An AI digital twin energy efficiency and state optimization method and system for a subway vehicle air conditioning system

Abstract

This invention discloses an AI digital twin energy efficiency and state optimization method and system for a subway vehicle air conditioning system. It comprehensively considers the air conditioning system's energy consumption, passenger comfort performance (PMV) within the carriage, as well as the operating losses of the air conditioning equipment and the carriage load fluctuation characteristics. It introduces an air conditioning system energy consumption correction factor, a comfort penalty factor, and a carriage load fluctuation coefficient to construct a multi-objective optimization objective function. Based on currently collected real-time air conditioning operating parameters, carriage environmental parameters, external climate parameters, and carriage load data, and in conjunction with a pre-constructed digital twin model of the subway vehicle air conditioning system, it uses the optimal values ​​of the air conditioning operating parameters as decision variables and employs a gradient descent algorithm to iteratively solve the function, obtaining and distributing the optimal values ​​of the air conditioning operating parameters. This invention achieves intelligent on-demand energy supply for the air conditioning system, minimizing operating energy consumption while ensuring passenger comfort, and simultaneously realizing real-time monitoring, fault warning, and full lifecycle optimization of the air conditioning system's operating status.

Description

Technical Field

[0001] This invention belongs to the field of air conditioning control and digital twin technology for rail transit vehicles, specifically relating to an AI digital twin energy efficiency and state optimization method and system for a subway vehicle air conditioning system. Background Technology

[0002] As the core carrier of urban public transportation, the subway's operating mileage and passenger volume continue to grow. The subway vehicle air conditioning system, as a key device ensuring passenger comfort, accounts for 30%-40% of the total energy consumption of subway vehicles, making it a crucial element in subway energy conservation and emission reduction. Currently, most subway vehicle air conditioning systems adopt the traditional fixed-frequency, fixed-airflow control mode, which has the following technical shortcomings:

[0003] 1. Low control precision, unable to dynamically adjust according to real-time operating conditions: Traditional control modes only start / stop or adjust gears based on preset temperature thresholds, without fully considering factors such as external climate changes, fluctuations in passenger density in the carriage, and dynamic changes in terminal load. This results in the air conditioning system's operating status not matching actual needs, or excessive energy supply leading to energy waste, or insufficient energy supply affecting passenger comfort.

[0004] 2. High energy consumption, making it difficult to balance comfort and energy consumption: Due to the lack of accurate load forecasting and dynamic optimization mechanisms, the air conditioning system is in a state of inefficient operation for a long time, and cannot achieve on-demand water volume, on-demand air volume, and on-demand cooling and heating supply, resulting in high energy consumption. At the same time, it is difficult to take into account the cooling and heating comfort of people, and cannot achieve a dynamic optimal balance between the two.

[0005] 3. Lack of full life cycle status perception and optimization capabilities: Traditional air conditioning systems can only achieve simple operation status monitoring, and cannot accurately simulate equipment operation status, provide fault warnings, or optimize the entire life cycle. Equipment maintenance mostly adopts a periodic inspection mode, which not only has high maintenance costs, but is also prone to failure and downtime, affecting the normal operation of the subway.

[0006] Therefore, developing a method for optimizing the energy efficiency and status of subway vehicle air conditioning systems to solve the problems of high energy consumption, poor comfort, and low control precision in existing technologies has become a technical challenge that urgently needs to be addressed by those skilled in the art. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides an AI digital twin energy efficiency and state optimization method and system for subway vehicle air conditioning systems, which can effectively solve the above-mentioned problems.

[0008] The technical solution adopted in this invention is as follows:

[0009] This invention provides an AI digital twin energy efficiency and state optimization method for a subway vehicle air conditioning system, comprising the following steps:

[0010] Step S1: Construct a PMV model for passenger comfort in subway cars applicable to subway vehicle scenarios;

[0011] Step S2: Construct an energy consumption model for the air conditioning system applicable to subway vehicle scenarios;

[0012] Step S3: Based on the passenger comfort PMV model and the air conditioning system energy consumption model, and comprehensively considering the air conditioning system energy consumption, passenger comfort PMV, air conditioning equipment operating losses, and passenger load fluctuation characteristics, an air conditioning system energy consumption correction factor, a comfort penalty factor, and a passenger load fluctuation coefficient are introduced to construct a multi-objective optimization objective function. :

[0013]

[0014] in: The total energy consumption of the air conditioning system is obtained based on the aforementioned air conditioning system energy consumption model. These represent air conditioning operating parameters, cabin environment parameters, external climate parameters, and cabin load data, respectively. The passenger comfort level is obtained based on the PMV model for passenger comfort in the carriage. This is the preset optimal comfort value; This refers to the operating loss indicators of air conditioning equipment. , For compressor frequency, This refers to the fan speed. This refers to the frequency of the water pump's variable frequency drive (VFD). ; , and These are the weights for total energy consumption of the air conditioning system, passenger comfort in the carriage, and operating losses of the air conditioning equipment and fluctuations in the carriage load. This is the energy consumption correction factor for the air conditioning system, relative to the outdoor temperature. Positive correlation ; As a comfort penalty factor; ; The coefficient for fluctuation of the carriage load. ; This is the difference between the load on the carriage and the rated load of the carriage. This represents the maximum load on the carriage.

[0015] Step S4: Based on the currently collected real-time air conditioning operating parameters, carriage environment parameters, external climate parameters and carriage load data, and in conjunction with the pre-constructed digital twin model of the subway vehicle air conditioning system, the optimal value of the air conditioning operating parameters is used as the decision variable, and the gradient descent algorithm is used to iteratively solve the multi-objective optimization objective function to obtain the optimal value of the air conditioning operating parameters.

[0016] Step S5: Determine the current operating condition of the air conditioner. If it is in cooling mode, use the on-demand cooling capacity calculation model to obtain the on-demand cooling capacity; if it is in heating mode, use the on-demand heating capacity calculation model to obtain the on-demand heating capacity.

[0017] Step S6, Control command issuance and execution: The target temperature of the air conditioning system is dynamically set according to the on-demand cooling capacity or on-demand heating capacity, and the optimal value of the air conditioning operating parameters is converted into a control command and issued to the air conditioning system.

[0018] Furthermore, the air conditioner operating parameter X includes the water pump inverter frequency. Fan speed and compressor frequency The environmental parameter E of the carriage includes the average temperature inside the carriage. Relative humidity inside the carriage and the total number of people in the carriage External climate parameters W include outdoor temperature. The carriage load data L includes the real-time cooling load of the carriages. Or the real-time heat load of the carriage .

[0019] Furthermore, the PMV model for passenger comfort inside the carriage is as follows:

[0020]

[0021] in: For the comfort of passengers inside the carriage; Human metabolic rate; For human mechanical work; The total heat dissipation of a single human body is calculated using the following formula:

[0022]

[0023]

[0024] in: For the partial pressure of water vapor in the carriage, , Average temperature inside the carriage Lower saturated vapor pressure; The relative humidity inside the carriage; For the surface temperature of clothing, in a subway setting, =35.7−0.028(M−W)−0.155 (35.7− ), For the thermal resistance of clothing; The radiative heat transfer coefficient is... ; The convective heat transfer coefficient is... , The air velocity inside the carriage.

