Cold and heat source configuration optimization method and system based on full life cycle

Abstract

The invention belongs to the technical field of cold and heat source configuration methods, and particularly relates to a cold and heat source configuration optimization method and system based on a full life cycle, and the method comprises the steps: S1, building use characteristic analysis; s2, designing a staged unit configuration scheme; s3, optimizing an intelligent control strategy; s4, carrying out full-life-cycle cost-benefit analysis; s5, optimal scheme decision making and dynamic adjustment; according to the method, a dynamic prediction model of a time sequence analysis and Transform-TCN hybrid architecture is adopted, historical load data can be deeply mined through time sequence analysis, and time characteristics and periodic rules in the data are captured; the long sequence modeling capability of the Transform and the local feature extraction capability of the TCN are combined by the Transform-TCN hybrid architecture, so that the long and short term dependency relationship in the load data can be better processed, and the prediction accuracy is improved.

Description

Technical Field

[0001] This invention belongs to the technical field of cold and heat source configuration methods, specifically a cold and heat source configuration optimization method and system based on the entire life cycle. Background Technology

[0002] The configuration of cold and heat sources is the core equipment combination used to provide cooling and heating in a central air conditioning system. Its selection needs to take into account load demand, energy conditions and economics. With social development, the consumption of energy such as cooling and heating is increasing year by year, and the total energy consumption increases every year. The total social energy consumption has become high. Optimizing the use of cold and heating energy in energy selection has become an effective way to improve energy efficiency and effectively reduce the total social energy consumption.

[0003] A method and system for optimizing the configuration of heating and cooling sources based on the entire life cycle, as disclosed in invention patent publication number CN116128236A, includes the following steps: acquiring existing energy resources around the project, including industrial waste heat, geological conditions, and power conditions. The key technical point is that when selecting a heating and cooling source configuration scheme, multiple factors such as industrial waste heat, geological conditions, and power conditions are considered. Simultaneously, carbon emission costs are incorporated into the total life-cycle cost of the heating and cooling sources, avoiding problems arising from considering only one factor in the configuration.

[0004] However, the above technologies often have the following drawbacks: Simply obtaining the project's heating and cooling loads without in-depth analysis of the impact of building usage characteristics on the heating and cooling loads, and lacking scientific and accurate load forecasting methods, makes it difficult to accurately grasp the heating and cooling load demands under different time periods and operating conditions, which can easily lead to energy waste or insufficient energy supply. Furthermore, the dynamics of building usage characteristics such as functional zoning adjustments, changes in population density, and external environments such as energy price fluctuations and carbon price changes were not incorporated into the optimization framework, resulting in poor system adaptability and decreased long-term operating efficiency. Most systems use fixed-frequency generators or simple start-stop control, which cannot dynamically adjust the output power according to real-time load demand, resulting in low energy efficiency and high operating costs for the generators under partial load.

[0005] Therefore, the present invention provides a method and system for optimizing the configuration of cold and heat sources based on the entire life cycle. Summary of the Invention

[0006] To overcome the shortcomings of existing technologies, this invention provides a method and system for optimizing the configuration of cold and heat sources based on the entire life cycle, thereby solving at least one technical problem mentioned in the background art.

[0007] The technical solution adopted by this invention to solve its technical problem is: a method for optimizing the configuration of cold and heat sources based on the entire life cycle, comprising the following steps: S1. Building Usage Characteristics Analysis: Obtain historical load data, usage time distribution, and functional zoning information of the target building. Combine this with the climate conditions of the building's location to establish a dynamic prediction model for the building's heating and cooling loads, and obtain the heating and cooling load demand curves based on this model. S2. Phased unit configuration scheme design: Based on the cooling and heating load demand curve, divide the basic load range and the peak load range, configure small-capacity units to meet the basic load demand, and configure large-capacity units as peak load supplements. S3. Intelligent control strategy optimization: Deploy variable frequency drive devices in small-capacity and large-capacity units, collect compressor or water pump speed, current and voltage parameters in real time, combine with the cooling and heating load demand curves, dynamically adjust the unit output power, and establish a multi-objective optimization rule library for unit start-up and power distribution. S4. Life Cycle Cost-Benefit Analysis: Construct a life cycle cost model that includes initial investment cost, operation and maintenance cost, energy consumption cost, equipment replacement cost, and carbon emission cost, and obtain the life cycle cost-benefit analysis results; S5. Optimal Solution Decision and Dynamic Adjustment: Based on the results of the full life cycle cost-benefit analysis, the configuration scheme with the lowest overall cost, carbon emission intensity lower than the industry average, and minimum carbon risk value is selected as the initial optimal scheme.

[0008] A further improvement of the present invention is that the building cooling and heating load dynamic prediction model adopts a combination of time series analysis and Transformer-TCN hybrid architecture. The input parameters include outdoor temperature, humidity, solar radiation intensity, building internal personnel density, and equipment heat generation. The output is the hourly cooling and heating load demand curve for the next 24 to 72 hours. Historical load data needs to undergo wavelet denoising and quantile mapping to eliminate outliers and seasonal biases.

