combination with cold source characteristics and equipment performance constraints of different types of water chilling units of all the sub-air-conditioning systems and different terminal requirements of underground equipment areas, offices and the like, an operation optimization method for cross-system cold source dynamic allocation and terminal parameter self-adaptive adjustment is established. Intelligent optimized operation of a large hydropower station complex building air conditioning system can be achieved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent operation technology of building air conditioning systems, specifically involving a method and system for multi-scenario heterogeneous fusion and multi-system collaborative optimization of air conditioning systems in large hydropower stations. Background Technology

[0002] Building air conditioning systems are a core component of building energy conservation and intelligent management, and their operational efficiency directly impacts the overall energy efficiency of the building and the user experience. While some progress has been made in intelligent operation technology for building air conditioning systems, common technical problems remain prominent: insufficient adaptability to multiple scenarios, lack of accurate perception and prediction of heterogeneous load demands from different functional areas (such as underground equipment spaces and above-ground office spaces); and imperfect cross-system collaboration mechanisms, with each subsystem (cooling source, terminal, and distribution) operating independently without forming a complete optimization chain, resulting in high system energy consumption, underutilization of equipment operating efficiency, and difficulty in balancing energy conservation goals with user comfort requirements.

[0003] As special-function buildings, large hydropower stations require air conditioning systems that simultaneously meet the heterogeneous demands of two typical scenarios: underground equipment areas need to maintain a stable temperature and humidity environment to ensure the safe operation of electrical equipment, and the heat dissipation load is relatively stable due to the heat storage characteristics of the rock walls; above-ground office or control room areas are greatly affected by fluctuations in personnel activity and meteorological factors, resulting in dynamic changes in load demand. However, traditional hydropower station air conditioning systems are not optimized for these different scenarios. Each subsystem operates independently, and the cooling source capacity cannot be complementaryly allocated across systems; equipment operation is not optimized based on its performance curves, often operating in low-efficiency conditions; and the parameters of terminal equipment are fixed, making it unable to adapt to load changes. This leads to insufficient environmental stability in underground equipment areas, affects the comfort of office personnel, and causes energy waste. Summary of the Invention

[0004] Addressing the common technical problems of insufficient adaptability to multiple scenarios and imperfect cross-system collaboration mechanisms in building air conditioning systems, as well as the shortcomings of existing large-scale hydropower station air conditioning systems that fail to optimize design for the heterogeneous demands of stable heat dissipation in underground areas, heat storage in rock walls, fluctuations in personnel in above-ground offices, and significant weather influences, resulting in independent operation of each subsystem leading to the inability of cold sources to complement each other, equipment not operating efficiently according to performance curves, and fixed terminal parameters that are difficult to adapt to load changes, this invention comprehensively considers the heterogeneous load characteristics of large-scale hydropower station air conditioning systems. Based on a full-link technical architecture of "multi-source data acquisition - heterogeneous load prediction - cross-system collaborative optimization - closed-loop feedback adjustment," it provides a method and system for multi-scenario heterogeneous fusion and multi-system collaborative optimization of large-scale hydropower station air conditioning systems.

[0005] To achieve the above-mentioned technical features, the objective of this invention is as follows: A method for multi-scenario heterogeneous fusion and multi-system collaborative optimization of a large-scale hydropower station air conditioning system, comprising the following steps: Step S1: Collect relevant operational data of the air conditioning system through the monitoring system, equipment sensors, and historical database within the hydropower station. Specifically, this includes operational data of each piece of equipment, environmental data of each office, and parameters and operational data of the air conditioning system's cold source and terminal equipment. Normalize the collected data and then use... The criteria are used to remove outliers; Step S2: Based on the preprocessed data in S1, a multi-scenario heterogeneous sub-model and an adaptive fusion building cooling load prediction model are established for the heterogeneous load characteristics formed by underground equipment areas where the load is mainly equipment heat dissipation and rock wall heat storage and heat transfer, and above-ground office hydropower stations where the load is mainly weather and human behavior. Step S3: Based on the preprocessed data in S1, capture the long-term interaction between meteorological factors and human behavior factors using the self-attention mechanism of Transformer, and establish a multi-factor coupled building load prediction model of meteorology and human behavior. Step S4: Based on the differences in the operating cycles of different subsystems in the central air conditioning system, time weights are assigned to the load prediction models for the underground equipment area and the office load prediction models in steps S2 and S3, and dynamically adjusted using the sliding window method. Step S5, based on the total air conditioning load obtained in step S4 Based on the characteristics of the cold source of different types of chillers in each sub-air conditioning system, the performance constraints of the equipment, and the different terminal requirements of underground equipment areas and offices, an operation optimization method for dynamic allocation of cold source and adaptive adjustment of terminal parameters across systems is established.

[0006] Preferably, step S1 specifically includes the following steps: Step S11: Obtain the real-time current, real-time power, and running time of the electrical components in the excitation transformer and electrical panel through the monitoring system; obtain the heat dissipation characteristic parameters of the excitation transformer and each electrical component through the equipment design manual. Step S12: Obtain the rock wall thickness of the hydropower station powerhouse. and material thermal conductivity Specific heat capacity ,density The parameters of the underground retaining structure, as well as the temperature inside the basement and the temperature outside the rock wall; the weather parameters of indoor and outdoor temperature and humidity and solar radiation intensity are obtained through the meteorological database; real-time personnel data in each area are obtained through the access control system, and the specific activity status of personnel is determined based on the smart work card; Step S13: Obtain data on the number of operating chillers and real-time load rate of each chiller unit in the air conditioning system, as well as the data on the cold source equipment with unit energy consumption; obtain data on the supply air temperature, air volume, water supply temperature, and fan coil unit speed of the combined air handling units in the factory building.