[0025] Furthermore, the energy consumption model of the air conditioning system is as follows:

[0026]

[0027] in: For compressor energy consumption, , For calibration coefficients, For compressor frequency, For the real-time cooling load of the carriage, Energy efficiency ratio (EER); For wind turbine energy consumption, , This refers to the rated power of the fan. This refers to the fan speed. This refers to the rated speed of the fan. For water pump energy consumption, , The rated power of the water pump This refers to the frequency of the water pump's variable frequency drive (VFD). To assist in energy consumption.

[0028] Furthermore, in step S5, the on-demand cooling calculation model and the on-demand heating calculation model are respectively:

[0029] On-demand cooling capacity calculation model:

[0030]

[0031] On-demand heat calculation model:

[0032]

[0033] in: and These are respectively on-demand cooling and on-demand heating; , , To correct the coefficients, calibration is performed using an AI algorithm. Outdoor temperature; This represents the total number of people in the carriage. This represents the average sensible heat dissipation per person. Real-time cooling load of the carriage; This is the real-time heat load of the carriage.

[0034] Furthermore, the construction and operation mode of the digital twin model of the subway vehicle air conditioning system is as follows:

[0035] Step A1: Construct a digital twin model of the subway vehicle air conditioning system; the digital twin model includes a physical entity layer, a virtual mapping layer, a data interaction layer, an AI algorithm layer, and a control execution layer;

[0036] Step A2, the physical entity layer is equipped with a sensor array for real-time collection of current air conditioning and environmental data, including air conditioning operating parameters X, carriage environmental parameters E, external climate parameters W, and carriage load data L;

[0037] Step A3: The data interaction layer preprocesses and standardizes the collected current air conditioning and environmental data, and then transmits it to the virtual mapping layer to drive the virtual mapping layer to perform real-time simulation, realize the mapping of the thermal environment of the carriage and the physical air conditioning system status, and perform real-time assessment of the passenger comfort PMV and the energy consumption of the air conditioning system.

[0038] Step A4: The AI ​​algorithm layer, based on the passenger comfort PMV, air conditioning system energy consumption assessment, and current air conditioning and environmental data, invokes the AI ​​energy-saving algorithm and combines historical operating data of the subway vehicle's air conditioning system to optimize the objective function for multiple objectives. The core coefficients are calibrated to obtain the calibrated multi-objective optimization objective function. ;

[0039] Step A5: The AI ​​algorithm layer invokes a preset AI optimization algorithm, and based on the currently input air conditioning and environmental data, uses a gradient descent algorithm to optimize the calibrated multi-objective objective function. The optimal values ​​of the air conditioner's operating parameters are obtained through iterative solutions; simultaneously, the on-demand cooling capacity is calculated based on the current operating conditions of the air conditioner. Or on demand calories ;

[0040] In step A6, the control execution layer converts the optimal value of the air conditioning operating parameters into control commands and sends them to the physical air conditioning system to adjust the operating parameters of the physical air conditioning system and achieve on-demand energy supply.

[0041] Furthermore, the virtual mapping layer includes a virtual geometric model, a physical model, and a behavioral model. The virtual geometric model, based on the structural parameters of the physical air conditioning system, constructs a 1:1 three-dimensional geometric model of the physical equipment, accurately reproducing the structural form and installation position of the air conditioning compressor, fan, and water pump. The physical model, based on the thermodynamic and fluid dynamic characteristics of the air conditioning system, constructs the physical equations for the operation of the air conditioning equipment, including the compressor cooling / heating equation, the fan airflow equation, and the water pump flow equation, accurately simulating the energy transfer and fluid flow processes of the air conditioning system. The behavioral model, based on historical and real-time operating data, is trained using machine learning algorithms to simulate the operating behavior of the air conditioning system under different external climates, different population densities, and different load conditions, improving the simulation accuracy of the virtual model.

[0042] Furthermore, it also includes:

[0043] Step A7, Data Feedback and Closed-Loop Optimization:

[0044] Real-time data collection of the adjusted physical air conditioning system operation is fed back to the data interaction layer to dynamically correct the digital twin model. At the same time, the feedback data is used as training samples to iteratively update the correction coefficients and weight coefficients of the AI ​​algorithm, thereby improving the optimization accuracy.

[0045] This invention also provides an AI digital twin energy efficiency and condition optimization system for a subway vehicle air conditioning system, comprising:

[0046] The PMV model building module for passenger comfort in subway cars is used to build a PMV model for passenger comfort in subway cars applicable to subway vehicle scenarios.

[0047] The air conditioning system energy consumption model building module is used to build an air conditioning system energy consumption model applicable to subway vehicle scenarios;

[0048] The multi-objective optimization objective function construction module is used to construct a multi-objective optimization objective function based on the passenger comfort PMV model and the air conditioning system energy consumption model, comprehensively considering the air conditioning system energy consumption, passenger comfort PMV, air conditioning equipment operating losses, and passenger load fluctuation characteristics. It introduces an air conditioning system energy consumption correction factor, a comfort penalty factor, and a passenger load fluctuation coefficient. :

[0049]

[0050] in: The total energy consumption of the air conditioning system is obtained based on the aforementioned air conditioning system energy consumption model. These represent air conditioning operating parameters, cabin environment parameters, external climate parameters, and cabin load data, respectively. The passenger comfort level is obtained based on the PMV model for passenger comfort in the carriage. This is the preset optimal comfort value; This refers to the operating loss indicators of air conditioning equipment. , For compressor frequency, This refers to the fan speed. This refers to the frequency of the water pump's variable frequency drive (VFD). ; , and These are the weights for total energy consumption of the air conditioning system, passenger comfort in the carriage, and operating losses of the air conditioning equipment and fluctuations in the carriage load. This is the energy consumption correction factor for the air conditioning system, relative to the outdoor temperature. Positive correlation ; As a comfort penalty factor; ; The coefficient for fluctuation of the carriage load. ; This is the difference between the load on the carriage and the rated load of the carriage. This represents the maximum load on the carriage.

[0051] The solution module is used to solve the multi-objective optimization objective function based on the currently collected real-time air conditioning operating parameters, carriage environmental parameters, external climate parameters and carriage load data, in conjunction with the pre-constructed digital twin model of the subway vehicle air conditioning system, with the optimal value of the air conditioning operating parameters as the decision variable, and using the gradient descent algorithm to iteratively solve the multi-objective optimization objective function to obtain the optimal value of the air conditioning operating parameters.

[0052] The on-demand heat calculation model is used to determine the current operating condition of the air conditioner. If it is in cooling condition, the on-demand cooling capacity calculation model is used to obtain the on-demand cooling capacity; if it is in heating condition, the on-demand heat calculation model is used to obtain the on-demand heat.

[0053] The control command issuance and execution module is used to dynamically set the target temperature of the air conditioning system according to the on-demand cooling capacity or on-demand heating capacity, and to convert the optimal value of the air conditioning operating parameters into control commands and issue them to the air conditioning system.

[0054] The AI ​​digital twin energy efficiency and state optimization method and system for subway vehicle air conditioning systems provided by this invention have the following advantages:

[0055] This invention enables intelligent on-demand energy supply for air conditioning systems, minimizing operating energy consumption while ensuring occupant comfort. It also enables real-time monitoring of the air conditioning system's operating status, fault warning, and full lifecycle optimization. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 The flowchart illustrates an AI digital twin energy efficiency and state optimization method for a subway vehicle air conditioning system provided by this invention. Detailed Implementation

[0058] To make the technical problems solved, the technical solutions, and the beneficial effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the invention.

