Water source heat pump air conditioning system emission reduction data evaluation method and system

Abstract

The present application relates to the technical field of data processing, and more particularly to a water source heat pump air conditioning system emission reduction data evaluation method and system, comprising data acquisition and processing, data completion, benchmark scenario modeling and simulation, actual emission calculation, emission reduction calculation and analysis, full life cycle emission calculation and result output; compared with the traditional full life cycle evaluation method, the regional factors such as specific resource conditions, environmental costs and policy intensity of the project location are ignored, resulting in that the evaluation results lack regional guidance significance; the present application introduces a regional influence weight adjustment model constructed based on a multilayer perception machine neural network, taking the differentiated parameters such as groundwater recharge cost, pollution control cost, recycling industry level and carbon policy intensity of the project location as input, and dynamically calculating the weight coefficients of each life cycle stage; the evaluation results can dynamically reflect where the value of implementing emission reduction is greater, and provide more accurate scientific basis for differentiated decision-making in different regions.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for evaluating emission reduction data of water source heat pump air conditioning systems. Background Technology

[0002] As a highly efficient building energy-saving technology, the carbon emission reduction achieved by water source heat pump air conditioning systems in actual operation is a core basis for measuring their environmental benefits, assessing project value, and supporting the formulation of relevant policies. However, the scientific assessment of this emission reduction is a complex and systematic project, which highly depends on high-quality historical and real-time operational data, a fair and reasonable benchmark, refined accounting methods, and consideration of regional external impacts.

[0003] Existing assessment methods have a number of significant shortcomings. When setting benchmarks, most methods employ static assumptions such as fixed efficiency coefficients or simple averages, failing to simulate the optimal operating conditions that conventional air conditioning systems might achieve under dynamic electricity prices, carbon prices, and meteorological conditions. This results in benchmarks that are either too lenient or too stringent, leading to unfair and unscientific emissions reduction assessments. In terms of calculation, existing methods often rely on long-term monthly or yearly totals for rough comparisons, failing to capture hourly fluctuations in emissions reductions within a single day, thus weakening the accuracy of the assessment results. More importantly, when assessing the environmental impact throughout the entire life cycle, they commonly use generic static databases and average parameters, completely ignoring the significant differences in water resource conditions, environmental governance costs, industrial chain maturity, and policy intensity at project locations. This renders the assessment conclusions detached from specific scenarios, making it difficult to support differentiated investment and decision-making.

[0004] This invention aims to solve the aforementioned problems and provides a method and system for evaluating emission reduction data of water source heat pump air conditioning systems. First, generative adversarial networks (GANs) are used to intelligently complete incomplete historical data with high fidelity, laying a reliable data foundation. Then, by constructing a building digital twin model and combining it with artificial intelligence optimization algorithms, the theoretically optimal operating state of a conventional air conditioning system during the evaluation period is simulated, thus establishing a dynamic and highly challenging benchmark scenario. Based on this, the actual hourly carbon emissions of the water source heat pump system are precisely compared with this dynamic benchmark to obtain the real-time emission reduction during the operating phase. Finally, a regional impact weight adjustment model based on neural networks is introduced to differentiate the weighting of carbon emissions at each stage of the entire life cycle, ensuring that the final result reflects the true environmental cost of a specific region. Ultimately, this achieves a full-chain evaluation and optimization from data, benchmarking, calculation to regional correction. Summary of the Invention

[0005] To overcome the problems mentioned in the background art, the present invention proposes a method and system for evaluating emission reduction data of water source heat pump air conditioning systems.

[0006] The technical solution of this invention is: a method for evaluating emission reduction data of a water source heat pump air conditioning system, comprising the following steps:

[0007] S11: Data acquisition and processing: Acquire building information data, historical operation data, meteorological environment data, and actual operation monitoring data of the water source heat pump air conditioning system of the target building, and clean and format the data.

[0008] S12: Data completion, which supplements incomplete data in the acquired data summary to form a complete building basic dataset;

[0009] S13: Baseline scenario modeling and simulation. Based on a complete building basic dataset, a digital twin energy consumption model of the target building under the scenario of using a conventional air conditioning system is constructed. Based on the digital twin energy consumption model and combined with grid carbon emission factor data, an artificial intelligence algorithm is used to simulate and calculate the hourly baseline energy consumption and corresponding hourly baseline carbon emissions of the conventional air conditioning system under the optimal operating strategy.

[0010] S14: Actual emissions calculation, based on the actual operation monitoring data of the water source heat pump air conditioning system, calculate its hourly actual carbon emissions during actual operation;

[0011] S15: Emission reduction calculation and analysis, compare the hourly baseline carbon emissions with the hourly actual carbon emissions, and calculate the dynamic emission reduction of the water source heat pump air conditioning system during the operation phase.

[0012] S16: Life cycle emissions calculation, calculates the carbon emissions of water source heat pump air conditioning system throughout its life cycle, and adjusts the weights based on regional differences to calculate the net emissions throughout the life cycle after regional correction.

[0013] S17: Results Output. Integrate the dynamic emission reduction during the operation phase with the net emission of the entire life cycle after regional correction, and generate and output a comprehensive emission reduction benefit assessment report for the water source heat pump air conditioning system.

[0014] As a preferred option, the data completion steps specifically include:

[0015] S21: Acquire historical energy consumption data for multiple complete operating cycles of the target building and similar buildings, and simultaneously collect corresponding meteorological data and building operation schedule data to form the original dataset;

[0016] S22: Clean, align, and normalize the original dataset to construct a training dataset for model training;

[0017] S23: Construct a data completion model based on generative adversarial networks, and use the training dataset to adversarially train the model until the discriminator can no longer stably distinguish between real data and generated data, thus obtaining a well-trained data completion model;

[0018] S24: Input the incomplete historical operation data of the target building, as well as the meteorological data and building operation schedule data for the corresponding time period, into the trained data completion model to generate complete time series data, replace and fill in the missing values ​​and outliers in the original data, and form a complete building basic dataset.

[0019] As a preferred approach, the data completion model is a conditional generative adversarial network. The generator of the data completion model takes into account a conditional vector derived from incomplete historical operational data, a random noise vector, and a weather and operational timetable vector corresponding to the time period. The discriminator of the data completion model takes into account real complete data samples, generated data samples output by the generator, conditional vectors, and weather and operational timetable vectors.