[0007] Preferably, step S2 specifically includes the following steps: Step S21: Based on equipment operating parameters and the unsteady heat transfer mechanism of underground building rock wall heat storage, construct an underground building load prediction model through data fusion; calculate the rock wall heat storage heat transfer load through an unsteady thermal resistance-heat capacity (RC) network, and divide the rock wall according to its actual location and structure. n Establish an RC network model for each of the three uniform layers. The unsteady heat transfer equation for each layer can be expressed as: ; In the formula, For the first i Temperature of the rock wall, °C; t For time, s; For the first i Layer and First i Thermal resistance between -1 layers, K / W; No. i Layer and First i+ Thermal resistance between layers, K / W; For the first i Heat capacity of layered rock walls ;in, Density of the rock wall For specific heat capacity, For the area of ​​the rock wall, m 2 ; Step S22: Determine the boundary conditions and initial conditions, and use the finite difference method to solve the differential equation in step S21 to obtain the temperature change of each rock wall layer over time, and then calculate the heat storage and heat transfer load of the rock wall. ; In the formula, The heat storage and heat transfer load of the rock wall is W; Temperature inside the basement, in °C; Inner layer temperature, °C; convective thermal resistance. K / W; The convective heat transfer coefficient is... ; A For the area of ​​the rock wall, m 2 ; Step S23: Based on the preprocessed data in step S11, calculate the heat dissipation load of each device: ; In the formula, For heat dissipation load; m The quantity of electrical equipment is in units. For the first i The heat dissipation coefficient of the equipment; For the first i Real-time power of the equipment, kW; For the first i The running time of the equipment h ; Further calculations yielded the cooling load of the building foundation in the underground equipment area: ; In the formula, For the building foundation cooling load; Step S24: Use the XGBoost algorithm to evaluate the residual error of the cooling load calculation model. Perform fitting calculations. The actual load of the underground equipment area is calculated from the cooling capacity output from the cold source and the terminal return air temperature. Historical office load over the past 24 hours; Input characteristic: cumulative device runtime Temperature on the outside of the rock wall Historical residual error Output: Predicted residual error value The XGBoost model was trained using historical data from the underground equipment area over the past year, and the parameters were optimized using grid search. The final cooling load for the underground equipment area is: ; In the formula, This is the final cooling load; Preferably, step S3 specifically includes the following steps: Step 31, Input Feature Construction: Outdoor Temperature Outdoor relative humidity Solar radiation intensity I Meteorological characteristics; office hours S Number of personnel N Personnel activity status F Personnel behavioral characteristics; office workload history over the past 24 hours Historical load characteristics; Step S32: Determine the Transformer model structure based on the above input features. Input layer: dimension is... Input vector; Encoder layer: a two-layer Transformer encoder, each layer containing one self-attention head for calculating attention weights between features, including the correlation between solar radiation and the number of people, and a feedforward neural network (FFN) for handling nonlinear relationships; Output layer: a fully connected layer and a linear activation layer, outputting the office building's cooling load. The hidden layer has a dimension of 64, and the self-attention mechanism is represented as follows: ; In the formula, Q For querying the matrix, K For the key matrix, V The value matrix is ​​obtained by linear transformation of the input vector. ; The model was trained using historical office data from the past year, with Adam as the optimizer and the mean squared error (MSE) of the loss function as follows: ; In the formula, m Indicates the number of samples; It is the first j One actual value; It is the first j One predicted value.

[0008] Preferably, step S4 specifically includes the following steps: Step S41: Set initial weights according to the air conditioning system's operating cycle and load characteristics: heating / cooling. ,in, Considering seasonal correction factors, the weight of meteorological features in the Transformer model is increased due to the high solar radiation intensity in summer, while the weight of human behavior features is reduced due to the low human activity level in winter. Step S42: Calculate the root mean square error of the sub-model using a 7-day sliding window. RMSE The weights are dynamically adjusted based on the magnitude of the error. ; If the underground equipment area load prediction model RMSE <5%, then Increase by 0.1, Reduce by 0.1; if the office load forecasting model RMSE <5%, then Increase by 0.1, Reduce by 0.1; In summary, the air conditioning load prediction model for large hydropower stations is expressed as follows: ; In the formula, The total air conditioning load is W; For the building load of the underground equipment area, W; For office building load, W.