[0059] The purpose of this invention is to overcome the shortcomings of existing subway vehicle air conditioning systems, such as low control precision, excessive energy consumption, difficulty in dynamically balancing comfort and energy consumption, and lack of full life cycle state perception and optimization capabilities. This invention provides an AI digital twin energy efficiency and state optimization method for subway vehicle air conditioning systems, enabling intelligent on-demand energy supply to the air conditioning system. While ensuring passenger comfort, it minimizes operating energy consumption and simultaneously achieves real-time monitoring, fault warning, and full life cycle optimization of the air conditioning system's operating status.

[0060] Specifically, the core objective of the scientific AI energy-saving algorithm formula system for subway vehicle air conditioning systems is to achieve on-demand cooling or heating supply by intelligently adjusting the operating status of each air conditioning device, minimizing the energy consumption of the air conditioning system while ensuring the comfort of the passengers, and achieving a dynamic optimal balance between "passenger comfort and energy consumption". The algorithm takes air conditioning operating parameters, passenger environment parameters, external climate parameters, and terminal load data as input, and finally outputs the optimal operating point of each device, the optimal system operating strategy, the on-demand cooling / heating supply scheme, and the energy-saving strategy optimization results.

[0061] This invention constructs a complete system of "comfort constraints + energy consumption model + multi-objective optimization + AI solution", which not only fits the actual situation of HVAC engineering, but also incorporates the dynamic optimization characteristics of AI algorithms.

[0062] See Figure 1 This invention provides an AI digital twin energy efficiency and state optimization method for a subway vehicle air conditioning system, comprising the following steps:

[0063] In this invention, the main parameter definitions (input parameter visualization) are shown in Table 1:

[0064] Table 1: Symbols and Physical Meanings of Input Parameters

[0065]

[0066] Step S1: Construct a PMV model for passenger comfort in subway cars applicable to subway vehicle scenarios;

[0067] The comfort constraint module uses the modified Fanger PMV model to calculate the comfort index PMV of passengers in the carriage, with the constraint condition PMV∈[-1,1].

[0068] The PMV model for passenger comfort inside the carriage is as follows:

[0069]

[0070] in: For the comfort of passengers inside the carriage; The metabolic rate is 1.2 met for subway passengers, where 1 met = 58.15 W / m². For human mechanical work (0met is taken when sitting or standing); The total heat dissipation of a single human body (including sensible heat and latent heat) is calculated using the following formula:

[0071]

[0072]

[0073] in: For the partial pressure of water vapor in the carriage, , Average temperature inside the carriage Lower saturated vapor pressure; The relative humidity inside the carriage; For the surface temperature of clothing, in a subway setting, =35.7−0.028(M−W)−0.155 (35.7− ), For the thermal resistance of clothing, 0.5clo is usually taken; The radiative heat transfer coefficient is... ; The convective heat transfer coefficient is... , The air velocity inside the carriage.

[0074] Step S2: Construct an energy consumption model for the air conditioning system applicable to subway vehicle scenarios;

[0075] The energy consumption model of the air conditioning system is as follows:

[0076]

[0077] in: For compressor energy consumption, , This is the calibration factor, equal to 0.015 kW / (Hz・kW). For compressor frequency, For the real-time cooling load of the carriage, The energy efficiency ratio (EER) is provided by the air conditioner manufacturer based on measured data. For wind turbine energy consumption, The energy consumption of a fan is directly proportional to the cube of its rotational speed. This refers to the rated power of the fan. This refers to the fan speed. This refers to the rated speed of the fan. For water pump energy consumption, The energy consumption of a water pump is directly proportional to the cube of the frequency. The rated power of the water pump is in kW, and 50 is the rated frequency in Hz. This refers to the frequency of the water pump's variable frequency drive (VFD). To assist in energy consumption, ≈0.5−1kW (valves, control systems, etc., take the average of actual measurements).

[0078] Step S3: Based on the passenger comfort PMV model and the air conditioning system energy consumption model, and comprehensively considering the air conditioning system energy consumption, passenger comfort PMV, air conditioning equipment operating losses, and passenger load fluctuation characteristics, an air conditioning system energy consumption correction factor, a comfort penalty factor, and a passenger load fluctuation coefficient are introduced to construct a multi-objective optimization objective function. .

[0079] Specifically, the multi-objective optimization module constructs a weighted single-objective optimization function, comprehensively considering the total energy consumption of the air conditioning system, comfort deviation, equipment operating losses, and load fluctuation characteristics. It introduces energy consumption correction factors, comfort penalty factors, and load fluctuation coefficients to construct a more complex and adaptable optimization function, transforming "minimizing energy consumption + optimizing comfort + minimizing equipment losses" into a weighted single objective (the weights can be dynamically adjusted).

[0080] Multi-objective optimization objective function The expression is as follows:

[0081]

[0082] in: The objective function is optimized for multiple objectives (balancing energy consumption and comfort).

[0083] The total energy consumption of the air conditioning system is obtained based on the aforementioned air conditioning system energy consumption model. These represent air conditioning operating parameters, cabin environment parameters, external climate parameters, and cabin load data, respectively.

[0084] The passenger comfort level is obtained based on the PMV model for passenger comfort in the carriage. The preset optimal comfort value, also known as the optimal predicted average vote value, is theoretically... This is the optimal comfort value;

[0085] To determine the operating loss index for air conditioning equipment, and considering dimensional uniformity, the actual fan speed is used. (Unit: r / min) Convert to Hz (1 r / min = 1 / 60 Hz) to match the compressor frequency and water pump frequency. , For compressor frequency, This refers to the fan speed. It is the frequency of the water pump, used to reduce the losses caused by high-frequency and high-load operation of the equipment and extend the service life of the equipment.

[0086] ; , and These are the weights for the total energy consumption of the air conditioning system, the comfort of passengers in the carriage, and the operating losses of the air conditioning equipment and the fluctuations in the carriage load; as one implementation method, =0.45 (energy consumption weight) =0.50 (comfort weight), (Equipment loss and load fluctuation weighting) can be fine-tuned;

[0087] This is the energy consumption correction factor for the air conditioning system, relative to the outdoor temperature. Positive correlation It is used to correct deviations in energy consumption calculations under different outdoor temperatures;

[0088] As a comfort penalty factor; The further the PMV deviates from the optimal value, the greater the penalty, thus strengthening the comfort constraint.

[0089] The coefficient for fluctuation of the carriage load. ; This is the difference between the load on the carriage and the rated load of the carriage. This represents the maximum load of the carriage, used to adapt to scenarios with dynamic load fluctuations.

[0090] Step S4: Based on the currently collected real-time air conditioning operating parameters, carriage environment parameters, external climate parameters and carriage load data, and in conjunction with the pre-constructed digital twin model of the subway vehicle air conditioning system, the optimal value of the air conditioning operating parameters is used as the decision variable, and the gradient descent algorithm is used to iteratively solve the multi-objective optimization objective function to obtain the optimal value of the air conditioning operating parameters.

[0091] In this invention, the air conditioner operating parameter X includes the water pump inverter frequency. Fan speed and compressor frequency The environmental parameter E of the carriage includes the average temperature inside the carriage. Relative humidity inside the carriage and the total number of people in the carriage External climate parameters W include outdoor temperature. The carriage load data L includes the real-time cooling load of the carriages. Or the real-time heat load of the carriage .