[0020] As a preferred embodiment, the baseline scenario modeling and simulation steps specifically include:

[0021] S31: Based on the complete building data set of the target building, construct and calibrate a digital twin energy consumption model that reflects the physical and energy consumption characteristics of the building, and replace the actual water source heat pump air conditioning system configuration with a preset virtual configuration of conventional air conditioning system;

[0022] S32: Acquire hourly grid carbon emission factor data corresponding to the assessment period and associate it with the energy consumption-carbon emission conversion module of the digital twin energy consumption model for dynamic boundary conditions;

[0023] S33: With minimizing operating costs as the optimization objective, artificial intelligence optimization algorithms are used to simulate and solve the optimal operation control strategy of the virtual configuration of the conventional air conditioning system throughout the entire evaluation period in the digital twin energy consumption model.

[0024] S34: Based on the optimal operation control strategy, run the digital twin energy consumption model to obtain the hourly simulated energy consumption value of the virtual configuration of the conventional air conditioning system, and combine it with the hourly grid carbon emission factor data to calculate the hourly baseline carbon emissions under the optimal strategy.

[0025] Preferably, the artificial intelligence optimization algorithm used in the baseline scenario modeling and simulation steps is a deep reinforcement learning algorithm, the specific execution process of which includes:

[0026] S41: State definition, which defines the building indoor and outdoor thermal environment parameters, system operating status, and current and predicted hourly grid carbon emission factors of the digital twin energy consumption model at the simulation time as the state variables of reinforcement learning;

[0027] S42: Action definition, defining the adjustable control variables of the virtual configuration of the conventional air conditioning system, including setpoint temperature, equipment start / stop and frequency conversion, as action quantities for reinforcement learning;

[0028] S43: Reward function definition. Set a reward function to reward low-energy consumption and low-carbon emission operating behaviors while meeting indoor thermal comfort constraints.

[0029] S44: Policy learning, in a simulated environment composed of a digital twin energy consumption model, trains a deep reinforcement learning algorithm to learn to output the optimal operation control strategy that satisfies the optimization objective.

[0030] As a preferred option, the actual emissions calculation steps include:

[0031] S51: Real-time monitoring and acquisition of hourly operating data of the water source heat pump air conditioning system during the evaluation period. The hourly operating data includes the total power consumption of the system, the power consumption of each major electrical equipment, and the consumption of other types of energy when the system includes non-electrically driven equipment.

[0032] S52: Based on hourly operating data, calculate the actual total energy consumption of the water source heat pump air conditioning system at each moment during the evaluation period;

[0033] S53: Obtain the hourly grid carbon emission factor corresponding to the assessment period, and calculate the carbon emissions of the electrical energy consumed by the water source heat pump system based on the hourly grid carbon emission factor;

[0034] S54: Cross-compare the calculated hourly actual carbon emissions with historical data for the same period, mark, verify and correct any abnormal data that exceeds the reasonable range, and form the final hourly actual carbon emission sequence.

[0035] As a preferred option, the life-cycle emissions calculation steps specifically include:

[0036] S61: Stage definition and initial calculation, define the entire life cycle stages of the water source heat pump air conditioning system, and use the life cycle assessment method to calculate the initial carbon emissions of each stage;

[0037] S62: Obtaining regional difference parameters, obtaining regional difference parameters of the project location;

[0038] S63: Dynamic weight coefficient calculation. Input the regional difference parameter into the preset regional influence weight adjustment model, and calculate and output the dynamic influence weight coefficient corresponding to each life cycle stage.

[0039] S64: Regionally corrected emissions calculation: Multiply the initial carbon emissions for each stage by the corresponding dynamic impact weight coefficients for each stage to obtain the regionally corrected carbon emissions for each stage, and then sum them to obtain the regionally corrected net emissions of the water source heat pump air conditioning system throughout its entire life cycle.

[0040] As a preferred option, the specific regional characteristic parameters include:

[0041] A11: Environmental resource cost parameters, including the unit cost of groundwater recharge reflecting the impact of groundwater on the environment and the unit treatment cost of major pollution types reflecting the local environmental sensitivity;

[0042] A12: Industry technical parameters, including the recycling rate of equipment and materials reflecting the level of resource recycling and the local recycling industry maturity index;

[0043] A13: Demand parameters, including local carbon emission intensity constraint coefficients.

[0044] Preferably, the regional influence weight adjustment model is an artificial neural network model based on a multilayer perceptron. This model is used to dynamically calculate the dynamic influence weight coefficients for each stage of the water source heat pump air conditioning system's lifecycle based on the regional differences in the project location. The construction of the regional influence weight adjustment model includes:

[0045] S71: Construct a multilayer perceptron network structure that includes an input layer, at least one hidden layer, and an output layer;

[0046] S72: Construct a training sample set using historical project data and expert knowledge. Each sample in the training sample set includes a set of normalized regional difference parameters as input features and a set of baseline weight coefficients corresponding to each life cycle stage determined by an optimization algorithm as labels.

[0047] S73: Supervised training of the multilayer perceptron model is performed using the training sample set until its loss function converges, resulting in a trained weight adjustment model.

[0048] The emission reduction data evaluation system for water source heat pump air conditioning systems includes:

[0049] The data processing module is used to acquire, clean, and complete data to form a basic building dataset;

[0050] The baseline scenario simulation module, connected to the data processing module, is used to build a digital twin model of a conventional air conditioning system and simulate its hourly baseline carbon emissions under optimal operation using AI.

[0051] The actual emissions calculation module is used to calculate the actual hourly carbon emissions of the water source heat pump system;

[0052] The comprehensive analysis module is connected to the baseline scenario simulation module and the actual emission calculation module, respectively. It is used to calculate the dynamic emission reduction during the operation phase, and to perform full life cycle carbon emission accounting and dynamic weight adjustment based on regional differences, and calculate the regionally corrected net emissions for the entire life cycle.

[0053] The report generation module, connected to the comprehensive analysis module, is used to integrate the results and generate a comprehensive emission reduction benefit assessment report.

[0054] The beneficial effects of this invention are:

[0055] 1. Compared to traditional life cycle assessment methods that typically use general, static databases for calculation, neglecting regional factors such as specific resource conditions, environmental costs, and policy intensity at the project site, resulting in assessment results lacking regional guidance significance, this invention introduces a regional impact weight adjustment model based on a multilayer perceptron neural network in life cycle assessment. This model uses differentiated parameters such as groundwater recharge costs, pollution control costs, recycling industry levels, and carbon policy intensity at the project site as inputs to dynamically calculate the weight coefficients for each life cycle stage. This method enables the assessment results to dynamically reflect where emission reduction is more valuable, providing a more accurate scientific basis for differentiated decision-making in different regions.