[0009] Preferably, step S5 specifically includes the following steps: Step S51: Through pipeline connectivity design, cross-system complementarity of cold source capacity of different air conditioning subsystems is achieved, a cold source capacity and energy consumption characteristic matrix is ​​constructed, and the maximum output capacity of each cold source device is determined. Current actual operating load Unit energy consumption Operating status: running / standby. The objective function is to minimize the total energy consumption of the air conditioning system and maximize the average efficiency of the equipment, with the following multiple objectives: ; In the formula, m The total number of cold source devices is [number] units. For the first j The output cooling capacity of the cooling source equipment meets the requirements. ; t For runtime, h ; For the first j Unit energy consumption of the cooling source equipment, KW / KW; For the first j The operating efficiency of the equipment; Define the constraints of the objective function, including terminal load demand constraints, cooling source capacity constraints, equipment start-up and shutdown constraints, and pipeline flow constraints. The sum of the output cooling capacity of all cooling source equipment must meet the total terminal load requirement. ; In the formula, The load margin factor reserves redundant cooling capacity; The output cooling capacity of each cold source device must not exceed its maximum capacity: ; The minimum start-up and shutdown time of the chiller unit is n Minimize the time spent starting and stopping the equipment to avoid frequent starts and stops that could cause wear and tear. like and ,but ; In the formula, For the first The last shutdown time of the chiller unit; The maximum flow rate in a cross-system pipeline must not exceed the pipeline's design capacity. ; In the formula, For the first k Flow rate of a cross-system pipeline ; Set its maximum design flow rate; Based on the above objective function and constraints, the Non-Dominated Sorting Genetic Algorithm (NSGA-II) is used to solve the multi-objective optimization problem to obtain the Pareto optimal solution. Real-number encoding is used, with each chromosome representing a set of output cooling capacity allocation schemes for the cold source equipment. Based on the objective function Calculate the fitness value of each chromosome, evaluate the quality of the solution using non-dominated sorting and crowding distance; finally output the optimal cold source allocation scheme. Step S52, based on the performance curves of different types of chillers: screw chiller units POP The curves and COP curves of air-cooled heat pumps are compared to optimize the equipment's operating load rate, ensuring the equipment operates within its high-efficiency range and is adjusted based on feedback from end-user demand. Performance curves for different types of chillers were constructed, and the load rate of screw chillers was analyzed. and POP The data were obtained by fitting a quadratic polynomial. POP Performance curve equation: ; In the formula, For load factor, ; POP Performance refers to the cooling capacity generated per kW of electrical energy consumed, expressed in KW / KW. a , b , c for POP Performance curve equation coefficients; Obtain the outdoor temperature of the heat pump unit from the equipment manual. The load factor is of COP The data were obtained by fitting a bivariate quadratic polynomial. COP Performance curve equation: ; In the formula, a , b , c , d for COP Performance curve equation coefficients; With the optimization objective of maximizing the operating efficiency of the chiller unit, the objective function is: ; The constraints include: load factor constraints, end-point load demand constraints, and equipment performance curve constraints, based on... POP The maximum range of the performance curve determines the load rate range of the screw chiller unit; based on COPThe maximum range of the performance curve determines the load rate range of the heat pump unit; Based on the above objective function and constraints, mixed-integer linear programming (MILP) is used to solve the problem, considering the total load at the end. The optimal load factor allocation scheme is obtained by solving the problem. Step S53: Based on the predicted load obtained in step S4, adjust the parameters of the terminal equipment to meet the needs of the end users and establish a closed-loop feedback system. If the terminal adjustment cannot meet the needs, it will be fed back to step S52 to adjust the equipment load rate. After adjusting the equipment load rate, evaluate its efficiency loss. If the loss rate exceeds the threshold, it will be fed back to step S51 to adjust the cold source capacity allocation.

[0010] Preferably, the feedback from step S53 to step S51 to adjust the cold source capacity allocation specifically includes the following steps: First, based on the current total terminal load... Total load forecast for the next 15 minutes The load change rate is calculated, and when it exceeds the threshold, the load is judged to be either rising or falling. For the parameter adjustment of combined air handling units, air volume adjustment takes precedence over supply air temperature adjustment. When the load increases, the air volume is increased first, provided that all constraints are met. If the load demand cannot be met, the supply air temperature is then reduced. For the parameter adjustment of fan coil units in office and control room areas, the human thermal comfort index PMV is calculated based on environmental and personnel parameters, and the fan coil unit speed and supply water temperature are adjusted according to the standard value to meet thermal comfort requirements. After the terminal equipment is adjusted, the temperature is verified and feedback is received through the monitoring devices of each air-conditioned area to perform rolling optimization.

[0011] Preferably, the unsteady thermal resistance-thermal capacity network model in S21 is constructed through the following steps: Step 21: Divide the rock wall into n uniform layers according to the rock wall thickness δ, with each layer having a thickness of... ,in, ; Step 22: Calculate the thermal resistance of each layer and heat capacity ;in, The thermal conductivity of the rock wall material. A For the area of ​​the rock wall, Density of the rock wall material c Specific heat capacity of the rock wall material; Step 23: Establish the unsteady-state heat transfer equation for each layer and set the boundary conditions: convective heat transfer between the inner layer and the indoor air, contact temperature between the outer layer and the soil, and initial conditions: the average temperature of the rock wall at the beginning of the year; Step 24: Solve the differential equations using the finite difference method to obtain the temperature change of each rock wall layer over time.

[0012] Preferably, the personnel activity status in S31 is calculated using the step count data from the smart work card, and the specific steps are as follows: Step 31: The smart work card collects the step count data of the personnel in real time; Step 32: Determine the activity status based on the step count threshold: steps < 50 steps / 10 minutes is considered sitting still, so assign a value of 1; 50 ≤ steps < 200 steps / 10 minutes is considered light activity, so assign a value of 2; steps ≥ 200 steps / 10 minutes is considered walking, so assign a value of 3.

[0013] Another aspect of the present invention provides a multi-scenario heterogeneous fusion and multi-system collaborative optimization system for air conditioning systems in large hydropower stations. The system is used to implement the method, comprising: Data acquisition module: used to collect heterogeneous data from multiple sources: temperature sensors, current transformers, and smart work cards; Load forecasting module: used to run a multi-scenario heteroproton model-run-cycle adaptive weighted fusion model to predict the total load at the end point; Multi-system collaborative optimization module: used to perform dynamic allocation of cross-system cold source capacity, efficient operation constrained by equipment performance curves, and adaptive adjustment of terminal parameters, including multi-objective optimization unit, equipment efficiency optimization unit, and closed-loop feedback unit; Execution module: Used to control the operating parameters of cold source equipment: screw chillers, heat pumps, and terminal equipment: combined air handling units, fan coil units.

[0014] Preferably, the multi-objective optimization unit is a solution unit of a non-dominated sorting genetic algorithm; The equipment efficiency optimization unit is a solution unit for mixed integer linear programming; The closed-loop feedback unit is a sensor network and PLC control system.