[0092] Specifically, the iterative solution module uses the gradient descent algorithm to find the optimal solution of the optimization function, obtaining the optimal value X* of the air conditioning operating parameters, i.e., the best operating point for each device. The iterative formula is:

[0093]

[0094] Where: η = 0.01 - 0.1 is the learning rate. Let t be the gradient of the objective function with respect to the air conditioning operating parameter X, and t be the number of iterations until the multi-objective optimization objective function is reached. Convergence or PMV entering the comfort zone.

[0095] Step S5: Determine the current operating condition of the air conditioner. If it is in cooling mode, use the on-demand cooling capacity calculation model to obtain the on-demand cooling capacity; if it is in heating mode, use the on-demand heating capacity calculation model to obtain the on-demand heating capacity.

[0096] Specifically, the core concept of on-demand cooling and heating is: based on the real-time changes in the cooling and heating load of subway cars, dynamically match the cooling and heating output of the air conditioning system to ensure that the PMV comfort index inside the car is maintained within the range of [-1,1], while avoiding energy waste caused by excessive energy supply and decreased comfort caused by insufficient energy supply, thus achieving the control requirements of "load matching and precise energy supply". Among them, precise energy supply under cooling conditions uses the optimal cooling load. As a quantitative benchmark, The optimal cooling load (unit: kW) of the air conditioning system is the optimal cooling capacity supply value calculated by a multi-objective optimization algorithm based on the real-time cooling load demand of the passenger compartment, PMV comfort constraints, and the goal of minimizing energy consumption. It serves as the core quantitative indicator for on-demand cooling. Correspondingly, precise energy supply under heating conditions is based on the optimal heat load. As a quantitative benchmark, The optimal heat load of the air conditioning system (unit: kW) is the optimal heat supply value calculated by a multi-objective optimization algorithm based on the real-time heat load demand of the passenger compartment, PMV comfort constraints, and the goal of minimizing energy consumption. As the core quantitative indicator of on-demand heating, the two ensure that the output cooling and heating capacity of the air conditioning system accurately matches the actual cooling and heating load demand of the passenger compartment, providing a clear load matching standard for subsequent algorithm solutions and parameter control.

[0097] Based on real-time load and comfort deviation, the optimal cooling / heating supply values ​​are calculated. The on-demand cooling capacity calculation model and the on-demand heating capacity calculation model are as follows:

[0098] On-demand cooling capacity (refrigeration) calculation model:

[0099]

[0100] On-demand heat (heating) calculation model:

[0101]

[0102] in: and These are respectively on-demand cooling and on-demand heating; , , The correction coefficients are calibrated using AI algorithms, with values ​​ranging from 0.1 to 0.5. The units are W / PMV, W / ℃, and dimensionless (corresponding to the dimensions of each correction term). Outdoor temperature (°C); This represents the total number of people in the carriage (unit: people), and after correction, its dimensions are consistent with those of other items. This represents the average sensible heat dissipation per person, for example, ; The real-time cooling load of the carriage (W); The real-time heat load (W) of the carriage.

[0103] In cooling mode, heat dissipation from people acts as an additional load, increasing the demand for cooling capacity. In heating mode, heat dissipation from people acts as an internal heat source, reducing the demand for heat.

[0104] The output mapping can be represented by Table 2:

[0105] Table 2: Output Result Mapping

[0106]

[0107] Step S6, Control command issuance and execution: The target temperature of the air conditioning system is dynamically set according to the on-demand cooling capacity or on-demand heating capacity, and the optimal value of the air conditioning operating parameters is converted into a control command and issued to the air conditioning system.

[0108] As one implementation method, the construction and operation of the digital twin model of the subway vehicle air conditioning system in this invention is as follows:

[0109] Step A1: Construct a digital twin model of the subway vehicle air conditioning system; the digital twin model is mapped one-to-one with the physical subway vehicle air conditioning system, and the digital twin model includes a physical entity layer, a virtual mapping layer, a data interaction layer, an AI algorithm layer, and a control execution layer;

[0110] Step A2, the physical entity layer is equipped with a sensor array for real-time collection of current air conditioning and environmental data, including air conditioning operating parameters X, carriage environmental parameters E, external climate parameters W, and carriage load data L;

[0111] The data collected by the vehicle speed sensor, personnel density sensor, and frequency sensor are uniformly collected and transmitted to this system by the vehicle TCMS system (train control and management system), while the other sensors directly collect the corresponding parameters to ensure the comprehensiveness and real-time nature of the input parameters.

[0112] The sensor array specifically includes temperature sensors, humidity sensors, speed sensors, frequency sensors, and personnel density sensors. The data collected by the vehicle speed sensors, personnel density sensors, and frequency sensors are uniformly collected and transmitted to the data interaction layer of this system by the vehicle TCMS system (Train Control and Management System). The temperature and humidity sensors directly collect their corresponding parameters. All sensors collect data at a frequency of 1-5Hz, enabling comprehensive perception of the air conditioning system's operating status, the carriage environment, external climate, and terminal load. The collected data undergoes preprocessing (including outlier removal, data smoothing, and standardization) to ensure accuracy and reliability, providing high-quality data support for subsequent optimization calculations.

[0113] Step A3: The data interaction layer preprocesses and standardizes the collected current air conditioning and environmental data, and then transmits it to the virtual mapping layer to drive the virtual mapping layer to perform real-time simulation, realize the mapping of the thermal environment of the carriage and the physical air conditioning system status, and perform real-time assessment of the passenger comfort PMV and the energy consumption of the air conditioning system.

[0114] The virtual mapping layer includes a virtual geometric model, a physical model, and a behavioral model;

[0115] The virtual geometric model is based on the structural parameters of the physical air conditioning system, and constructs a 1:1 three-dimensional geometric model of the physical equipment to accurately reproduce the structural form and installation position of the air conditioning compressor, fan, water pump, valve and other equipment.

[0116] The physical model is based on the thermodynamic and fluid dynamic characteristics of the air conditioning system and constructs the physical equations for the operation of the air conditioning equipment, including compressor cooling / heating equations, fan air volume equations, water pump flow equations, etc., to accurately simulate the energy transfer and fluid flow process of the air conditioning system.

[0117] The behavioral model is based on historical and real-time operating data and is trained using machine learning algorithms to simulate the operating behavior of the air conditioning system under different external climates, different population densities, and different load conditions, thereby improving the simulation accuracy of the virtual model.

[0118] Step A4: The AI ​​algorithm layer, based on the passenger comfort PMV, air conditioning system energy consumption assessment, and current air conditioning and environmental data, invokes the AI ​​energy-saving algorithm and combines historical operating data of the subway vehicle's air conditioning system to optimize the objective function for multiple objectives. The core coefficients are calibrated to obtain the calibrated multi-objective optimization objective function. ;

[0119] The PMV model for passenger comfort in the subway car uses a modified Fanger PMV model to calculate the PMV index for passenger comfort within the car, with the constraint that PMV ∈ [-1, 1] to ensure passenger comfort. This model is modified to account for the activity characteristics of subway car passengers (mainly sitting and standing). Verification showed that the PMV value corresponding to this range is -0.6 to 0.5, which is well within the comfortable range, improving the accuracy of comfort calculation.