[0056] 2. Compared to existing technologies that typically use fixed efficiency coefficients or static operating assumptions to set baselines, this method cannot simulate the dynamic response of conventional air conditioning systems under real-time electricity prices, carbon prices, and meteorological conditions, potentially leading to unscientific comparison benchmarks. This invention constructs a digital twin model of the target building and utilizes artificial intelligence algorithms such as deep reinforcement learning to simulate the baseline carbon emissions of conventional air conditioning systems under the "theoretically optimal operating strategy." This method establishes a scientific, reasonable, and highly challenging dynamic comparison benchmark, ensuring that the emission reduction benefits of water source heat pumps are derived in comparison with the "best performing" conventional system, making the evaluation results more credible.

[0057] 3. Compared with existing assessment methods that are mostly based on monthly or annual totals for macro-level estimation, which cannot capture the subtle fluctuations caused by changes in load and grid carbon intensity within a day, and may thus mask the true benefits in actual operation; this invention compares the hourly baseline carbon emissions with the actual carbon emissions when calculating emission reductions during the operation phase; this method achieves high temporal resolution for accurate measurement of emission reduction benefits, and can clearly reveal the specific emission reduction performance of the water source heat pump system under different time periods and operating conditions, providing data support for refined operation optimization;

[0058] 4. Compared to existing technologies that generally use traditional methods such as linear interpolation or statistical regression to process missing data, these methods struggle to capture the complex nonlinear relationships between energy consumption and weather and operational patterns, leading to discrepancies between the completed data and the actual situation. This invention innovatively employs an intelligent completion model based on conditional generative adversarial networks. This model learns the high-dimensional distribution of real data through adversarial training, enabling it to generate high-quality continuous time series data that conforms to the physical laws of building energy consumption and is context-dependent. This method effectively solves the modeling bottleneck caused by data quality issues, laying a solid data foundation for the subsequent construction of accurate digital twin models. Attached Figure Description

[0059] Figure 1 The diagram shown is a flowchart illustrating the emission reduction data evaluation method for the water source heat pump air conditioning system of the present invention.

[0060] Figure 2 The diagram shown is a structural schematic of the emission reduction data evaluation system for the water source heat pump air conditioning system of the present invention. Detailed Implementation

[0061] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0062] Please see Figure 1 This invention provides an embodiment: a method for evaluating emission reduction data of a water source heat pump air conditioning system, comprising the following steps:

[0063] Step 1: Data Acquisition and Processing

[0064] Acquire building information data, historical operation data, meteorological environment data, and actual operation monitoring data of the water source heat pump air conditioning system of the target building, and clean and format the data.

[0065] Specifically, various data on the target building are collected from building energy management systems, sensor networks, design drawings, and meteorological databases, including: static building information data, such as building area, thermal parameters of the building envelope, and functional zoning; dynamic historical operating data, such as hourly heating and cooling loads and air conditioning system energy consumption over the past 1-3 years; meteorological environmental data, such as hourly dry-bulb temperature, wet-bulb temperature, and solar radiation intensity during the assessment period; and actual operation monitoring data of the water source heat pump air conditioning system, including hourly power consumption, flow rate, and inlet and outlet water temperatures of key equipment such as the main unit, water pumps, and cooling towers.

[0066] Step 2: Data Completion

[0067] Because historical operational data often contains missing, anomalies, or noise, an intelligent completion method based on generative adversarial networks is adopted to supplement the incomplete data in the acquired data summary, forming a complete building foundation dataset, specifically including:

[0068] S21: Acquire historical energy consumption data for multiple complete operating cycles of the target building and similar buildings, and simultaneously collect corresponding meteorological data and building operation schedule data to form the original dataset;

[0069] S22: Clean, align, and normalize the original dataset to construct a training dataset for model training;

[0070] S23: Construct a data completion model based on generative adversarial networks, and use the training dataset to adversarially train the model until the discriminator can no longer stably distinguish between real data and generated data, thus obtaining a well-trained data completion model;

[0071] S24: Input the incomplete historical operation data of the target building, as well as the meteorological data and building operation schedule data for the corresponding time period, into the trained data completion model to generate complete time series data, replace and fill in the missing values ​​and outliers in the original data, and form a complete building basic dataset.

[0072] The historical energy consumption data includes hourly electricity consumption, hourly cooling load and hourly heating load; the meteorological data includes hourly dry-bulb temperature, wet-bulb temperature and solar irradiance; and the building operation schedule data includes hourly personnel density and equipment activation status.

[0073] The cleaning, alignment, and normalization processes for the original dataset specifically include:

[0074] Data cleaning: Removing outliers from the original dataset that are clearly outside the physically reasonable range;

[0075] Data alignment: Historical energy consumption data, meteorological data, and building operation schedule data are aligned in time sequence according to a unified timestamp;

[0076] Data normalization: Mapping all aligned data to a uniform numerical range to complete data standardization.

[0077] Specifically, the data completion model is a conditional generative adversarial network. The generator of the data completion model takes into account a conditional vector derived from incomplete historical operational data, a random noise vector, and a weather and operational timetable vector corresponding to the time period. The discriminator of the data completion model takes into account real complete data samples, generated data samples output by the generator, conditional vectors, and weather and operational timetable vectors.

[0078] The adversarial training process of the data completion model includes:

[0079] Train the discriminant with fixed generator parameters: Label the real and complete data in the training dataset as true, and label the data generated by the generator based on incomplete conditions as false, and train the discriminant to accurately distinguish between the two.

[0080] Training the generator with fixed discriminator parameters: The goal is to optimize the generator's parameters so that the generated data deceives the discriminator and makes it classify the data as true.

[0081] The discriminator and generator are trained in a loop until the preset number of training rounds is reached, or the discriminator's judgment accuracy approaches the level of random guessing.

[0082] Among them, the building basic dataset is used to construct a digital twin model of the energy consumption of the target building for subsequent baseline simulation calculations of emission reduction.