[0015] The present invention has the following beneficial effects: This invention provides a method for multi-scenario heterogeneous fusion and multi-system collaborative optimization of air conditioning systems in large hydropower stations. This method, through a full-link architecture of "multi-source data acquisition - heterogeneous load prediction - cross-system collaborative optimization - closed-loop feedback adjustment," addresses the problem of insufficient adaptability of existing systems to heterogeneous scenarios by considering the heterogeneous needs of different air-conditioning areas in large hydropower stations. It optimizes key aspects such as dynamic allocation of cooling source capacity and adaptive adjustment of terminal parameters, enabling precise adaptation to multiple scenarios and efficient collaborative operation of air conditioning systems in large hydropower stations. This helps improve equipment operating efficiency, reduce system energy consumption, and ensure the safety of electrical equipment and the comfort of personnel. It is of great significance for energy conservation, emission reduction, and intelligent management of complex functional building energy systems in large hydropower stations. Attached Figure Description

[0016] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0017] Figure 1 This is a flowchart of the method of the present invention.

[0018] Figure 2 This is a schematic diagram of the unsteady thermal resistance-thermal capacity (RC) network model of the method of the present invention.

[0019] Figure 3 This is the technical framework of the method of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] like Figure 1 This invention discloses a method for heterogeneous fusion and multi-system collaborative optimization of air conditioning systems in large hydropower stations, comprising the following steps: Step S1: 1) Obtain real-time current, real-time power, and operating time (from start / stop records of the equipment PLC controller) of the excitation transformer and electrical components in the electrical control panel through the monitoring system; obtain equipment heat dissipation characteristic parameters such as the heat dissipation coefficient of the excitation transformer and each electrical component through the equipment design manual (e.g., the heat dissipation coefficient of the excitation transformer). This means that 60% of the input power is converted into heat dissipation. 2) Obtain the rock wall thickness of the hydropower station powerhouse. and material thermal conductivity Specific heat capacity ,density Parameters of the underground retaining structure, as well as the temperature inside the basement and the temperature outside the rock wall; weather parameters such as indoor and outdoor temperature and humidity, and solar radiation intensity are obtained through the meteorological station database; real-time personnel data in each area are obtained through the access control system (using binary variables, office hours 1 represents office hours from 8:00 to 18:00; other times are 0), and the specific activity status of personnel is determined based on the smart work card.

[0022] 3) Obtain data on the number of operating chillers and real-time load rate, unit energy consumption, and other cold source equipment data for each chiller unit in the air conditioning system; obtain data on the supply air temperature, air volume, water supply temperature, fan coil unit speed, and other terminal equipment data for the combined air handling units in the factory buildings.

[0023] Step S2: 1) A load prediction model for underground structures based on the unsteady-state heat transfer mechanism and data fusion of equipment operating parameters and underground rock wall heat storage. The heat transfer load of the rock wall heat storage is calculated using an unsteady-state thermal resistance-heat capacity (RC) network. The rock wall is divided into sections according to its actual location and structure. n An RC network model is established for each of the three uniform layers. The unsteady heat transfer equation for each layer can be expressed as: ; In the formula, For the first i Temperature of the rock wall, °C; t For time, s; The thermal resistance between the i-th layer and the (i-1)-th layer is expressed in K / W. No. i Layer and First i+1 Thermal resistance between layers, K / W; For the first i Heat capacity of layered rock walls ;in, Density of the rock wall For specific heat capacity, For the area of ​​the rock wall, m 2 ; 2) Determine the boundary conditions and initial conditions ( When the temperature of each rock wall layer is equal to the initial temperature, the differential equation in step S21 is solved using the finite difference method to obtain the temperature change of each rock wall layer over time, and then the heat storage and heat transfer load of the rock wall is calculated: ; In the formula, The heat storage and heat transfer load of the rock wall is W; Temperature inside the basement, in °C; Inner layer temperature, °C; convective thermal resistance. K / W; The convective heat transfer coefficient is... ; A For the area of ​​the rock wall, m 2 ; 3) Based on the preprocessed data in step S11, calculate the heat dissipation load of each device: ; In the formula, For heat dissipation load; m The quantity of electrical equipment is in units. For the first i The heat dissipation coefficient of the equipment; For the first i Real-time power of the equipment, kW; For the firsti The running time of the equipment h ; Further calculations yielded the cooling load of the building foundation in the underground equipment area: ; In the formula, For the building foundation cooling load; 4) The XGBoost algorithm was used to correct the residual error of the cooling load calculation model. Perform fitting calculations. The actual load of the underground equipment area is calculated from the cooling capacity output from the cold source and the terminal return air temperature. Historical office load over the past 24 hours; Input characteristic: cumulative device runtime Temperature on the outside of the rock wall Historical residual error Output: Predicted residual error value The XGBoost model was trained using historical data from the underground equipment area over the past year, and the parameters were optimized using a grid search (e.g., learning rate 0.1, tree depth 6, estimators=100). The final cooling load for the underground equipment area is: ; In the formula, This is the final cooling load; Step S3: 1) Input feature construction: outdoor temperature Outdoor relative humidity Solar radiation intensity I Meteorological characteristics; office hours S Number of personnel N Personnel activity status F Personnel behavioral characteristics; office workload history over the past 24 hours Historical load characteristics.