[0009] A further improvement of the present invention lies in the phased unit configuration scheme design phase: The types of units include air source heat pumps, ground source heat pumps, chillers, and gas boilers, and include at least one variable frequency control device; The rated power of small-capacity units is 30%-50% of the building's maximum cooling and heating load; The rated power of the large-capacity unit is 60%-80% of the building's maximum cooling and heating load, and it complements the capacity of the small-capacity unit. When the two are used together, they cover 100% of the peak load demand. Furthermore, the capacity ratio between small-capacity and large-capacity units can be dynamically adjusted according to the building load characteristics.

[0010] A further improvement of the present invention lies in the intelligent control strategy optimization stage: The SAC algorithm in deep reinforcement learning is adopted. The deviation between load demand and actual output, the rate of change of deviation, and unit energy consumption are used as state inputs, and the output frequency adjustment command of the frequency converter and the start-up and shutdown decision of the unit are output. The rule base is generated offline through genetic algorithm optimization and contains ≥100 rules, covering load surges, equipment start-up and shutdown, fault switching, and energy price fluctuations.

[0011] A further improvement of the present invention lies in the life-cycle cost-benefit analysis stage: Carbon emission costs are calculated based on local carbon trading market prices and the carbon emission coefficient per unit energy consumption of the generating unit. The carbon emission coefficient is determined by measured unit energy consumption and standard coal conversion coefficient, and a dynamic carbon price prediction model is introduced. Using a multi-scenario tree and nested Monte Carlo simulation method, more than 5,000 cost scenarios are generated by randomly sampling parameters such as unit operating life, energy price fluctuations, equipment failure rate, and carbon price fluctuation uncertainty. Calculate the net present value, internal rate of return, and carbon risk value under each scenario, and screen the allocation schemes with carbon risk value ≥ 0, internal rate of return higher than the industry benchmark rate of return, and carbon risk value lower than the threshold. The dynamic carbon price prediction model uses an LSTM network at the bottom layer to process historical carbon price data, an ARIMA model in the middle layer to capture linear trends, and an attention mechanism at the top layer to fuse the outputs of both models, which is used to predict the carbon price trend in the next 12 months.

[0012] A further improvement of the present invention is that the dynamic adjustment includes: Establish a unit operation data feedback mechanism to collect actual operating energy consumption, equipment failure frequency, maintenance costs, and carbon emission data every quarter, and update the parameters of the full life cycle cost model and the carbon price prediction model. When the actual operating data deviates from the prediction model by more than the set percentage, steps S1-S4 are executed again. By introducing digital twin technology, a virtual mirror of the building's heating and cooling system is constructed. The simulation model is driven by real-time data to verify the effectiveness of the optimization strategy. The digital twin model is built using Unity3D or Unreal Engine, integrating building thermal models, unit performance models, and control strategies, and supports VR / AR interactive optimization.

[0013] A further improvement of the present invention is that, in the digital twin model, the building thermal model is constructed using a method based on finite element analysis; The unit performance model is obtained by fitting measured data and includes the unit's energy efficiency ratio and power output parameters at different load rates. The control strategy module integrates the SAC algorithm and multi-objective optimization rule base from the intelligent control module to achieve synchronous simulation optimization with the actual system.

[0014] A cold and heat source configuration optimization system based on the entire life cycle, applied to a cold and heat source configuration optimization method based on the entire life cycle, includes: Data acquisition and preprocessing module: Used to collect and process building-related data, it includes multiple sensors and transmits the collected data to the load forecasting module at a frequency of once per minute; Load forecasting module: used to build a dynamic forecasting model of building heating and cooling loads, receive data transmitted from the data acquisition and preprocessing module, output hourly load demand curves, and transmit the load demand curves to the unit configuration module; Unit configuration module: used to design phased unit configuration schemes based on load demand curves, generate a capacity combination list, and transmit the configuration schemes to the intelligent control module; Intelligent control module: used to deploy SAC algorithm and multi-objective optimization rule base, receive configuration scheme transmitted by unit configuration module and load demand curve transmitted by load forecast module, adjust unit output power and start / stop status in real time, and output frequency converter control commands to unit equipment; Cost Analysis Module: Used to build a full life cycle cost model, receive unit operation data, perform simulations, and output a cost-benefit risk analysis report to the decision optimization module; Decision optimization module: Selects the optimal configuration scheme based on cost analysis results, supports dynamic adjustment of strategies, includes a scheme library sub-module, and selects and adjusts schemes based on cost analysis reports and actual operation data feedback; User interaction module: It provides a visual interface to display relevant data, supports manual user intervention, and interacts with the decision optimization module. Edge computing module: Deployed on local servers or industrial gateways to perform data preprocessing, interact with data acquisition and preprocessing modules and load forecasting modules, and provide computing support; Cloud platform module: Deployed in public or private cloud, it supports multi-building cluster management, interacts with edge computing module and decision optimization module for data, and provides storage and computing support; Mobile application module: Developed based on Android or iOS platforms, it supports remote monitoring and data interaction with cloud platform modules.