[0083] In this embodiment, to address the issue of incomplete historical energy consumption data for buildings, complete historical energy consumption, meteorological, and operational schedule data for the target building and similar buildings are first collected to form a training set. After data cleaning, time-series alignment, and normalization preprocessing, a conditional generative adversarial network (GAN) is constructed as a data completion model. The generator takes incomplete data, meteorological, and operational schedule data as input conditions and learns the distribution patterns of real data through adversarial training. After training, the model is used to complete the missing or abnormal historical operational data of the target building, generating high-quality, continuous time-series data that conforms to the physical laws of building energy consumption, forming a complete building basic dataset. Compared with traditional interpolation or statistical imputation methods, this method, through the powerful generative capabilities of GANs, can capture the complex nonlinear relationships between energy consumption, meteorology, and operational patterns, thereby performing high-fidelity, context-aware completion of missing data. This effectively solves the modeling bottleneck caused by data quality issues and lays a solid data foundation for the subsequent construction of reliable digital twin models and accurate emission reduction benchmark calculations.

[0084] Step 3: Baseline Scenario Modeling and Simulation

[0085] Based on a complete building dataset, a digital twin energy consumption model of the target building under the scenario of using a conventional air conditioning system is constructed. Based on this digital twin energy consumption model and combined with grid carbon emission factor data, artificial intelligence algorithms are used to simulate and calculate the hourly baseline energy consumption and corresponding hourly baseline carbon emissions of the conventional air conditioning system under the optimal operating strategy. Specifically, this includes:

[0086] S31: Based on the complete building dataset of the target building, construct and calibrate a digital twin energy consumption model reflecting the building's physical and energy consumption characteristics, and replace the actual water source heat pump air conditioning system configuration with a preset virtual configuration of a conventional air conditioning system, specifically including:

[0087] Model building: Using a complete building basic dataset, a three-dimensional geometric model of the target building is built in the simulation platform, the thermal parameters of the building envelope and the internal thermal disturbance timetable are set, and the equipment performance curves, capacity and control logic of the conventional air conditioning system are configured in a virtual configuration.

[0088] Model calibration: The known historical meteorological data and operation schedule data in the complete building basic dataset are used as input to drive the digital twin energy consumption model. The energy consumption simulation data output by the model is compared with the real historical energy consumption data in the same time period. By adjusting the key model parameters, the error between the two is made lower than the preset accuracy threshold, thus completing the model calibration.

[0089] S32: Acquire hourly grid carbon emission factor data corresponding to the assessment period and associate it with the energy consumption-carbon emission conversion module of the digital twin energy consumption model for dynamic boundary conditions;

[0090] S33: With minimizing operating costs as the optimization objective, artificial intelligence optimization algorithms are used to simulate and solve the optimal operation control strategy of the virtual configuration of the conventional air conditioning system throughout the entire evaluation period in the digital twin energy consumption model.

[0091] S34: Based on the optimal operation control strategy, run the digital twin energy consumption model to obtain the hourly simulated energy consumption value of the virtual configuration of the conventional air conditioning system, and combine it with the hourly grid carbon emission factor data to calculate the hourly baseline carbon emissions under the optimal strategy. The principle formula is as follows:

[0092] ;

[0093] in, This represents the hourly baseline carbon emissions at time t. The total power consumption of the virtual configuration of the conventional air conditioning system at time t is obtained by simulation using the digital twin energy consumption model. Let be the carbon emission factor of the power grid at time t.

[0094] Among them, the conventional air conditioning system is virtually configured as an air-cooled heat pump system, or a composite system consisting of a multi-split system and a gas boiler or electric boiler.

[0095] In this embodiment, the artificial intelligence optimization algorithm used is a deep reinforcement learning algorithm, and its specific execution process includes:

[0096] S41: State definition, which defines the building indoor and outdoor thermal environment parameters, system operating status, and current and predicted hourly grid carbon emission factors of the digital twin energy consumption model at the simulation time as the state variables of reinforcement learning;

[0097] S42: Action definition, defining the adjustable control variables of the virtual configuration of the conventional air conditioning system, including setpoint temperature, equipment start / stop and frequency conversion, as action quantities for reinforcement learning;

[0098] S43: Reward function definition. Set a reward function to reward low-energy consumption and low-carbon emission operating behaviors while meeting indoor thermal comfort constraints.

[0099] S44: Policy learning, in a simulated environment composed of a digital twin energy consumption model, trains a deep reinforcement learning algorithm to learn to output the optimal operation control strategy that satisfies the optimization objective.

[0100] In this embodiment, based on a complete building dataset, a high-fidelity energy consumption model of the target building is constructed and calibrated in a digital twin platform, and the air conditioning system configuration is replaced with a preset conventional air conditioning system. Dynamic hourly grid carbon emission factors are input into the model as key boundary conditions. Using artificial intelligence optimization algorithms such as deep reinforcement learning, and under the constraint of satisfying indoor thermal comfort, the globally optimal operating control strategy of the conventional air conditioning system throughout the entire evaluation period is simulated and optimized in the digital twin environment with the goal of minimizing operating costs. Finally, hourly energy consumption is simulated based on this strategy, and the hourly benchmark carbon emission sequence is calculated by combining it with the grid carbon factor. Its beneficial effect is that, compared to traditional baseline methods using fixed efficiency coefficients or static operating assumptions, this method, through digital twins and AI optimization, simulates the "theoretically optimal" performance of the conventional air conditioning system in response to real-time grid carbon prices, thereby establishing a dynamic, reasonable, and highly challenging emission reduction benchmark. This not only avoids the problem of artificially exaggerating the emission reduction benefits of water source heat pumps due to poor operation of the comparison objects, but also ensures that the evaluation results are more scientific, fair and credible, providing a reliable basis for accurately quantifying the emission reduction during the operation phase.

[0101] Step 4: Actual Emissions Calculation

[0102] Based on actual operation monitoring data of the water source heat pump air conditioning system, the hourly actual carbon emissions during actual operation are calculated, specifically including:

[0103] S51: Real-time monitoring and acquisition of hourly operating data of the water source heat pump air conditioning system during the evaluation period. The hourly operating data includes the total power consumption of the system, the power consumption of each major electrical equipment, and the energy consumption of other types of energy when the system includes non-electrically driven equipment. The hourly operating data is collected and recorded in real time by smart meters, flow meters, temperature sensors and energy meters deployed on the main unit, water pump, cooling tower and terminal circulation loop of the water source heat pump air conditioning system.

[0104] S52: Based on hourly operating data, calculate the actual total energy consumption of the water source heat pump air conditioning system at each moment during the evaluation period;

[0105] S53: Obtain the hourly grid carbon emission factor corresponding to the assessment period, and calculate the carbon emissions of the electrical energy consumed by the water source heat pump system based on the hourly grid carbon emission factor;

[0106] S54: Cross-compare the calculated hourly actual carbon emissions with historical data for the same period, mark, verify and correct any abnormal data that exceeds the reasonable range, and form the final hourly actual carbon emission sequence.