[0024] 2) Determine the Transformer model structure based on the above input features. Input layer: dimension is... Input vector; Encoder layer: A two-layer Transformer encoder, each layer containing one self-attention head to calculate attention weights between features, such as the relationship between solar radiation and the number of people, and a feedforward neural network (FFN) to handle non-linear relationships. Output layer: A fully connected layer (64 hidden dimensions) and a linear activation layer, outputting the office building's cooling load. The self-attention mechanism can be represented as: ; In the formula, Q For querying the matrix,K For the key matrix, V The value matrix is ​​obtained by linear transformation of the input vector. ; The model was trained using historical office data from the past year, with Adam as the optimizer and the mean squared error (MSE) of the loss function as follows: ; In the formula, m Indicates the number of samples; It is the first j One actual value; It is the first j One predicted value.

[0025] Step S4: 1) Set initial weights according to the air conditioning system's operating cycle and load characteristics (heating / cooling). Initial weights For example, the air conditioning system of a certain hydropower station has three subsystems: Central Air Conditioning #1 (cooling from May to October): The underground equipment area has a high load proportion. ; Central air conditioning units #1-#3 (cooling year-round): The underground excitation transformer room has a high load ratio. Central Air Conditioner Unit #2 and Unit #4 (Heating from November to March, cooling at other times): The central control room and lobby have a high load ratio during the heating season. ;Cooling period Central air conditioning unit #3 (heating from November to March): High load ratio on the basement level of the office building. .

[0026] 2) Considering seasonal correction factors, the weight of meteorological features in the Transformer model is increased due to the high solar radiation intensity in summer (e.g., the linear transformation coefficient is increased from 0.3 to 0.5); the weight of human behavior features is reduced due to the low human activity in winter (e.g., the linear transformation coefficient is reduced from 0.4 to 0.2).

[0027] 3) The root mean square error (RMSE) of the sub-model is calculated using a 7-day sliding window, and the weights are dynamically adjusted according to the magnitude of the error.

[0028] ; If the underground equipment area load prediction model RMSE <5%, then Increase by 0.1, Reduce by 0.1; if the office load forecasting model RMSE <5%, then Increase by 0.1, Reduce by 0.1; In summary, the air conditioning load prediction model for large hydropower stations is expressed as follows: ; In the formula, The total air conditioning load is W; For the building load of the underground equipment area, W; For office building load, W.

[0029] Step S5: 1) Through pipeline connectivity design, cross-system complementarity of cold source capacity of different air conditioning subsystems is achieved, a cold source capacity and energy consumption characteristic matrix is ​​constructed, and the maximum output capacity of each cold source device is determined. Current actual operating load Unit energy consumption Operating status (operating / standby). The objective function is: Minimizing the total energy consumption of the air conditioning system and maximizing the average efficiency of the equipment. ; In the formula, m The total number of cold source devices is [number] units. For the first j The output cooling capacity of the cooling source equipment meets the requirements. ; t For runtime, h ; For the first j Unit energy consumption of the cooling source equipment, KW / KW; For the first j The operating efficiency of the equipment; Define the constraints of the objective function, including terminal load demand constraints, cooling source capacity constraints, equipment start-up and shutdown constraints, and pipeline flow constraints. The sum of the output cooling capacity of all cooling source equipment must meet the total terminal load requirement. ; In the formula, The load margin factor reserves redundant cooling capacity; The output cooling capacity of each cold source device must not exceed its maximum capacity: ; The minimum start-up and shutdown time of the chiller unit is n Minimize the time spent starting and stopping the equipment to avoid frequent starts and stops that could cause wear and tear. like and ,but ; In the formula, For the first The last shutdown time of the chiller unit; The maximum flow rate in a cross-system pipeline must not exceed the pipeline's design capacity. ; In the formula, For the first k Flow rate of a cross-system pipeline ; Set its maximum design flow rate; Based on the above objective function and constraints, the Non-Dominated Sorting Genetic Algorithm (NSGA-II) is used to solve the multi-objective optimization problem, obtaining the Pareto optimal solution. Real-number encoding is used during encoding, with each chromosome representing a set of cold source equipment output cooling capacity allocation schemes. Based on the objective function Calculate the fitness value of each chromosome, evaluate the quality of the solution using non-dominated sorting and crowding distance, and finally output the optimal cold source allocation scheme.

[0030] For example, in the output of a Pareto optimal solution, the three screw compressors of air conditioning system #1 operate at 75% load rate. , POP =4.9); Heat pump #1 of air conditioning system #2 is operating at 70% load rate ( , COP =3.2). The overall total energy consumption is reduced by 2.3% compared to the minimum solution of the single-objective optimization, and the average equipment efficiency is improved by 6%.

[0031] 2) Based on the performance curves of different types of chillers (screw chillers) POP Curves and air-cooled heat pumps COP (Curve) to optimize equipment operating load rate, so that the equipment operates in the high-efficiency range and accepts feedback adjustments from end-user demand.

[0032] Performance curves for different types of chillers were constructed, and the load rate of screw chillers was analyzed. and POP The data were obtained by fitting a quadratic polynomial. POP Performance curve equation: ; In the formula, For load factor, ; POP Performance refers to the cooling capacity generated per kW of electrical energy consumed, expressed in KW / KW. a , b , c for POP Performance curve equation coefficients; Obtain the outdoor temperature of the heat pump unit from the equipment manual. The load factor is of COP The data were obtained by fitting a bivariate quadratic polynomial. COP Performance curve equation: ; In the formula, a , b , c , d for COP Performance curve equation coefficients; With the optimization objective of maximizing the operating efficiency of the chiller unit, the objective function is: ; The constraints include: load factor constraints, end-point load demand constraints, and equipment performance curve constraints, based on... POP The maximum range of the performance curve determines the load rate range of the screw chiller unit; based on COP The maximum range of the performance curve determines the load rate range of the heat pump unit; Based on the above objective function and constraints, mixed-integer linear programming (MILP) is used to solve the problem, considering the total load at the end. The optimal load factor allocation scheme is obtained by solving the problem. 3) Based on the predicted load obtained in step S4, adjust the terminal equipment parameters to meet the needs of end users and establish a closed-loop feedback system. If the terminal adjustment cannot meet the demand, it will be fed back to step S52 to adjust the equipment load rate. After adjusting the equipment load rate, evaluate its efficiency loss. If the loss rate exceeds the threshold, it will be fed back to step S51 to adjust the cold source capacity allocation. First, based on the current total terminal load... Total load forecast for the next 15 minutes Predicting the end-load trend for the next 15 minutes Exceeding the threshold At that time, the load is judged to be rising / falling / stable.