[0015] A further improvement of this invention is that, after the data acquisition and preprocessing module acquires data, it transmits it to the load forecasting module to generate a load demand curve. The unit configuration module designs a configuration scheme based on this curve. The intelligent control module adjusts the unit in real time according to the load demand curve and the configuration scheme. The cost analysis module performs cost analysis based on the unit operation data. The decision optimization module selects and adjusts the scheme based on the cost analysis results. The user interaction module provides an operation interface for users. The edge computing module and the cloud platform module provide computing and storage support for each module. The mobile application module facilitates remote operation by users. The modules communicate with each other via the gRPC protocol. The data storage adopts a hybrid architecture of time-series database InfluxDB, relational database PostgreSQL, and object storage MinIO.

[0016] A further improvement of this invention is that it adopts a microservice architecture, with each module independently deployed on edge computing devices and cloud servers. The modules communicate with each other via the gRPC protocol, and the data storage adopts a hybrid architecture of time-series database InfluxDB, relational database PostgreSQL, and object storage MinIO.

[0017] The beneficial effects of this invention are as follows: 1. This invention employs a dynamic prediction model combining time series analysis and a Transformer-TCN hybrid architecture. Time series analysis enables in-depth mining of historical load data, capturing temporal characteristics and periodic patterns. The Transformer-TCN hybrid architecture combines the long-sequence modeling capabilities of Transformer with the local feature extraction capabilities of TCN, better handling long-term and short-term dependencies in load data and improving prediction accuracy. Simultaneously, wavelet denoising effectively eliminates outliers and noise interference, while quantile mapping addresses seasonal biases, further enhancing data quality and prediction accuracy, and preventing energy waste or insufficient power supply. A complementary configuration of small-capacity and large-capacity units is adopted, along with the deployment of variable frequency drives. The small-capacity units operate under low load conditions to meet basic power needs, while the large-capacity units operate under high load conditions to provide sufficient power support. Through the variable frequency drive, the output power of the units can be dynamically adjusted according to the actual load, ensuring optimal efficiency under different load conditions and reducing operating costs. 2. This invention monitors the unit's operating data in real time, such as power, temperature, and pressure, and feeds this data back to the predictive model and control strategy. The model parameters are updated and optimized quarterly to better adapt the model to the actual operating conditions of the system. A digital twin model built with Unity3D / UnrealEngine is introduced, integrating the building thermal model, the unit performance model, and the control strategy. The digital twin model can simulate the actual system's operating state in real time, the building thermal model can accurately simulate the building's heat transfer process, the unit performance model can reflect the actual operating performance of the unit, and the control strategy is dynamically adjusted based on real-time data. It supports VR / AR interactive optimization. Through VR / AR technology, operators can intuitively observe the operation of the digital twin model, perform interactive operations and optimization adjustments, and verify the optimization strategy through simulation driven by real-time data. The control strategy of the actual system is adjusted and optimized based on the simulation results to ensure that the system can operate stably for a long time without degradation of operating efficiency. 3. This invention introduces a dynamic carbon price prediction model. LSTM can handle long-term dependencies in time series data, ARIMA can perform stationarity analysis and prediction on time series, and the attention mechanism can highlight key information and improve the accuracy of prediction. By combining these three models, the future trend of carbon price changes can be accurately predicted. By combining multi-scenario trees and Monte Carlo simulation to generate more than 5,000 cost scenarios, multi-scenario trees can construct different future scenarios, and Monte Carlo simulation can perform random sampling and simulation analysis on these scenarios, thereby comprehensively considering the impact of various uncertainties on costs. Attached Figure Description

[0018] The invention will now be further described with reference to the accompanying drawings.

[0019] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 , Example 1 This embodiment provides a method for optimizing the configuration of cold and heat sources based on their entire life cycle, including the following steps: S1. Building Usage Characteristics Analysis: Historical load data, usage time distribution, and functional zoning information of the target building are obtained. Combined with the local climate conditions, a dynamic prediction model for building heating and cooling loads is constructed using a hybrid architecture combining time series analysis and the Transformer-TCN framework. Input parameters include outdoor temperature, humidity, solar radiation intensity, building occupancy density, and equipment heat generation. Wavelet denoising and quantile mapping are applied to the historical load data to eliminate outliers and seasonal biases. The output is hourly heating and cooling load demand curves for the next 24 to 72 hours. In the Transformer-TCN hybrid architecture, the Transformer module utilizes its self-attention mechanism to capture long-distance dependencies in the load data, taking into account the mutual influence between loads at different time periods. The TCN module effectively expands the receptive field through dilated convolution operations, capturing local features and short-term variation patterns of the load data. The hybrid approach involves combining the Transformer and TCN modules in series or parallel to fully leverage their advantages in processing time series data and improve the model's prediction accuracy.