[0120] Air conditioning system energy consumption model: compressor energy consumption Fan energy consumption Water pump energy consumption Energy consumption of auxiliary equipment The model incorporates unified calculations to accurately reflect the energy consumption distribution of the air conditioning system. It is tailored to the operating characteristics of the subway air conditioning system, with the energy consumption of fans and water pumps following the "cubic law" and the energy consumption of compressors linked to real-time load and energy efficiency ratio, ensuring the accuracy of energy consumption calculations.

[0121] Step A5: The AI ​​algorithm layer invokes a preset AI optimization algorithm, and based on the currently input air conditioning and environmental data, uses a gradient descent algorithm to optimize the calibrated multi-objective objective function. The optimal values ​​of the air conditioner's operating parameters are obtained through iterative solutions; simultaneously, the on-demand cooling capacity is calculated based on the current operating conditions of the air conditioner. Or on demand calories ;

[0122] In this step, a weighted single-objective optimization function is constructed, breaking through the traditional optimization logic of solely focusing on energy consumption and comfort. It comprehensively incorporates the total energy consumption of the air conditioning system, comfort deviation, equipment operating losses, and load fluctuation characteristics. An energy consumption correction factor, a comfort penalty factor, and a load fluctuation coefficient are introduced, transforming the three objectives of "minimizing energy consumption, optimizing comfort, and minimizing equipment losses" into a single optimization objective. The relationship between these three objectives is dynamically balanced through weighted coefficients. In the subway scenario, comfort is prioritized while also considering energy conservation and equipment lifespan. Therefore, the following settings are implemented: (Energy consumption weight) (Comfort weight) (Equipment loss and load fluctuation weighting). This indicates that the objective function Minimization, that is, achieving optimal energy consumption while meeting comfort constraints and reducing equipment losses, involves setting each correction factor and coefficient to fit the dynamic operating conditions of subway air conditioning, thereby improving the practicality and accuracy of the optimization function.

[0123] Iterative Solution: The gradient descent algorithm is used to iteratively solve the optimization function to obtain the optimal values ​​of the air conditioning operating parameters (water pump inverter frequency, fan speed, and compressor frequency). That is, the optimal operating point of each device; during the iteration process, it is determined in real time whether PMV has entered the comfort zone and the objective function. Whether the convergence occurs is checked to ensure the rationality and practicality of the optimization results.

[0124] Operating condition assessment (cooling / heating): Based on the on-demand cooling / heating supply formula, output an energy supply plan that adapts to the current operating condition.

[0125] In step A6, the control execution layer converts the optimal value of the air conditioning operating parameters into control commands and sends them to the physical air conditioning system to adjust the operating parameters of the physical air conditioning system and achieve on-demand energy supply.

[0126] This step involves the issuance and execution of control commands: the optimization results from multi-objective optimization are transformed into control commands and issued to the physical air conditioning system to adjust the operating status of each device and achieve on-demand water and air volume supply. The control execution layer adopts a distributed control architecture, which transforms the optimized operating parameters of each device (such as compressor frequency, fan speed, etc.) into executable control commands. Through devices such as frequency converters and controllers, the operating status of each device in the physical air conditioning system is precisely adjusted. Combined with the on-demand cooling / heating supply scheme, on-demand cooling capacity, on-demand heating capacity, on-demand water volume, and on-demand air volume supply are achieved. In cooling mode, the cooling capacity is increased according to the personnel density, and in heating mode, the heating capacity is reduced according to the personnel density to avoid over-supply or under-supply of energy.

[0127] Step A7, Data Feedback and Closed-Loop Optimization:

[0128] Real-time data collection of the adjusted physical air conditioning system operation is fed back to the data interaction layer to dynamically correct the digital twin model. At the same time, the feedback data is used as training samples to iteratively update the correction coefficients and weight coefficients of the AI ​​algorithm, thereby improving the optimization accuracy.

[0129] Specifically, real-time operational data of the physical air conditioning system after adjustments and passenger compartment comfort data are collected and fed back to the data interaction layer. The data interaction layer preprocesses and standardizes the collected input parameters before transmitting them to the virtual mapping layer, driving the digital twin model to perform real-time simulation and reproduce the operating status of the physical air conditioning system.

[0130] The data interaction layer adopts an architecture that combines edge computing and cloud collaboration. The edge is responsible for real-time data acquisition, preprocessing, and rapid transmission, while the cloud is responsible for data storage, analysis, and model optimization, ensuring the real-time performance and stability of data transmission. After receiving the data, the virtual mapping layer uses the collaborative simulation of geometric, physical, and behavioral models to reproduce the operating status of the physical air conditioning system in real time, including the operating parameters of each device, energy consumption data, and cabin environmental parameters, providing a virtual simulation platform for AI optimization.

[0131] This invention also provides an AI digital twin energy efficiency and condition optimization system for a subway vehicle air conditioning system, comprising:

[0132] The PMV model building module for passenger comfort in subway cars is used to build a PMV model for passenger comfort in subway cars applicable to subway vehicle scenarios.

[0133] The air conditioning system energy consumption model building module is used to build an air conditioning system energy consumption model applicable to subway vehicle scenarios;

[0134] The multi-objective optimization objective function construction module is used to construct a multi-objective optimization objective function based on the passenger comfort PMV model and the air conditioning system energy consumption model, comprehensively considering the air conditioning system energy consumption, passenger comfort PMV, air conditioning equipment operating losses, and passenger load fluctuation characteristics. It introduces an air conditioning system energy consumption correction factor, a comfort penalty factor, and a passenger load fluctuation coefficient. :

[0135]

[0136] in: The total energy consumption of the air conditioning system is obtained based on the aforementioned air conditioning system energy consumption model. These represent air conditioning operating parameters, cabin environment parameters, external climate parameters, and cabin load data, respectively. The passenger comfort level is obtained based on the PMV model for passenger comfort in the carriage. This is the preset optimal comfort value; This refers to the operating loss indicators of air conditioning equipment. , For compressor frequency, This refers to the fan speed. This refers to the frequency of the water pump's variable frequency drive (VFD). ; , and These are the weights for total energy consumption of the air conditioning system, passenger comfort in the carriage, and operating losses of the air conditioning equipment and fluctuations in the carriage load. This is the energy consumption correction factor for the air conditioning system, relative to the outdoor temperature. Positive correlation ; As a comfort penalty factor; ; The coefficient for fluctuation of the carriage load. ; This is the difference between the load on the carriage and the rated load of the carriage. This represents the maximum load on the carriage.

[0137] The solution module is used to solve the multi-objective optimization objective function based on the currently collected real-time air conditioning operating parameters, carriage environmental parameters, external climate parameters and carriage load data, in conjunction with the pre-constructed digital twin model of the subway vehicle air conditioning system, with the optimal value of the air conditioning operating parameters as the decision variable, and using the gradient descent algorithm to iteratively solve the multi-objective optimization objective function to obtain the optimal value of the air conditioning operating parameters.

[0138] The on-demand heat calculation model is used to determine the current operating condition of the air conditioner. If it is in cooling condition, the on-demand cooling capacity calculation model is used to obtain the on-demand cooling capacity; if it is in heating condition, the on-demand heat calculation model is used to obtain the on-demand heat.

[0139] The control command issuance and execution module is used to dynamically set the target temperature of the air conditioning system according to the on-demand cooling capacity or on-demand heating capacity, and to convert the optimal value of the air conditioning operating parameters into control commands and issue them to the air conditioning system.