[0033] ; For parameter adjustment of combined air handling units, air volume adjustment takes precedence over supply air temperature adjustment. When the load increases, the air volume is increased first, provided that all constraints are met. If the load demand cannot be met, the supply air temperature is then reduced. The specific adjustment logic is shown in Table 1 below. Table 1 Adjustment Logic

[0034] For fan coil unit parameter adjustments in office and control room areas, the Human Thermal Comfort Index (PMV) is calculated based on environmental and personnel parameters, and the fan coil unit speed and water supply temperature are adjusted according to standard values ​​to meet thermal comfort requirements. After the terminal equipment is adjusted, the temperature is verified and feedback is obtained through monitoring devices in each air-conditioned area to ensure that the requirements are met, and rolling optimization is carried out.

[0035] Using Fanger priv Model calculation of human thermal comfort index: ; In the formula, M Metabolic rate of individuals (while sitting) M =1.0met, when walking M =1.5met); A Office area, in m²; The load is expressed as kW / m².

[0036] Table 2 Fan Coil Unit Parameter Adjustment Logic

[0037] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for heterogeneous fusion and multi-system collaborative optimization of air conditioning systems in large hydropower stations, characterized in that, Includes the following steps: Step S1: Collect relevant operational data of the air conditioning system through the monitoring system, equipment sensors, and historical database within the hydropower station. Specifically, this includes operational data of each piece of equipment, environmental data of each office, and parameters and operational data of the air conditioning system's cold source and terminal equipment. Normalize the collected data and then use... The criteria are used to remove outliers; Step S2: Based on the preprocessed data in S1, a multi-scenario heterogeneous sub-model and an adaptive fusion building cooling load prediction model are established for the heterogeneous load characteristics formed by underground equipment areas where the load is mainly equipment heat dissipation and rock wall heat storage and heat transfer, and above-ground office hydropower stations where the load is mainly weather and human behavior. Step S3: Based on the preprocessed data in S1, capture the long-term interaction between meteorological factors and human behavior factors using the self-attention mechanism of Transformer, and establish a multi-factor coupled building load prediction model of meteorology and human behavior. Step S4: Based on the differences in the operating cycles of different subsystems in the central air conditioning system, time weights are assigned to the load prediction models for the underground equipment area and the office load prediction models in steps S2 and S3, and dynamically adjusted using the sliding window method. Step S5, based on the total air conditioning load obtained in step S4 Based on the characteristics of the cold source of different types of chillers in each sub-air conditioning system, the performance constraints of the equipment, and the different terminal requirements of underground equipment areas and offices, an operation optimization method for dynamic allocation of cold source and adaptive adjustment of terminal parameters across systems is established.

2. The method for multi-scenario heterogeneous fusion and multi-system collaborative optimization of a large hydropower station air conditioning system according to claim 1, characterized in that, Step S1 specifically includes the following steps: Step S11: Obtain the real-time current, real-time power, and running time of the electrical components in the excitation transformer and electrical panel through the monitoring system; obtain the heat dissipation characteristic parameters of the excitation transformer and each electrical component through the equipment design manual. Step S12: Obtain the rock wall thickness of the hydropower station powerhouse. and material thermal conductivity Specific heat capacity ,density The parameters of the underground retaining structure, as well as the temperature inside the basement and the temperature outside the rock wall; the weather parameters of indoor and outdoor temperature and humidity and solar radiation intensity are obtained through the meteorological database; real-time personnel data in each area are obtained through the access control system, and the specific activity status of personnel is determined based on the smart work card; Step S13: Obtain data on the number of operating chillers and real-time load rate of each chiller unit in the air conditioning system, as well as the data on the cold source equipment with unit energy consumption; obtain data on the supply air temperature, air volume, water supply temperature, and fan coil unit speed of the combined air handling units in the factory building.