[0022] Historical load data is obtained from the building's energy management system or dedicated load monitoring equipment, covering a period of time with a recording frequency of hourly or higher. Usage time distribution is determined by the building's operation and management records, personnel entry and exit data, etc., such as the business hours of commercial buildings and the working hours of office buildings. Functional zoning information is based on the building's floor plan and actual usage, clarifying the purpose of each area, such as office area, catering area, shopping mall area, etc. Climate conditions are obtained from local meteorological stations or meteorological data service platforms, including historical and forecast data of outdoor meteorological parameters such as temperature, humidity, and solar radiation intensity.

[0023] It should be noted that S1 constructs an accurate dynamic prediction model of cooling and heating loads by acquiring multi-faceted information about the target building, providing basic data for subsequent unit configuration and optimization. This model can take into account the impact of various factors on the building's cooling and heating loads, making the prediction results closer to the actual situation. For example, for a commercial complex located in a cold northern region, analysis of its historical load data reveals that the demand for cooling and heating loads increases significantly during peak business hours in winter, such as weekends and holidays. The usage time distribution shows that the load is higher during daytime business hours and lower at night. Functional zoning information indicates that the catering area has a relatively large load demand due to the high density of people and the heat generated by equipment. Combining local climate conditions, the model constructed using a hybrid architecture of time series analysis and Transformer-TCN can accurately predict the cooling and heating load demand for different periods in the future. Furthermore, when acquiring information on the density of people inside a building, in addition to historical data, personnel counting sensors can be installed to monitor changes in the number of people in real time, so as to more accurately reflect the impact of personnel density on heating and cooling loads. For the heat generation of equipment, a detailed equipment list should be established, recording information such as the power and operating time of each piece of equipment, and the data should be updated in a timely manner according to the upgrading of equipment.

[0024] S2. Phased Unit Configuration Design: Based on the cooling and heating load demand curves, the system is divided into basic load ranges and peak load ranges. Unit types include air source heat pumps, ground source heat pumps, chillers, and gas boilers, and must include at least one variable frequency drive (VFD) device. Small-capacity units are configured to meet basic load demands, with a rated power of 30%-50% of the building's maximum cooling and heating load. Large-capacity units are configured to supplement peak loads, with a rated power of 60%-80% of the building's maximum cooling and heating load, and their capacity complements that of the small-capacity units. When both units operate together, they cover 100% of the peak load demand. The capacity ratio of small-capacity units to large-capacity units can be dynamically adjusted according to the building's load characteristics. The unit types include air source heat pumps, ground source heat pumps, chillers, and gas boilers, and must include at least one variable frequency drive (VFD) device. Air source heat pumps utilize heat energy from the air for cooling or heating, are suitable for most climate conditions, and have advantages such as easy installation and relatively low operating costs. Ground source heat pumps exchange energy through heat energy from underground soil or water bodies, offering high energy efficiency, but with higher installation costs and certain geological requirements. Chillers are primarily used for cooling, achieving temperature reduction through refrigerant circulation, and are suitable for centralized cooling needs in large buildings. Gas boilers use natural gas as fuel to provide heating, and are used in winter or when heating demand is high. The VFD device is selected from the above units and must possess at least one VFD function to dynamically adjust the output power according to load demand.

[0025] The rated power of small-capacity units is 30%-50% of the building's maximum cooling and heating load; The rated power of the large-capacity unit is 60%-80% of the building's maximum cooling and heating load, and it complements the capacity of the small-capacity unit. When the two are used together, they cover 100% of the peak load demand. The capacity ratio of small-capacity units to large-capacity units can be dynamically adjusted according to the building load characteristics. For example, for office buildings, small-capacity units account for 40%-60%; for commercial complexes, small-capacity units account for 30%-50%; and for industrial plants, small-capacity units account for 20%-40%.

[0026] It should be noted that S2 rationally divides the base load range and peak load range according to the heating and cooling load demand curve, and configures units of corresponding types and capacities to ensure efficient operation under different load conditions, while meeting the needs of load changes. For example, for the aforementioned commercial complex, based on the predicted heating and cooling load demand curve, the base load range is determined to be 30%-50% of the building's maximum heating and cooling load, and small-capacity air source heat pump units are configured, with a rated power of 40% of the building's maximum heating and cooling load; the peak load range is 60%-80%, and large-capacity gas boiler units are configured, with a rated power of 70% of the building's maximum heating and cooling load, and their capacity complements that of the small-capacity units. When the two operate together, they can cover 100% of the peak load demand. During the peak business hours in summer, when the load reaches its peak, the large-capacity gas boiler units start up and operate together with the small-capacity air source heat pump units to meet the heating and cooling needs of the commercial complex. Furthermore, when selecting the type of heat pump unit, in addition to considering common types such as air source heat pumps, ground source heat pumps, chillers, and gas boilers, a comprehensive evaluation should also be conducted based on local energy policies and resource availability. For example, if there are abundant geothermal resources in the area and policies encourage the use of ground source heat pumps, then ground source heat pump units should be given priority when selecting the unit type. At the same time, the capacity configuration of the unit should take into account the possible future expansion or functional adjustment of the building, and reserve a certain capacity margin.