[0140] The present invention will be further described in detail below with reference to specific embodiments, so as to enable those skilled in the art to clearly understand the technical solution of the present invention and to implement it:

[0141] This embodiment provides an AI digital twin energy efficiency and state optimization method for a subway car air conditioning system, applied to the air conditioning system of subway line 3 in a certain city. The air conditioning system of the subway cars on this line uses variable frequency compressors, variable frequency fans, and variable frequency water pumps, and the rated passenger capacity of the carriages is 200 people. The specific implementation steps are as follows:

[0142] S1. Construct a digital twin model of the subway vehicle air conditioning system:

[0143] Based on the structural parameters of the subway vehicle's air conditioning system (compressor model, fan power, water pump flow rate, etc.), a 1:1 three-dimensional geometric model of the physical equipment was constructed; based on thermodynamic and fluid dynamic characteristics, a physical model was constructed, in which the compressor energy efficiency ratio was determined. Fit coefficient , , , Based on one year of historical operating data of the subway air conditioning system on this line, a behavioral model was trained to simulate the operating behavior under different working conditions. The digital twin model includes a physical entity layer, a virtual mapping layer, a data interaction layer, an AI algorithm layer, and a control execution layer. Each layer works together to achieve real-time mapping between the physical entity and the virtual model.

[0144] S2. Input parameter acquisition:

[0145] Frequency sensors are installed on the compressors, fans, and water pumps of the physical air conditioning system; temperature sensors, humidity sensors, and personnel density sensors are installed inside the carriages; and a temperature sensor is installed outside the vehicle. Data collected by the vehicle speed sensor, personnel density sensor, and frequency sensor is provided to this system by the vehicle's TCMS (Train Control and Management System), with the acquisition frequency set to 2Hz. Real-time acquired input parameters (including those provided by the TCMS system) include: water pump frequency converter frequency. (20-50Hz), actual operating speed of the fan (800-1500r / min), compressor frequency (30-60Hz) Average temperature inside the carriage (22-28℃), relative humidity inside the carriage (40%-60%) Total number of passengers in the carriage (10-50 people), outdoor temperature (10-35℃), Real-time cooling load of the carriage (10-50kW); among which After being collected by personnel density sensors, the total number of people is converted into the total number of people by combining the area of ​​the carriage, ensuring that the dimensions are consistent with the on-demand energy supply model.

[0146] S3 and AI optimization solutions:

[0147] The AI ​​algorithm layer calls the AI ​​energy-saving algorithm. The specific process is as follows, with supplementary calibration examples of the core coefficients of the algorithm to ensure that the optimization process is feasible and reproducible:

[0148] S31. Algorithm Coefficient Calibration: Using one year of historical operating data of the subway air conditioning system (covering different seasons and different passenger flow periods), the coefficients were calibrated using the random forest algorithm to obtain the specific values ​​of each correction coefficient and fitting coefficient. The calibration process is as follows:

[0149] (1) Calibrating the fitting coefficient of compressor energy efficiency ratio (COP): 1000 sets of historical data (compressor frequency f, outdoor temperature T, cabin temperature T and corresponding measured COP values) were selected. The least squares method was used to fit the data, and a=0.02, a=-0.05, a=0.03 and a=2.8 were obtained. The goodness of fit R²=0.96, ensuring that the COP calculation error is ≤2%;

[0150] (2) Correction coefficient , , Calibration: 500 sets of comfort and load data under different passenger flows and outdoor temperatures were selected. With the goal of minimizing PMV deviation and optimizing energy consumption, the calibration was achieved through training with a gradient descent algorithm. , , After calibration, the calculation error for on-demand cooling / heating supply is ≤3%;

[0151] (3) Weighting coefficients , , Calibration: Based on the operational needs of the subway (prioritizing comfort while considering energy consumption and equipment lifespan), calibration was performed using the analytic hierarchy process (AHP). Ten typical operating conditions (peak passenger flow, off-peak passenger flow, extreme high temperature, and extreme low temperature) were selected for verification, and the final calibration was determined. (Energy consumption weight) (Comfort weight) (Equipment loss and load fluctuation weighting) ensures that PMV can be maintained in the comfort range under all operating conditions, while energy consumption is reduced by more than 15% compared with the traditional mode, and equipment operating loss is reduced by more than 10%.

[0152] S32. Comfort constraint calculation: The modified PMV model is adopted, M=1.2 met, W=0 met, I=0.5 clo. Combined with the working conditions of this embodiment (carriage temperature 22-28℃, relative humidity 40%-60%), the total heat dissipation of the human body is taken as a specific value of 80 W / m², and the PMV value is calculated, with PMV constrained to [-1,1].

[0153] S33. Energy consumption calculation: Based on the energy consumption model formula. , , , , Calculate the total energy consumption of the air conditioning system. ;

[0154] S34, Multi-objective optimization: Settings , , The correction factors are calculated based on the operating conditions of this embodiment. Adjust dynamically according to outdoor temperature. Dynamically adjusted according to PMV deviation. (Dynamically adjusted according to load fluctuations), construct an optimization function J to... For the goal;

[0155] Supplementary example of specific operating condition factor calculation (fitting the actual measured operating conditions of this embodiment): Select a typical peak passenger flow condition, known parameters: outdoor temperature Current PMV = 0.8, Real-time cooling load Maximum load compressor frequency Actual fan speed Water pump frequency The factors are calculated as follows:

[0156] Simultaneously, calculate on-demand cooling supply. ,in , , , , Take the total number of passengers in the carriage under this operating condition (40 people) to ensure uniformity of measurement; calculate the energy saving rate. Energy consumption of traditional fixed frequency and fixed air volume control Based on.

[0157] Comfort penalty factor : Because PMV deviates from the optimal value of 0.8, the penalty is increased to push comfort back to the optimal level;

[0158] Equipment operating loss indicators In accordance with the requirements for dimensional uniformity, the actual speed of the fan is... Convert to Hz (1400 / 60 ≈ 23.33Hz), and substitute into the corrected formula: This reflects the current level of equipment wear and tear.

[0159] To ensure the accuracy of the above factor calculation results, each step was verified. The specific verification process is as follows:

[0160] 1. Energy consumption correction factor Substitution , The calculation is correct and it fits perfectly. Definition of positive correlation with outdoor temperature;

[0161] 2. Comfort penalty factor Substitution , , The calculation is correct and conforms to the logic that "the greater the deviation from PMV, the greater the penalty."

[0162] 3. Load fluctuation coefficient : , , The calculation is correct and it is suitable for operating conditions where the load does not reach the rated value.

[0163] 4. Equipment operating loss indicators Substitute into the corrected formula, , (After conversion ≈ 23.33Hz) , , , The summation is 5069.44, and multiplying by 0.001 gives 5.06944. The calculation is correct, and the dimensions are consistent with the other items (all are dimensionless), which is in line with the definition of equipment loss index.

[0164] Based on the above calculation results, the objective function under the current working condition can be obtained by substituting them into the optimization function. The optimal balance between energy consumption, comfort, and equipment wear is achieved through iterative solutions.