3. The method for multi-scenario heterogeneous fusion and multi-system collaborative optimization of a large hydropower station air conditioning system according to claim 1, characterized in that, Step S2 specifically includes the following steps: Step S21: Based on equipment operating parameters and the unsteady heat transfer mechanism of underground building rock wall heat storage, construct an underground building load prediction model through data fusion; calculate the rock wall heat storage heat transfer load through an unsteady thermal resistance-heat capacity (RC) network, and divide the rock wall according to its actual location and structure. n Establish an RC network model for each of the three uniform layers. The unsteady heat transfer equation for each layer can be expressed as: ； In the formula, For the first i Temperature of the rock wall, °C; t For time, s; For the first i Layer and First i Thermal resistance between -1 layers, K / W; No. i Layer and First i+ Thermal resistance between layers, K / W; For the first i Heat capacity of layered rock walls ;in, Density of the rock wall For specific heat capacity, For the area of ​​the rock wall, m 2 ; Step S22: Determine the boundary conditions and initial conditions, and use the finite difference method to solve the differential equation in step S21 to obtain the temperature change of each rock wall layer over time, and then calculate the heat storage and heat transfer load of the rock wall. ； In the formula, The heat storage and heat transfer load of the rock wall is W; Temperature inside the basement, in °C; Inner layer temperature, °C; convective thermal resistance. K / W; The convective heat transfer coefficient is... ; A For the area of ​​the rock wall, m 2 ; Step S23: Based on the preprocessed data in step S11, calculate the heat dissipation load of each device: ； In the formula, For heat dissipation load; m The quantity of electrical equipment is in units. For the first i The heat dissipation coefficient of the equipment; For the first i Real-time power of the equipment, kW; For the first i The running time of the equipment h ; Further calculations yielded the cooling load of the building foundation in the underground equipment area: ； In the formula, For the building foundation cooling load; Step S24: Use the XGBoost algorithm to evaluate the residual error of the cooling load calculation model. Perform fitting calculations. The actual load of the underground equipment area is calculated from the cooling capacity output from the cold source and the terminal return air temperature. Historical office load over the past 24 hours; Input characteristic: cumulative device runtime Temperature on the outside of the rock wall Historical residual error Output: Predicted residual error value The XGBoost model was trained using historical data from the underground equipment area over the past year, and the parameters were optimized using grid search. The final cooling load for the underground equipment area is: ； In the formula, This represents the final cooling load.

4. The method for multi-scenario heterogeneous fusion and multi-system collaborative optimization of a large hydropower station air conditioning system according to claim 3, characterized in that, Step S3 specifically includes the following steps: Step 31, Input Feature Construction: Outdoor Temperature Outdoor relative humidity Solar radiation intensity I Meteorological characteristics; office hours S Number of personnel N Personnel activity status F Personnel behavioral characteristics; office workload history over the past 24 hours Historical load characteristics; Step S32: Determine the Transformer model structure based on the above input features. Input layer: dimension is... Input vector; Encoder layer: a two-layer Transformer encoder, each layer containing one self-attention head for calculating attention weights between features, including the correlation between solar radiation and the number of people, and a feedforward neural network (FFN) for handling nonlinear relationships; Output layer: a fully connected layer and a linear activation layer, outputting the office building's cooling load. The hidden layer has a dimension of 64, and the self-attention mechanism is represented as follows: ； In the formula, Q For querying the matrix, K For the key matrix, V The value matrix is ​​obtained by linear transformation of the input vector. ; The model was trained using historical office data from the past year, with Adam as the optimizer and the mean squared error (MSE) of the loss function as follows: ； In the formula, m Indicates the number of samples; It is the first j One actual value; It is the first j One predicted value.

5. The method for multi-scenario heterogeneous fusion and multi-system collaborative optimization of a large hydropower station air conditioning system according to claim 4, characterized in that, Step S4 specifically includes the following steps: Step S41: Set initial weights according to the air conditioning system's operating cycle and load characteristics: heating / cooling. ,in, Considering seasonal correction factors, the weight of meteorological features in the Transformer model is increased due to the high solar radiation intensity in summer, while the weight of human behavior features is reduced due to the low human activity level in winter. Step S42: Calculate the root mean square error of the sub-model using a 7-day sliding window. RMSE The weights are dynamically adjusted based on the magnitude of the error. ； If the underground equipment area load prediction model RMSE < 5%, then Increase by 0.1, Reduce by 0.1; if the office load forecasting model RMSE < 5%, then Increase by 0.1, Reduce by 0.1; In summary, the air conditioning load prediction model for large hydropower stations is expressed as follows: ； In the formula, The total air conditioning load is W; For the building load of the underground equipment area, W; For office building load, W.

6. The method for multi-scenario heterogeneous fusion and multi-system collaborative optimization of a large hydropower station air conditioning system according to claim 1, characterized in that, Step S5 specifically includes the following steps: Step S51: Through pipeline connectivity design, cross-system complementarity of cold source capacity of different air conditioning subsystems is achieved, a cold source capacity and energy consumption characteristic matrix is ​​constructed, and the maximum output capacity of each cold source device is determined. Current actual operating load Unit energy consumption Operating status: running / standby. The objective function is to minimize the total energy consumption of the air conditioning system and maximize the average efficiency of the equipment, with the following multiple objectives: ； In the formula, m The total number of cold source devices is [number] units. For the first j The output cooling capacity of the cooling source equipment meets the requirements. ; t For runtime, h ; For the first j Unit energy consumption of the cooling source equipment, KW / KW; For the first j The operating efficiency of the equipment; Define the constraints of the objective function, including terminal load demand constraints, cooling source capacity constraints, equipment start-up and shutdown constraints, and pipeline flow constraints. The sum of the output cooling capacity of all cooling source equipment must meet the total terminal load requirement. ； In the formula, The load margin factor reserves redundant cooling capacity; The output cooling capacity of each cold source device must not exceed its maximum capacity: ； The minimum start-up and shutdown time of the chiller unit is n Minimize the time spent starting and stopping the equipment to avoid frequent starts and stops that could cause wear and tear. like and ,but ; In the formula, For the first The last shutdown time of the chiller unit; The maximum flow rate in a cross-system pipeline must not exceed the pipeline's design capacity. ； In the formula, For the first k Flow rate of a cross-system pipeline ; Set its maximum design flow rate; Based on the above objective function and constraints, the Non-Dominated Sorting Genetic Algorithm (NSGA-II) is used to solve the multi-objective optimization problem to obtain the Pareto optimal solution. Real-number encoding is used, with each chromosome representing a set of output cooling capacity allocation schemes for the cold source equipment. Based on the objective function Calculate the fitness value of each chromosome, evaluate the quality of the solution using non-dominated sorting and crowding distance; finally output the optimal cold source allocation scheme. Step S52, based on the performance curves of different types of chillers: screw chiller units PLV The curves and COP curves of air-cooled heat pumps are compared to optimize the equipment's operating load rate, ensuring the equipment operates within its high-efficiency range and is adjusted based on feedback from end-user demand. Performance curves for different types of chillers were constructed, and the load rate of screw chillers was analyzed. and PLV The data were obtained by fitting a quadratic polynomial. PLV Performance curve equation: ； In the formula, For load factor, ; PLV Performance refers to the cooling capacity generated per kW of electrical energy consumed, expressed in KW / KW. a , b , c for PLV Performance curve equation coefficients; Obtain the outdoor temperature of the heat pump unit from the equipment manual. The load factor is of COP The data were obtained by fitting a bivariate quadratic polynomial. COP Performance curve equation: ； In the formula, a , b , c , d for COP Performance curve equation coefficients; With the optimization objective of maximizing the operating efficiency of the chiller unit, the objective function is: ； The constraints include: load factor constraints, end-point load demand constraints, and equipment performance curve constraints, based on... PLV The maximum range of the performance curve determines the load rate range of the screw chiller unit; based on COP The maximum range of the performance curve determines the load rate range of the heat pump unit; Based on the above objective function and constraints, mixed-integer linear programming (MILP) is used to solve the problem, considering the total load at the end. The optimal load factor allocation scheme is obtained by solving the problem. Step S53: Based on the predicted load obtained in step S4, adjust the parameters of the terminal equipment to meet the needs of the end users and establish a closed-loop feedback system. If the terminal adjustment cannot meet the needs, it will be fed back to step S52 to adjust the equipment load rate. After adjusting the equipment load rate, evaluate its efficiency loss. If the loss rate exceeds the threshold, it will be fed back to step S51 to adjust the cold source capacity allocation.