[0107] The specific formula for calculating the hourly actual carbon emissions of a water source heat pump air conditioning system in hour t is as follows:

[0108] ;

[0109] in, The actual hourly carbon emissions of the water source heat pump air conditioning system in hour t. Let t be the total power consumption of the system in hour t. Let be the carbon emission factor of the power grid in hour t.

[0110] Specifically, when cross-referencing the calculated hourly actual carbon emissions with historical data from the same period, and marking, verifying, and correcting abnormal data that exceeds a reasonable range, the process includes:

[0111] Threshold verification: Set reasonable upper and lower limits for hourly carbon emissions per unit building area, and mark data points that exceed these thresholds;

[0112] Trend verification: Check the continuity and smoothness of the hourly actual carbon emission series and mark abnormal jump points such as sudden increases and decreases;

[0113] Verification and correction: For marked abnormal data points, trace back the original operating data and equipment logs. If it is confirmed to be a data acquisition error or a transmission error, then use interpolation of data from adjacent time points for estimation and correction.

[0114] In this embodiment, intelligent metering devices deployed at key nodes of the water source heat pump system monitor and collect hourly energy consumption data in real time. Based on this actual operating data, combined with dynamic hourly grid carbon emission factors, the actual carbon emissions at each moment of system operation are accurately calculated. To ensure data quality, the calculation results are cross-referenced with historical data, thresholds are set, trend analysis is performed, and abnormal data points are traced, verified, and interpolated for correction, ultimately forming an accurate and reliable hourly actual carbon emission sequence. Its advantages are: this method abandons the traditional approach of using theoretical efficiency or average values ​​for rough estimation; through real-time monitoring and dynamic factor calculation, it achieves accurate and high-frequency metering of carbon emissions from the water source heat pump system; the strict data verification and correction mechanism effectively eliminates data noise caused by metering errors or equipment malfunctions, thus providing a solid, high-quality data foundation for subsequent fair and credible comparative analysis with the "baseline scenario."

[0115] Step 5: Emission Reduction Calculation and Analysis

[0116] By comparing the hourly baseline carbon emissions with the hourly actual carbon emissions, the dynamic emission reduction of the water source heat pump air conditioning system during operation is calculated. Specifically, by comparing and calculating the hourly baseline carbon emissions, which represent the optimal conventional air conditioning scheme, with the hourly actual carbon emissions calculated based on actual monitoring data, point by point on the same evaluation time scale, the real-time emission reduction of the water source heat pump system at each moment is obtained. These are then accumulated over the entire evaluation period to calculate the total dynamic emission reduction achieved by the water source heat pump system relative to the optimal conventional scenario during operation. This accurately and intuitively reflects the carbon emission reduction contribution of the technology in daily operation.

[0117] Step Six: Life Cycle Emissions Calculation

[0118] A full life-cycle carbon emission accounting was performed on the water source heat pump air conditioning system, and regional differences were considered for weighting adjustments to calculate the regionally corrected net life-cycle emissions, which specifically include:

[0119] S61: Stage definition and initial accounting, define the entire life cycle stages of the water source heat pump air conditioning system, and use the life cycle assessment method to calculate the initial carbon emissions of each stage;

[0120] S62: Obtaining regional difference parameters, obtaining regional difference parameters of the project location;

[0121] S63: Dynamic weight coefficient calculation. Input the regional difference parameter into the preset regional influence weight adjustment model, and calculate and output the dynamic influence weight coefficient corresponding to each life cycle stage.

[0122] S64: Regionally corrected emissions calculation: Multiply the initial carbon emissions for each stage by the corresponding dynamic impact weight coefficients for each stage to obtain the regionally corrected carbon emissions for each stage, and then sum them to obtain the regionally corrected net emissions of the water source heat pump air conditioning system throughout its entire life cycle.

[0123] The entire lifecycle stage specifically includes:

[0124] Production stage: encompasses the manufacturing process of all equipment and materials, including main unit, water pumps, pipelines, and materials;

[0125] Transportation phase: This covers the entire process of transporting equipment and materials from the production site to the project site;

[0126] Installation phase: covers energy and material consumption during on-site system installation and commissioning;

[0127] Operational phase: This phase covers the system's energy consumption over its entire service life, and the data is derived from actual operational monitoring and calculations.

[0128] Maintenance phase: This includes material and transportation costs incurred during regular maintenance and parts replacement;

[0129] The end-of-life recycling phase covers the dismantling, waste disposal, and material recycling or disposal processes when the system is scrapped.

[0130] Specifically, the regional characteristic parameters include:

[0131] A11: Environmental resource cost parameters, including the unit cost of groundwater recharge reflecting the impact of groundwater on the environment and the unit treatment cost of major pollution types reflecting the local environmental sensitivity;

[0132] A12: Industry technical parameters, including the recycling rate of equipment and materials reflecting the level of resource recycling and the local recycling industry maturity index;

[0133] A13: Demand parameters, including local carbon emission intensity constraint coefficients.

[0134] In this embodiment, the regional influence weight adjustment model is an artificial neural network model based on a multilayer perceptron. It is used to dynamically calculate the dynamic influence weight coefficients for each lifecycle stage of the water source heat pump air conditioning system based on the regional difference parameters of the project location. The construction of the regional influence weight adjustment model includes:

[0135] S71: Construct a multilayer perceptron network structure that includes an input layer, at least one hidden layer, and an output layer;

[0136] S72: Construct a training sample set using historical project data and expert knowledge. Each sample in the training sample set includes a set of normalized regional difference parameters as input features and a set of baseline weight coefficients corresponding to each life cycle stage determined by an optimization algorithm as labels.

[0137] S73: Supervised training of the multilayer perceptron model is performed using the training sample set until its loss function converges, resulting in a trained weight adjustment model.

[0138] The principle and process of the regional influence weight adjustment model in calculating weight parameters are as follows:

[0139] When evaluating a specific project, the regional differences in the target project location are normalized to form an input vector. This input vector is then fed into a trained weight adjustment model. The input layer of the weight adjustment model receives this vector, the hidden layer performs nonlinear transformation and feature fusion on it, and the output layer outputs a set of values ​​through a normalization function. This set of values ​​serves as the dynamic influence weight coefficients corresponding to each stage of the equipment's life cycle: production, transportation, installation, operation, maintenance, and disposal. The sum of all coefficients is 1.