[0027] S3. Intelligent Control Strategy Optimization: Variable frequency drive devices are deployed in both small-capacity and large-capacity units. By collecting real-time compressor or water pump speed, current, and voltage parameters, combined with heating and cooling load demand curves, the SAC algorithm in deep reinforcement learning is used. The deviation between load demand and actual output, the rate of change of deviation, and unit energy consumption are used as state inputs. The output is frequency adjustment commands for the inverter and unit start-up and shutdown decisions. A multi-objective optimization rule base for unit start-up and power allocation is established. The rule base is generated offline through genetic algorithm optimization and contains ≥100 rules, covering load changes, equipment start-up and shutdown, fault switching, and energy price fluctuations. The reward function of the SAC algorithm is designed as follows: in, For load matching error, For power adjustment range, For carbon emissions, For a safe state set, , , and These are all weighting coefficients used to balance targets such as load matching, power regulation, carbon emissions, and safety status. It should be noted that the S3, by deploying variable frequency drive devices and employing the SAC algorithm, dynamically adjusts the unit's output power and start / stop status based on real-time load demand, improving the unit's energy efficiency under partial load. Simultaneously, it can quickly respond to sudden load changes, ensuring system stability. For example, during the operation of a commercial complex, parameters such as compressor or water pump speed, current, and voltage are collected in real time, combined with heating and cooling load demand curves. When the load suddenly increases, the SAC algorithm, based on the deviation between load demand and actual output, the rate of change of deviation, and unit energy consumption, quickly outputs frequency adjustment commands to the inverter, increasing the unit's output power. When the load decreases, it promptly reduces the unit's output power or shuts it down to avoid energy waste. For instance, when a large event is held in the shopping mall, and the sudden increase in people leads to a rise in load, the intelligent control system adjusts the unit's operating status in a short time to maintain the indoor temperature within a comfortable range.

[0028] To further improve the adaptability of intelligent control strategies, this method also includes establishing a historical operating database of the unit to analyze the unit's operating performance and energy consumption under different operating conditions, providing more data support for the training of the SAC algorithm. At the same time, for the multi-objective optimization rule base, in addition to covering operating conditions such as load changes, equipment start-up and shutdown, fault switching, and energy price fluctuations, the impact of meteorological factors such as ambient temperature and humidity on unit operation can also be considered, making the control strategy more comprehensive and optimized.

[0029] S4. Life Cycle Cost-Benefit Analysis: Construct a life cycle cost model that includes initial investment costs, operation and maintenance costs, energy consumption costs, equipment replacement costs, and carbon emission costs. Initial investment costs include one-time costs related to project initiation, such as the purchase cost of unit equipment, installation and commissioning costs, and building renovation costs. Operation and maintenance costs cover daily operating costs such as energy consumption costs, equipment maintenance costs, and personnel management costs during unit operation. Energy consumption costs are calculated based on the unit's energy consumption and energy prices, representing the total energy consumption cost over the project's entire life cycle. Equipment replacement costs consider the lifespan of the unit equipment and the need for upgrades, estimating the cost of replacing equipment over the project's entire life cycle. Carbon emission costs are based on local carbon trading market prices and the unit's unit energy consumption. Carbon emission coefficients are calculated using measured unit energy consumption and standard coal conversion coefficients. A dynamic carbon price prediction model is introduced, with the bottom layer using an LSTM network to process historical carbon price data, the middle layer using an ARIMA model to capture linear trends, and the top layer using an attention mechanism to fuse the outputs of both models to predict carbon price trends over the next 12 months. A multi-scenario tree and nested Monte Carlo simulation method is employed to randomly sample parameters such as unit operating life, energy price fluctuations, equipment failure rates, and carbon price fluctuation uncertainty, generating over 5000 cost scenarios. Net present value, internal rate of return, and carbon risk value are calculated for each scenario. Configuration schemes with a carbon risk value ≥ 0, an internal rate of return higher than the industry benchmark rate of return, and a carbon risk value lower than a threshold are selected to obtain the full life-cycle cost-benefit analysis results. It should be noted that S4 comprehensively considers various cost factors of a project by constructing a full life cycle cost model, and selects configuration schemes that are both economically and environmentally sound through dynamic carbon price prediction models and scenario simulations. For example, for a commercial complex's heating and cooling source configuration project, a full life cycle cost model is constructed that includes initial investment costs such as unit procurement, installation costs, operation and maintenance costs, energy consumption during unit operation, maintenance costs, energy consumption costs, equipment replacement costs, and carbon emission costs. The dynamic carbon price prediction model predicts the carbon price trend for the next 12 months, and 5,000 cost scenarios are generated by combining multi-scenario trees and Monte Carlo simulations. After calculation and analysis, configuration schemes with a carbon risk value ≥ 0, an internal rate of return higher than the industry benchmark rate of return, and a carbon risk value lower than the threshold are selected. For example, a certain combination of unit configuration and control strategy is selected to minimize the project's cost throughout its entire life cycle, while keeping the carbon emission intensity lower than the industry average. Furthermore, when calculating carbon emission costs, in addition to considering carbon emissions during unit operation, carbon emissions from unit production, transportation, and installation should also be considered to make the calculation of carbon emission costs more comprehensive and accurate. At the same time, for dynamic carbon price prediction models, more influencing factors can be incorporated to improve the accuracy of carbon price prediction.