7. The method for multi-scenario heterogeneous fusion and multi-system collaborative optimization of a large hydropower station air conditioning system according to claim 6, characterized in that, The feedback from step S53 to step S51, adjusting the cold source capacity allocation, specifically includes the following steps: First, based on the current total terminal load... Total load forecast for the next 15 minutes The load change rate is calculated, and when it exceeds the threshold, the load is judged to be either rising or falling. For the parameter adjustment of combined air handling units, air volume adjustment takes precedence over supply air temperature adjustment. When the load increases, the air volume is increased first, provided that all constraints are met. If the load demand cannot be met, the supply air temperature is then reduced. For the parameter adjustment of fan coil units in office and control room areas, the human thermal comfort index PMV is calculated based on environmental and personnel parameters, and the fan coil unit speed and supply water temperature are adjusted according to the standard value to meet thermal comfort requirements. After the terminal equipment is adjusted, the temperature is verified and feedback is received through the monitoring devices of each air-conditioned area to perform rolling optimization.

8. The method for multi-scenario heterogeneous fusion and multi-system collaborative optimization of a large hydropower station air conditioning system according to claim 3, characterized in that, The unsteady thermal resistance-thermal capacity network model in S21 is constructed through the following steps: Step 21: Divide the rock wall into n uniform layers according to the rock wall thickness δ, with each layer having a thickness of... ,in, ; Step 22: Calculate the thermal resistance of each layer and heat capacity ;in, The thermal conductivity of the rock wall material. A For the area of ​​the rock wall, Density of the rock wall material c Specific heat capacity of the rock wall material; Step 23: Establish the unsteady-state heat transfer equation for each layer and set the boundary conditions: convective heat transfer between the inner layer and the indoor air, contact temperature between the outer layer and the soil, and initial conditions: the average temperature of the rock wall at the beginning of the year; Step 24: Solve the differential equations using the finite difference method to obtain the temperature change of each rock wall layer over time.

9. The method for multi-scenario heterogeneous fusion and multi-system collaborative optimization of a large hydropower station air conditioning system according to claim 3, characterized in that, The personnel activity status in S31 is calculated using the step count data from the smart work card. The specific steps are as follows: Step 31: The smart work card collects the step count data of the personnel in real time; Step 32: Determine the activity status based on the step count threshold: steps < 50 steps / 10 minutes is considered sitting still, so assign a value of 1; 50 ≤ steps < 200 steps / 10 minutes is considered light activity, so assign a value of 2; steps ≥ 200 steps / 10 minutes is considered walking, so assign a value of 3.

10. A multi-scenario heterogeneous fusion and multi-system collaborative optimization system for air conditioning systems in large hydropower stations, characterized in that, The system is used to implement the method according to any one of claims 1-9, comprising: Data acquisition module: used to collect heterogeneous data from multiple sources: temperature sensors, current transformers, and smart work cards; Load forecasting module: used to run a multi-scenario heteroproton model-run-cycle adaptive weighted fusion model to predict the total load at the end point; Multi-system collaborative optimization module: used to perform dynamic allocation of cross-system cold source capacity, efficient operation constrained by equipment performance curves, and adaptive adjustment of terminal parameters, including multi-objective optimization unit, equipment efficiency optimization unit, and closed-loop feedback unit; Execution module: Used to control the operating parameters of cold source equipment: screw chillers, heat pumps, and terminal equipment: combined air handling units, fan coil units.

11. The multi-scenario heterogeneous fusion and multi-system collaborative optimization system for air conditioning systems in large hydropower stations according to claim 10, characterized in that, The multi-objective optimization unit is the solution unit of the non-dominated sorting genetic algorithm; The equipment efficiency optimization unit is a solution unit for mixed integer linear programming; The closed-loop feedback unit is a sensor network and PLC control system.

Images