[0165] S35. Iterative solution: Set the learning rate η=0.05, and iterate t until J converges (convergence condition is |JJ|<0.01), and obtain the optimal operating point X*=[f,n,f,α]* for each device;

[0166] Simultaneously, calculate on-demand cooling supply. ,in , , , ;

[0167] S4. Control Execution:

[0168] The control execution layer translates the optimized operating point into control commands, which are then sent to the inverters and controllers of the air conditioning system to adjust the water pump frequency, fan speed, and compressor frequency, thereby achieving on-demand energy supply. For example, when the passenger density in the carriage... When PMV > 1, increase fan speed and compressor frequency to increase cooling capacity (in line with the on-demand cooling formula; increased personnel density leads to increased cooling capacity). Increase cooling demand (to ensure PMV returns to the comfort range); when outdoor temperature decreases and load decreases, reduce water pump frequency and compressor frequency to reduce energy consumption.

[0169] S5, Digital Twin Model Simulation:

[0170] The data interaction layer performs outlier removal (using the 3σ criterion), data smoothing (using the moving average method), and standardization on the collected input parameters. The data is then transmitted to the virtual mapping layer to drive the digital twin model for real-time simulation, reproducing the operating state of the physical air conditioning system. The simulation error is controlled within 2%.

[0171] S6, Closed-loop optimization:

[0172] Real-time data collection of the physical air conditioning system's adjusted operating data (such as actual energy consumption and PMV value) is fed back to the data interaction layer to dynamically correct the digital twin model, keeping the simulation error within 2%. At the same time, the feedback data is used as training samples to iteratively update the correction coefficients and weight coefficients of the AI ​​algorithm, improving the optimization accuracy.

[0173] According to actual measurements, in this embodiment, the energy saving rate of the subway vehicle air conditioning system reaches 22%, the PMV value in the carriage is stable in the comfortable range of [-0.8, 0.7], the equipment failure rate is reduced by 35%, and the maintenance cost is reduced by 28%, which significantly improves the energy efficiency and operational stability of the subway air conditioning system.

[0174] In summary, the AI ​​digital twin energy efficiency and state optimization method for subway vehicle air conditioning systems provided by this invention can be implemented in one of the following ways:

[0175] S1. Construct a digital twin model of the subway vehicle air conditioning system. The digital twin model is mapped one-to-one with the physical subway vehicle air conditioning system, including a physical entity layer, a virtual mapping layer, an AI algorithm layer, a data interaction layer, and a control execution layer.

[0176] S2. Input parameters are collected in real time through the sensor array of the physical entity layer. The input parameters include air conditioning operation parameters, subway car environmental parameters, external climate parameters, and terminal load data. The sensor array includes temperature sensors, humidity sensors, speed sensors, frequency sensors, and personnel density sensors. The data collected by the vehicle speed sensor, personnel density sensor, and frequency sensor are provided to this system by the vehicle TCMS system (train control and management system). All sensors collect data at a frequency of 1-5Hz. Data preprocessing includes outlier removal, data smoothing, and standardization.

[0177] S3, the AI ​​algorithm layer calls the preset AI energy-saving algorithm, with the goal of minimizing energy consumption under comfort constraints, to optimize and solve the operating state of the digital twin model simulation, and output the optimal operating point of each device, the optimal system operating strategy, the on-demand cooling / heating supply scheme and the energy-saving strategy optimization results;

[0178] The AI ​​energy-saving algorithm includes a comfort constraint module, an energy consumption model module, a multi-objective optimization module, and an iterative solution module. The multi-objective optimization module constructs a weighted single-objective optimization function with the objectives of minimizing energy consumption and maximizing comfort. The iterative solution module uses the gradient descent algorithm to find the optimal solution of the optimization function, obtaining the optimal values ​​X* of the air conditioning operating parameters, i.e., the optimal operating points of each device. The system's optimal operating strategies include compressor frequency conversion adjustment strategies and fan speed adjustment strategies, formulating dynamic adjustment rules for different operating conditions based on the optimal operating points of each device.

[0179] S4. The control execution layer converts the optimization results output by the AI ​​algorithm layer into control commands and sends them to the physical air conditioning system to adjust the operating status of each device and achieve on-demand water and air supply.

[0180] S5 collects real-time operational data of the physical air conditioning system after adjustments and passenger compartment comfort data, feeds it back to the data interaction layer, dynamically corrects the digital twin model, realizes iterative updates of the optimization algorithm, and forms a closed-loop optimization.

[0181] S6. After preprocessing and standardizing the collected input parameters through the data interaction layer, the data is transmitted to the virtual mapping layer to drive the digital twin model to perform real-time simulation and reproduce the operating status of the physical air conditioning system.

[0182] This invention may further include a fault diagnosis submodule, which identifies abnormal operating conditions of air conditioning equipment by comparing simulation data from a digital twin model with physical operating data, and generates fault warning signals and maintenance suggestions.

[0183] Compared with the prior art, the present invention has the following advantages:

[0184] 1. Achieving the optimal dynamic balance between energy consumption and comfort: This invention takes AI energy-saving algorithm as the core, combines digital twin technology, collects multi-dimensional parameters in real time, and achieves on-demand water volume, on-demand air volume, and on-demand cooling and heating supply of air conditioning system through multi-objective optimization under comfort constraints. Under the premise of ensuring that PMV is maintained in the comfort range [-1,1], the energy consumption of air conditioning system is reduced to the maximum extent.

[0185] 2. Improve control precision and intelligence: Real-time simulation of the physical air conditioning system is achieved through digital twin model. Combined with iterative solution of AI algorithm, the optimal operating point of each device and the optimal operating strategy of the system can be accurately output. This avoids the lag and blindness of traditional control mode and realizes intelligent regulation of the entire process of air conditioning system. The control precision is improved by more than 40% compared with traditional mode.

[0186] 3. Achieve full lifecycle status optimization and predictive maintenance: The digital twin model can reproduce the operating status of the air conditioning system in real time. Combined with the fault diagnosis submodule, it can accurately identify equipment anomalies, generate fault warnings and maintenance suggestions, transform the traditional periodic maintenance mode into predictive maintenance, reduce maintenance costs, extend equipment service life, and improve the operational stability and reliability of the subway air conditioning system.

[0187] 4. Strong adaptability and scalability: The digital twin model of this invention can be adjusted according to the structural parameters of the air conditioning system of different subway vehicles. The AI ​​optimization algorithm can be continuously iterated and upgraded through data training to adapt to subway vehicles with different lines, different climate conditions and different passenger flow characteristics, which has strong practicality and promotion value.