[0140] The output layer of the regional influence weight adjustment model uses the Softmax function as the normalization function to ensure that all output weight coefficients are non-negative and sum to 1.

[0141] In this embodiment, firstly, a life cycle assessment (LCA) approach is used to perform carbon accounting on the entire process of a water source heat pump air conditioning system, from equipment production, transportation, installation, operation, maintenance to end-of-life recycling, to obtain the initial carbon emissions at each stage. Simultaneously, a regional impact weight adjustment model based on a multilayer perceptron neural network is introduced. This model takes differentiated parameters such as groundwater recharge costs, pollution control costs, recycling industry levels, and carbon policy intensity in the project location as inputs, and calculates and outputs a set of dynamic impact weight coefficients. Finally, the initial carbon emissions at each stage are multiplied by their corresponding weights and summed to obtain the regionally adjusted net emissions over the entire life cycle. This method overcomes the limitations of traditional LCA, which only uses static and general data for accounting. By quantifying the impact of regional resource, environmental, economic, and policy externalities on the system's true environmental costs and applying differentiated weights to each stage, the assessment results dynamically reflect where emission reduction value is greater. This provides a more accurate, fairer, and more practically guiding scientific basis for differentiated investment decisions, technology selection, and policy formulation in different regions.

[0142] Step 7: Output Results

[0143] By integrating dynamic emission reductions during the operational phase with regionally adjusted net emissions over the entire lifecycle, a comprehensive emission reduction benefit assessment report for the water source heat pump air conditioning system is generated and output, specifically including:

[0144] S81: Receive and integrate dynamic emission reductions during the operation phase and regionally corrected net emissions over the entire life cycle, and calculate the net emission reduction over the entire life cycle of the water source heat pump system relative to the baseline scenario.

[0145] S82: The integrated and calculated results data, together with the system's key operating parameters and regional adjustment factor values, are organized according to a preset report template to automatically generate a comprehensive emission reduction benefit assessment report that includes text, data tables, and trend charts.

[0146] S83: Output the generated comprehensive emission reduction benefit assessment report in the specified format and provide a visualization interface.

[0147] The calculation method for the net emission reduction of the water source heat pump system relative to the baseline scenario is as follows: subtract the regionally corrected net emission of the water source heat pump system from the total carbon emissions of the baseline conventional air conditioning system throughout its entire life cycle, and then add the dynamic emission reduction during the operation phase.

[0148] The comprehensive emission reduction benefit assessment report includes the following parts: system and project overview, assessment methods and data source description, hourly and daily emission reduction analysis charts during the operation phase, analysis of carbon emission contribution and regional weight impact at each stage of the entire life cycle, summary of key emission reduction benefit indicators, uncertainty analysis and main conclusions.

[0149] In this embodiment, the direct emission reductions during the operational phase and the regionally adjusted net emissions over the entire life cycle are integrated. The final net emission reduction relative to the baseline scenario is calculated by subtracting the regionally adjusted net emissions of this system from the total life cycle emissions of the baseline scenario and adding the operational emission reductions. This core result, along with key operational parameters and regional adjustment factors, is then automatically organized according to a preset template to generate a structured comprehensive assessment report. This report includes a system overview, methodology description, hourly emission reduction trend charts, life cycle stage and regional impact analysis, a summary of core indicators, and conclusions. A visual interface is provided for result display and interaction. The beneficial effects are: this method not only achieves a crucial integration from data to insights at the technical and environmental levels, but also transforms complex carbon emission accounting and dynamic assessment results into intuitive, comprehensive, and actionable decision support documents through automated report generation and visual interaction. This greatly improves the usability and communication efficiency of the assessment results, providing authoritative, transparent, and persuasive final evidence for project certification, performance demonstration, and policy formulation.

[0150] like Figure 2As shown, this embodiment also provides an emission reduction data evaluation system for water source heat pump air conditioning systems, including:

[0151] The data processing module is used to acquire, clean, and complete data to form a basic building dataset;

[0152] The baseline scenario simulation module, connected to the data processing module, is used to build a digital twin model of a conventional air conditioning system and simulate its hourly baseline carbon emissions under optimal operation using AI.

[0153] The actual emissions calculation module is used to calculate the actual hourly carbon emissions of the water source heat pump system;

[0154] The comprehensive analysis module is connected to the baseline scenario simulation module and the actual emission calculation module, respectively. It is used to calculate the dynamic emission reduction during the operation phase, and to perform full life cycle carbon emission accounting and dynamic weight adjustment based on regional differences, and calculate the regionally corrected net emissions for the entire life cycle.

[0155] The report generation module, connected to the comprehensive analysis module, is used to integrate the results and generate a comprehensive emission reduction benefit assessment report.

[0156] Example 1: Emission Reduction Assessment of Water Source Heat Pump System in Commercial Buildings in a Hot-Summer, Cold-Winter Region

[0157] This embodiment focuses on an office building in Shanghai with a floor area of ​​20,000 square meters, whose air conditioning system has been converted to a groundwater-source heat pump system. The evaluation period is a complete cooling season. First, historical hourly data for the cooling season was collected from the building's energy management system, sensor network, and meteorological database. This included building envelope parameters, indoor occupant and equipment schedules, outdoor dry and wet bulb temperatures, and energy consumption and inlet / outlet water temperatures of the system's main unit and pumps. Due to a sensor malfunction, a significant amount of historical cooling load data was missing for two weeks in July. To address this, the evaluation system used complete annual operating data from three similar office buildings in the same area as a training set. Through a pre-trained conditional generative adversarial network model, using the building's existing meteorological data and operating schedule as conditions, a continuous, high-quality cooling load sequence consistent with physical laws was generated, improving data integrity to 99.5% and forming a reliable building-based dataset.

[0158] Based on this complete dataset, a high-fidelity energy consumption model of the building was established on a digital twin platform. In the baseline scenario, the air conditioning system was virtually configured as an efficient air-cooled heat pump system. After model calibration, actual hourly meteorological data for the evaluation period was input, and a deep reinforcement learning algorithm was used for optimization. The algorithm, aiming to minimize operating electricity and carbon costs, dynamically optimized the start-up and shutdown of the air-cooled heat pump and the temperature setpoint while meeting indoor temperature constraints, simulating its theoretically optimal operating strategy. Ultimately, the total baseline carbon emissions of the virtual system during the evaluation period were calculated to be 285 tons of CO2 equivalent. Simultaneously, the actual total electricity consumption of the water source heat pump system was obtained through on-site smart meter monitoring. Combined with the hourly carbon emission factor of the Shanghai power grid, its actual operating carbon emissions were calculated to be 210 tons. Through hourly comparison, it was found that the system achieved an immediate dynamic emission reduction of 75 tons during the cooling season through efficient operation.