[0030] S5. Optimal Solution Decision and Dynamic Adjustment: Based on the full life cycle cost-benefit analysis results, select the configuration scheme with the lowest overall cost, carbon emission intensity lower than the industry average, and minimum carbon risk value as the initial optimal scheme; establish a unit operation data feedback mechanism, collect actual operating energy consumption, equipment failure frequency, maintenance costs, and carbon emission data every quarter, and update the full life cycle cost model parameters and carbon price prediction model; when the deviation between actual operating data and prediction model exceeds a set percentage, such as 10%, repeat steps S1-S4; introduce digital twin technology to construct a virtual image of the building's heating and cooling source system, and verify the effectiveness of the optimization strategy by driving the simulation model with real-time data; the digital twin model is built using Unity3D or Unreal Engine, integrating the building thermal model, unit performance model, and control strategy, and supports VR / AR interactive optimization.

[0031] It should be noted that S5 selects the initial optimal solution based on the results of the full life cycle cost-benefit analysis. By establishing a unit operation data feedback mechanism and introducing digital twin technology, it achieves dynamic adjustment and verification of the optimization strategy, ensuring the long-term efficient and stable operation of the system. For example, based on the full life cycle cost-benefit analysis results, the configuration scheme with the lowest overall cost, carbon emission intensity lower than the industry average, and lowest carbon risk value is selected as the initial optimal solution. During project operation, data such as actual operating energy consumption, equipment failure frequency, maintenance costs, and carbon emissions are collected quarterly to update the parameters of the full life cycle cost model and the carbon price prediction model. When the deviation between the actual operating data and the prediction model exceeds 10%, the previous steps are repeated for optimization and adjustment. Simultaneously, digital twin technology is introduced to construct a virtual mirror of the commercial complex's heating and cooling source system. The effectiveness of the optimization strategy is verified through real-time data-driven simulation models. For example, the digital twin model simulates the system operation under different control strategies, providing a reference for adjusting the actual system.

[0032] Furthermore, when establishing a unit operation data feedback mechanism, the accuracy and timeliness of the data should be ensured. Automated data acquisition systems can be used to reduce human intervention and improve data quality. For digital twin models, in addition to integrating building thermal models, unit performance models, and control strategies, integration with other building systems such as electrical and water supply and drainage systems can be considered to achieve more comprehensive system simulation and optimization. At the same time, digital twin models should support remote access and control, facilitating system monitoring and optimization by managers anytime and anywhere.

[0033] Example 2 This embodiment provides: a cold and heat source configuration optimization system based on the entire life cycle, including: Data acquisition and preprocessing module: Used to collect and process building-related data, it includes multiple sensors and transmits the collected data to the load forecasting module at a frequency of once per minute; Load forecasting module: used to build a dynamic forecasting model of building heating and cooling loads, receive data transmitted from the data acquisition and preprocessing module, output hourly load demand curves, and transmit the load demand curves to the unit configuration module; Unit configuration module: used to design phased unit configuration schemes based on load demand curves, generate a capacity combination list, and transmit the configuration schemes to the intelligent control module; Intelligent control module: used to deploy SAC algorithm and multi-objective optimization rule base, receive configuration scheme transmitted by unit configuration module and load demand curve transmitted by load forecast module, adjust unit output power and start / stop status in real time, and output frequency converter control commands to unit equipment; Cost Analysis Module: Used to build a full life cycle cost model, receive unit operation data, perform simulations, and output a cost-benefit risk analysis report to the decision optimization module; Decision optimization module: Selects the optimal configuration scheme based on cost analysis results, supports dynamic adjustment of strategies, includes a scheme library sub-module, and selects and adjusts schemes based on cost analysis reports and actual operation data feedback; User interaction module: It provides a visual interface to display relevant data, supports manual user intervention, and interacts with the decision optimization module. Edge computing module: Deployed on local servers or industrial gateways to perform data preprocessing, interact with data acquisition and preprocessing modules and load forecasting modules, and provide computing support; Cloud platform module: Deployed in public or private cloud, it supports multi-building cluster management, interacts with edge computing module and decision optimization module for data, and provides storage and computing support; Mobile application module: Developed based on Android or iOS platforms, it supports remote monitoring and data interaction with cloud platform modules.