[0159] A further full life-cycle assessment was conducted. The initial carbon emissions of the water source heat pump system (including the main unit, pumps, and pipelines) were calculated for each stage from production, transportation, installation, operation, maintenance to disposal and recycling. Simultaneously, regional characteristic parameters of the Shanghai area were collected, such as the reinjection costs due to high groundwater protection requirements, the mature scrap metal recycling industry, and strict carbon emission intensity control policies. These parameters were input into a trained regional impact weighting model, which output a set of dynamic weights, significantly increasing the weight proportion of the operation and maintenance phases in the full life-cycle impact. After weighted calculation, the system's regionally adjusted net emissions over its entire life-cycle were determined to be 1000 tons. In comparison, the baseline air-cooled heat pump system, after the same regional weighting adjustment, had a total life-cycle emission of 1500 tons. Overall, the water source heat pump project not only achieved an immediate emission reduction of 75 tons during the operation phase, but also a net emission reduction of 575 tons over its entire life-cycle. Finally, all process data, comparative charts, and conclusions were automatically integrated to generate a comprehensive assessment report including hourly emission reduction trend charts, full life-cycle stage contribution analysis, and key benefit indicators.

[0160] Example 2: Emission Reduction Assessment of Water Source Heat Pump System in a Hotel Building in a Cold Region

[0161] This embodiment focuses on a hotel building in Shenyang City with a floor area of ​​15,000 square meters, which uses a groundwater-source heat pump system to meet its year-round heating and cooling needs. The evaluation period is a full year. During the data acquisition phase, hourly load, energy consumption, and meteorological data for the entire year were collected. Some winter heat load data was abnormal due to communication interruptions. A generative adversarial network method similar to that in Embodiment 1 was used, with concurrent meteorological data and hotel occupancy schedules as conditions, to successfully complete the missing heat load data. The complete annual dataset provided reliable input for subsequent simulations.

[0162] In the baseline scenario modeling, a combined conventional air conditioning system consisting of a multi-split system and a gas-fired boiler was configured for the hotel's digital twin model. Optimized by a deep reinforcement learning algorithm, the system dynamically adjusts the input ratio of the multi-split heating system and the gas-fired boiler based on temperature, gas price, and electricity price during winter, simulating the lowest-cost operating strategy. Combining the hourly carbon factor of the Northeast regional power grid and the carbon intensity coefficient of natural gas, the theoretically optimal annual carbon emissions for this baseline system were calculated to be 420 tons. The actual operating energy consumption of the water source heat pump system was obtained through sub-metering devices, and its annual electricity consumption corresponds to actual carbon emissions of 300 tons. Through hourly comparison, its annual dynamic emission reduction was determined to be 120 tons.

[0163] In the full life cycle accounting phase, in addition to calculating the initial emissions of each stage of the system, the regional differences in Shenyang were taken into account. For example, the long winters in the area result in a large proportion of energy consumption during operation, but the recycling industry chain for scrapped equipment is relatively weak. The regional impact weight adjustment model, based on the input parameters, provides weight coefficients different from those in Implementation Example 1, appropriately increasing the weights of the operation and scrapping / recycling stages. After this correction, the net emissions of the water source heat pump system throughout its entire life cycle are 1250 tons, while the baseline combined air conditioning system is 1800 tons. Considering the dynamic emission reduction of 120 tons during the operation phase, the total net emission reduction of this project compared to the baseline scenario reaches 670 tons throughout its entire life cycle. The report generation module integrates and visualizes the above calculation process, regional weight impact analysis, and core emission reduction benefit indicators, automatically generating a detailed assessment report for the hotel project, clearly demonstrating the significant emission reduction benefits of water source heat pump technology in cold regions.

[0164] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A method for evaluating emission reduction data of a water source heat pump air conditioning system, characterized in that: Includes the following steps: S11: Data acquisition and processing: Acquire building information data, historical operation data, meteorological environment data, and actual operation monitoring data of the water source heat pump air conditioning system of the target building, and clean and format the data. S12: Data completion, which supplements incomplete data in the acquired data summary to form a complete building basic dataset; S13: Baseline scenario modeling and simulation. Based on a complete building basic dataset, a digital twin energy consumption model of the target building under the scenario of using a conventional air conditioning system is constructed. Based on the digital twin energy consumption model and combined with grid carbon emission factor data, an artificial intelligence algorithm is used to simulate and calculate the hourly baseline energy consumption and corresponding hourly baseline carbon emissions of the conventional air conditioning system under the optimal operating strategy. S14: Actual emissions calculation, based on the actual operation monitoring data of the water source heat pump air conditioning system, calculate its hourly actual carbon emissions during actual operation; S15: Emission reduction calculation and analysis, compare the hourly baseline carbon emissions with the hourly actual carbon emissions, and calculate the dynamic emission reduction of the water source heat pump air conditioning system during the operation phase. S16: Life cycle emissions calculation, calculates the carbon emissions of water source heat pump air conditioning system throughout its life cycle, and adjusts the weights based on regional differences to calculate the net emissions throughout the life cycle after regional correction. S17: Results output, integrate the dynamic emission reduction during the operation phase with the net emission of the entire life cycle after regional correction, generate and output a comprehensive emission reduction benefit assessment report of the water source heat pump air conditioning system; The data completion steps specifically include: S21: Acquire historical energy consumption data for multiple complete operating cycles of the target building and similar buildings, and simultaneously collect corresponding meteorological data and building operation schedule data to form the original dataset; S22: Clean, align, and normalize the original dataset to construct a training dataset for model training; S23: Construct a data completion model based on generative adversarial networks, and use the training dataset to adversarially train the model until the discriminator can no longer stably distinguish between real data and generated data, thus obtaining a well-trained data completion model; S24: Input the incomplete historical operation data of the target building, as well as the meteorological data and building operation schedule data for the corresponding time period, into the trained data completion model to generate complete time series data, replace and fill in the missing values ​​and outliers in the original data, and form a complete building basic dataset.

2. The method for evaluating emission reduction data of a water source heat pump air conditioning system according to claim 1, characterized in that: The data completion model is a conditional generative adversarial network. The generator of the data completion model takes into account a conditional vector derived from incomplete historical operational data, a random noise vector, and a weather and operational timetable vector corresponding to the time period. The discriminator of the data completion model takes into account real complete data samples, generated data samples output by the generator, conditional vectors, and weather and operational timetable vectors.

3. The method for evaluating emission reduction data of a water source heat pump air conditioning system according to claim 2, characterized in that: The specific steps of baseline scenario modeling and simulation include: S31: Based on the complete building data set of the target building, construct and calibrate a digital twin energy consumption model that reflects the physical and energy consumption characteristics of the building, and replace the actual water source heat pump air conditioning system configuration with a preset virtual configuration of conventional air conditioning system; S32: Acquire hourly grid carbon emission factor data corresponding to the assessment period and associate it with the energy consumption-carbon emission conversion module of the digital twin energy consumption model for dynamic boundary conditions; S33: With minimizing operating costs as the optimization objective, artificial intelligence optimization algorithms are used to simulate and solve the optimal operation control strategy of the virtual configuration of the conventional air conditioning system throughout the entire evaluation period in the digital twin energy consumption model. S34: Based on the optimal operation control strategy, run the digital twin energy consumption model to obtain the hourly simulated energy consumption value of the virtual configuration of the conventional air conditioning system, and combine it with the hourly grid carbon emission factor data to calculate the hourly baseline carbon emissions under the optimal strategy.

4. The method for evaluating emission reduction data of a water source heat pump air conditioning system according to claim 3, characterized in that: The artificial intelligence optimization algorithm used in the baseline scenario modeling and simulation steps is a deep reinforcement learning algorithm, and its specific execution process includes: S41: State definition, which defines the building indoor and outdoor thermal environment parameters, system operating status, and current and predicted hourly grid carbon emission factors of the digital twin energy consumption model at the simulation time as the state variables of reinforcement learning; S42: Action definition, defining the adjustable control variables of the virtual configuration of the conventional air conditioning system, including setpoint temperature, equipment start / stop and frequency conversion, as action quantities for reinforcement learning; S43: Reward function definition. Set a reward function to reward low-energy consumption and low-carbon emission operating behaviors while meeting indoor thermal comfort constraints. S44: Policy learning, in a simulated environment composed of a digital twin energy consumption model, trains a deep reinforcement learning algorithm to learn to output the optimal operation control strategy that satisfies the optimization objective.

5. The method for evaluating emission reduction data of a water source heat pump air conditioning system according to claim 4, characterized in that: The actual emissions calculation steps include: S51: Real-time monitoring and acquisition of hourly operating data of the water source heat pump air conditioning system during the evaluation period. The hourly operating data includes the total power consumption of the system, the power consumption of each major electrical equipment, and the consumption of other types of energy when the system includes non-electrically driven equipment. S52: Based on hourly operating data, calculate the actual total energy consumption of the water source heat pump air conditioning system at each moment during the evaluation period; S53: Obtain the hourly grid carbon emission factor corresponding to the assessment period, and calculate the carbon emissions of the electrical energy consumed by the water source heat pump system based on the hourly grid carbon emission factor; S54: Cross-compare the calculated hourly actual carbon emissions with historical data for the same period, mark, verify and correct any abnormal data that exceeds the reasonable range, and form the final hourly actual carbon emission sequence.

6. The method for evaluating emission reduction data of a water source heat pump air conditioning system according to claim 5, characterized in that: The specific steps for calculating life-cycle emissions include: S61: Stage definition and initial calculation, define the entire life cycle stages of the water source heat pump air conditioning system, and use the life cycle assessment method to calculate the initial carbon emissions of each stage; S62: Obtaining regional difference parameters, obtaining regional difference parameters of the project location; S63: Dynamic weight coefficient calculation. Input the regional difference parameter into the preset regional influence weight adjustment model, and calculate and output the dynamic influence weight coefficient corresponding to each life cycle stage. S64: Regionally corrected emissions calculation: Multiply the initial carbon emissions for each stage by the corresponding dynamic impact weight coefficients for each stage to obtain the regionally corrected carbon emissions for each stage, and then sum them to obtain the regionally corrected net emissions of the water source heat pump air conditioning system throughout its entire life cycle.

7. The method for evaluating emission reduction data of a water source heat pump air conditioning system according to claim 6, characterized in that: The specific regional characteristic parameters include: A11: Environmental resource cost parameters, including the unit cost of groundwater recharge reflecting the impact of groundwater on the environment and the unit treatment cost of major pollution types reflecting the local environmental sensitivity; A12: Industry technical parameters, including the recycling rate of equipment and materials reflecting the level of resource recycling and the local recycling industry maturity index; A13: Demand parameters, including local carbon emission intensity constraint coefficients.

8. The method for evaluating emission reduction data of a water source heat pump air conditioning system according to claim 7, characterized in that: The regional impact weight adjustment model is an artificial neural network model based on a multilayer perceptron. It is used to dynamically calculate the dynamic impact weight coefficients of each life cycle stage of the water source heat pump air conditioning system according to the regional difference parameters of the project location. The construction of the regional impact weight adjustment model includes: S71: Construct a multilayer perceptron network structure that includes an input layer, at least one hidden layer, and an output layer; S72: Construct a training sample set using historical project data and expert knowledge. Each sample in the training sample set includes a set of normalized regional difference parameters as input features and a set of baseline weight coefficients corresponding to each life cycle stage determined by an optimization algorithm as labels. S73: Supervised training of the multilayer perceptron model is performed using the training sample set until its loss function converges, resulting in a trained weight adjustment model.

9. A data evaluation system for emission reduction of a water source heat pump air conditioning system, used to implement the data evaluation method for emission reduction of a water source heat pump air conditioning system as described in any one of claims 1-8, characterized in that: include: The data processing module is used to acquire, clean, and complete data to form a basic building dataset; The baseline scenario simulation module, connected to the data processing module, is used to build a digital twin model of a conventional air conditioning system and simulate its hourly baseline carbon emissions under optimal operation using AI. The actual emissions calculation module is used to calculate the actual hourly carbon emissions of the water source heat pump system; The comprehensive analysis module is connected to the baseline scenario simulation module and the actual emission calculation module, respectively. It is used to calculate the dynamic emission reduction during the operation phase, and to perform full life cycle carbon emission accounting and dynamic weight adjustment based on regional differences, and calculate the regionally corrected net emissions for the entire life cycle. The report generation module, connected to the comprehensive analysis module, is used to integrate the results and generate a comprehensive emission reduction benefit assessment report.

Images