[0034] After collecting data, the data acquisition and preprocessing module transmits it to the load forecasting module to generate a load demand curve. The unit configuration module designs a configuration scheme based on this curve. The intelligent control module adjusts the unit in real time according to the load demand curve and the configuration scheme. The cost analysis module performs cost analysis based on the unit operation data. The decision optimization module selects and adjusts the scheme based on the cost analysis results. The user interaction module provides an operating interface for users. The edge computing module and cloud platform module provide computing and storage support for each module. The mobile application module facilitates remote operation by users. The modules communicate with each other via the gRPC protocol. Data storage adopts a hybrid architecture of time-series database InfluxDB, relational database PostgreSQL, and object storage MinIO.

[0035] The system adopts a microservice architecture, with each module deployed independently on edge computing devices and cloud servers. The modules communicate with each other via the gRPC protocol, and the data storage adopts a hybrid architecture of time-series database InfluxDB, relational database PostgreSQL, and object storage MinIO.

[0036] The terms "front," "back," "left," "right," "top," and "bottom" all refer to the figures in the accompanying drawings. Figure 1 Based on the perspective of the observer, the side of the device facing the observer is defined as the front, the left side of the observer is defined as the left, and so on.

[0037] In the description of this invention, it should be understood that the terms "center", "longitudinal", "lateral", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting the scope of protection of this invention.

[0038] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing the configuration of cold and heat sources based on the entire life cycle, characterized in that: Includes the following steps: S1. Building Usage Characteristics Analysis: Obtain historical load data, usage time distribution, and functional zoning information of the target building. Combine this with the climate conditions of the building's location to establish a dynamic prediction model for the building's heating and cooling loads, and obtain the heating and cooling load demand curves based on this model. S2. Phased unit configuration scheme design: Based on the cooling and heating load demand curve, divide the basic load range and the peak load range, configure small-capacity units to meet the basic load demand, and configure large-capacity units as peak load supplements. S3. Intelligent control strategy optimization: Deploy variable frequency drive devices in small-capacity and large-capacity units, collect compressor or water pump speed, current and voltage parameters in real time, combine with the cooling and heating load demand curves, dynamically adjust the unit output power, and establish a multi-objective optimization rule library for unit start-up and power distribution. S4. Life Cycle Cost-Benefit Analysis: Construct a life cycle cost model that includes initial investment cost, operation and maintenance cost, energy consumption cost, equipment replacement cost, and carbon emission cost, and obtain the life cycle cost-benefit analysis results; S5. Optimal Solution Decision and Dynamic Adjustment: Based on the results of the full life cycle cost-benefit analysis, the configuration scheme with the lowest overall cost, carbon emission intensity lower than the industry average, and minimum carbon risk value is selected as the initial optimal scheme.

2. The method for optimizing the configuration of cold and heat sources based on the entire life cycle as described in claim 1, characterized in that: The building cooling and heating load dynamic prediction model adopts a combination of time series analysis and Transformer-TCN hybrid architecture. The input parameters include outdoor temperature, humidity, solar radiation intensity, building occupancy density, and equipment heat generation. The output is the hourly cooling and heating load demand curve for the next 24 to 72 hours. Historical load data needs to undergo wavelet denoising and quantile mapping to eliminate outliers and seasonal biases.

3. The method for optimizing the configuration of cold and heat sources based on the entire life cycle as described in claim 1, characterized in that: The phased unit configuration scheme design phase: The types of units include air source heat pumps, ground source heat pumps, chillers, and gas boilers, and include at least one variable frequency control device; The rated power of small-capacity units is 30%-50% of the building's maximum cooling and heating load; The rated power of the large-capacity unit is 60%-80% of the building's maximum cooling and heating load, and it complements the capacity of the small-capacity unit. When the two are used together, they cover 100% of the peak load demand. Furthermore, the capacity ratio between small-capacity and large-capacity units can be dynamically adjusted according to the building load characteristics.

4. The method for optimizing the configuration of cold and heat sources based on the entire life cycle as described in claim 1, characterized in that: The intelligent control strategy optimization phase: The SAC algorithm in deep reinforcement learning is adopted. The deviation between load demand and actual output, the rate of change of deviation, and unit energy consumption are used as state inputs, and the output frequency adjustment command of the frequency converter and the start-up and shutdown decision of the unit are output. The rule base is generated offline through genetic algorithm optimization and contains ≥100 rules, covering load surges, equipment start-up and shutdown, fault switching, and energy price fluctuations.

5. The method for optimizing the configuration of cold and heat sources based on the entire life cycle as described in claim 1, characterized in that: The life-cycle cost-benefit analysis phase: Carbon emission costs are calculated based on local carbon trading market prices and the carbon emission coefficient per unit energy consumption of the generating unit. The carbon emission coefficient is determined by measured unit energy consumption and standard coal conversion coefficient, and a dynamic carbon price prediction model is introduced. Using a multi-scenario tree and nested Monte Carlo simulation method, more than 5,000 cost scenarios are generated by randomly sampling parameters such as unit operating life, energy price fluctuations, equipment failure rate, and carbon price fluctuation uncertainty. Calculate the net present value, internal rate of return, and carbon risk value under each scenario, and screen the allocation schemes with carbon risk value ≥ 0, internal rate of return higher than the industry benchmark rate of return, and carbon risk value lower than the threshold. The dynamic carbon price prediction model uses an LSTM network at the bottom layer to process historical carbon price data, an ARIMA model in the middle layer to capture linear trends, and an attention mechanism at the top layer to fuse the outputs of both models, which is used to predict the carbon price trend in the next 12 months.

6. The method for optimizing the configuration of cold and heat sources based on the entire life cycle as described in claim 1, characterized in that: The dynamic adjustment includes: Establish a unit operation data feedback mechanism to collect actual operating energy consumption, equipment failure frequency, maintenance costs, and carbon emission data every quarter, and update the parameters of the full life cycle cost model and the carbon price prediction model. When the actual operating data deviates from the prediction model by more than the set percentage, steps S1-S4 are executed again. By introducing digital twin technology, a virtual mirror of the building's heating and cooling system is constructed. The simulation model is driven by real-time data to verify the effectiveness of the optimization strategy. The digital twin model is built using Unity3D or Unreal Engine, integrating building thermal models, unit performance models, and control strategies, and supports VR / AR interactive optimization.

7. The method for optimizing the configuration of cold and heat sources based on the entire life cycle as described in claim 6, characterized in that: In the digital twin model, the building thermal model is constructed using a finite element analysis method; The unit performance model is obtained by fitting measured data and includes the unit's energy efficiency ratio and power output parameters under different load rates.

8. A cold and heat source configuration optimization system based on the entire life cycle, applied to the cold and heat source configuration optimization method based on the entire life cycle as described in any one of claims 1-7, characterized in that, include: Data acquisition and preprocessing module: Used to collect and process building-related data, it includes multiple sensors and transmits the collected data to the load forecasting module at a frequency of once per minute; Load forecasting module: used to build a dynamic forecasting model of building heating and cooling loads, receive data transmitted from the data acquisition and preprocessing module, output hourly load demand curves, and transmit the load demand curves to the unit configuration module; Unit configuration module: used to design phased unit configuration schemes based on load demand curves, generate a capacity combination list, and transmit the configuration schemes to the intelligent control module; Intelligent control module: used to deploy SAC algorithm and multi-objective optimization rule base, receive configuration scheme transmitted by unit configuration module and load demand curve transmitted by load forecast module, adjust unit output power and start / stop status in real time, and output frequency converter control commands to unit equipment; Cost Analysis Module: Used to build a full life cycle cost model, receive unit operation data, perform simulations, and output a cost-benefit risk analysis report to the decision optimization module; Decision optimization module: Selects the optimal configuration scheme based on cost analysis results, supports dynamic adjustment of strategies, includes a scheme library sub-module, and selects and adjusts schemes based on cost analysis reports and actual operation data feedback; User interaction module: It provides a visual interface to display relevant data, supports manual user intervention, and interacts with the decision optimization module. Edge computing module: used to deploy on local servers or industrial gateways to perform data preprocessing, interact with data acquisition and preprocessing modules and load forecasting modules, and provide computing support; Cloud platform module: Deployed in public or private cloud, it supports multi-building cluster management, interacts with edge computing module and decision optimization module for data, and provides storage and computing support; Mobile application module: Developed based on Android or iOS platforms, it supports remote monitoring and data interaction with cloud platform modules.

9. A cold and heat source configuration optimization system based on the entire life cycle as described in claim 8, characterized in that: The data acquisition and preprocessing module collects data and transmits it to the load forecasting module to generate a load demand curve. The unit configuration module designs a configuration scheme based on this curve. The intelligent control module adjusts the unit in real time according to the load demand curve and the configuration scheme. The cost analysis module performs cost analysis based on the unit operation data. The decision optimization module selects and adjusts the scheme based on the cost analysis results. The user interaction module provides an operation interface for users. The edge computing module and cloud platform module provide computing and storage support for each module. The mobile application module facilitates remote operation by users. The modules communicate with each other via the gRPC protocol. Data storage adopts a hybrid architecture of time-series database InfluxDB, relational database PostgreSQL, and object storage MinIO.

10. A cold and heat source configuration optimization system based on the entire life cycle as described in claim 8, characterized in that: It adopts a microservice architecture, with each module deployed independently on edge computing devices and cloud servers. The modules communicate with each other via the gRPC protocol, and the data storage adopts a hybrid architecture of time-series database InfluxDB, relational database PostgreSQL, and object storage MinIO.

Images