[0188] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for optimizing the energy efficiency and state of a subway vehicle air conditioning system using an AI digital twin, characterized in that: Includes the following steps: Step S1: Construct a PMV model for passenger comfort in subway cars applicable to subway vehicle scenarios; Step S2: Construct an energy consumption model for the air conditioning system applicable to subway vehicle scenarios; Step S3: Based on the passenger comfort PMV model and the air conditioning system energy consumption model, and comprehensively considering the air conditioning system energy consumption, passenger comfort PMV, air conditioning equipment operating losses, and passenger load fluctuation characteristics, an air conditioning system energy consumption correction factor, a comfort penalty factor, and a passenger load fluctuation coefficient are introduced to construct a multi-objective optimization objective function. : ； in: The total energy consumption of the air conditioning system is obtained based on the aforementioned air conditioning system energy consumption model. These represent air conditioning operating parameters, cabin environment parameters, external climate parameters, and cabin load data, respectively. The passenger comfort level is obtained based on the PMV model for passenger comfort in the carriage. This is the preset optimal comfort value; This refers to the operating loss indicators of air conditioning equipment. , For compressor frequency, This refers to the fan speed. This refers to the frequency of the water pump's variable frequency drive (VFD). ; , and These are the weights for total energy consumption of the air conditioning system, passenger comfort in the carriage, and operating losses of the air conditioning equipment and fluctuations in the carriage load. This is the energy consumption correction factor for the air conditioning system, relative to the outdoor temperature. Positive correlation ; As a comfort penalty factor; ; The coefficient for fluctuation of the carriage load. ; This is the difference between the load on the carriage and the rated load of the carriage. This represents the maximum load on the carriage. Step S4: Based on the currently collected real-time air conditioning operating parameters, carriage environment parameters, external climate parameters and carriage load data, and in conjunction with the pre-constructed digital twin model of the subway vehicle air conditioning system, the optimal value of the air conditioning operating parameters is used as the decision variable, and the gradient descent algorithm is used to iteratively solve the multi-objective optimization objective function to obtain the optimal value of the air conditioning operating parameters. Step S5: Determine the current operating condition of the air conditioner. If it is in cooling mode, use the on-demand cooling capacity calculation model to obtain the on-demand cooling capacity; if it is in heating mode, use the on-demand heating capacity calculation model to obtain the on-demand heating capacity. Step S6, Control command issuance and execution: The target temperature of the air conditioning system is dynamically set according to the on-demand cooling capacity or on-demand heating capacity, and the optimal value of the air conditioning operating parameters is converted into a control command and issued to the air conditioning system.

2. The AI ​​digital twin energy efficiency and state optimization method for a subway vehicle air conditioning system according to claim 1, characterized in that, Air conditioner operating parameter X includes water pump inverter frequency. Fan speed and compressor frequency The environmental parameter E of the carriage includes the average temperature inside the carriage. Relative humidity inside the carriage and the total number of people in the carriage External climate parameters W include outdoor temperature. The carriage load data L includes the real-time cooling load of the carriages. Or the real-time heat load of the carriage .

3. The AI ​​digital twin energy efficiency and state optimization method for a subway vehicle air conditioning system according to claim 1, characterized in that, The PMV model for passenger comfort inside the carriage is as follows: ； in: For the comfort of passengers inside the carriage; Human metabolic rate; For human mechanical work; The total heat dissipation of a single human body is calculated using the following formula: ； ； in: For the partial pressure of water vapor in the carriage, , Average temperature inside the carriage Lower saturated vapor pressure; The relative humidity inside the carriage; For the surface temperature of clothing, in a subway setting, =35.7−0.028(M−W)−0.155 (35.7− ), For the thermal resistance of clothing; The radiative heat transfer coefficient is... ; The convective heat transfer coefficient is... , The air velocity inside the carriage.

4. The AI ​​digital twin energy efficiency and state optimization method for a subway vehicle air conditioning system according to claim 1, characterized in that, The energy consumption model of the air conditioning system is as follows: ； in: For compressor energy consumption, , For calibration coefficients, For compressor frequency, For the real-time cooling load of the carriage, Energy efficiency ratio (EER); For wind turbine energy consumption, , This refers to the rated power of the fan. This refers to the fan speed. This refers to the rated speed of the fan. For water pump energy consumption, , The rated power of the water pump This refers to the frequency of the water pump's variable frequency drive (VFD). To assist in energy consumption.

5. The AI ​​digital twin energy efficiency and state optimization method for a subway vehicle air conditioning system according to claim 1, characterized in that, In step S5, the on-demand cooling calculation model and the on-demand heating calculation model are as follows: On-demand cooling capacity calculation model: ； On-demand heat calculation model: ； in: and These are respectively on-demand cooling and on-demand heating; , , To correct the coefficients, calibration is performed using an AI algorithm. Outdoor temperature; This represents the total number of people in the carriage. This represents the average sensible heat dissipation per person. Real-time cooling load of the carriage; This is the real-time heat load of the carriage.

6. The AI ​​digital twin energy efficiency and state optimization method for a subway vehicle air conditioning system according to claim 1, characterized in that, The construction and operation mode of the digital twin model of the subway vehicle air conditioning system is as follows: Step A1: Construct a digital twin model of the subway vehicle air conditioning system; the digital twin model includes a physical entity layer, a virtual mapping layer, a data interaction layer, an AI algorithm layer, and a control execution layer; Step A2, the physical entity layer is equipped with a sensor array for real-time collection of current air conditioning and environmental data, including air conditioning operating parameters X, carriage environmental parameters E, external climate parameters W, and carriage load data L; Step A3: The data interaction layer preprocesses and standardizes the collected current air conditioning and environmental data, and then transmits it to the virtual mapping layer to drive the virtual mapping layer to perform real-time simulation, realize the mapping of the thermal environment of the carriage and the physical air conditioning system status, and perform real-time assessment of the passenger comfort PMV and the energy consumption of the air conditioning system. Step A4: The AI ​​algorithm layer, based on the passenger comfort PMV, air conditioning system energy consumption assessment, and current air conditioning and environmental data, invokes the AI ​​energy-saving algorithm and combines historical operating data of the subway vehicle's air conditioning system to optimize the objective function for multiple objectives. The core coefficients are calibrated to obtain the calibrated multi-objective optimization objective function. ; Step A5: The AI ​​algorithm layer invokes a preset AI algorithm and, based on the currently input air conditioning and environmental data, uses a gradient descent algorithm to optimize the calibrated multi-objective objective function. The optimal values ​​of the air conditioner's operating parameters are obtained through iterative solutions; simultaneously, the on-demand cooling capacity is calculated based on the current operating conditions of the air conditioner. Or on demand calories ; In step A6, the control execution layer converts the optimal value of the air conditioning operating parameters into control commands and sends them to the physical air conditioning system to adjust the operating parameters of the physical air conditioning system and achieve on-demand energy supply.

7. The AI ​​digital twin energy efficiency and state optimization method for a subway vehicle air conditioning system according to claim 6, characterized in that, The virtual mapping layer includes a virtual geometric model, a physical model, and a behavioral model. The virtual geometric model is based on the structural parameters of the physical air conditioning system, constructing a 1:1 three-dimensional geometric model of the physical equipment to accurately reproduce the structural form and installation position of the air conditioning compressor, fan, and water pump. The physical model is based on the thermodynamic and fluid dynamic characteristics of the air conditioning system, constructing the physical equations for the operation of the air conditioning equipment, including the compressor cooling / heating equation, the fan airflow equation, and the water pump flow equation, to accurately simulate the energy transfer and fluid flow process of the air conditioning system. The behavioral model is based on historical operating data and real-time data, trained through machine learning algorithms, to simulate the operating behavior of the air conditioning system under different external climates, different population densities, and different load conditions, thereby improving the simulation accuracy of the virtual model.

8. The AI ​​digital twin energy efficiency and state optimization method for a subway vehicle air conditioning system according to claim 6, characterized in that, Also includes: Step A7, Data Feedback and Closed-Loop Optimization: Real-time data collection of the adjusted physical air conditioning system operation is fed back to the data interaction layer to dynamically correct the digital twin model. At the same time, the feedback data is used as training samples to iteratively update the correction coefficients and weight coefficients of the AI ​​algorithm, thereby improving the optimization accuracy.

9. An AI digital twin energy efficiency and state optimization system for a subway vehicle air conditioning system, characterized in that, include: The PMV model building module for passenger comfort in subway cars is used to build a PMV model for passenger comfort in subway cars applicable to subway vehicle scenarios. The air conditioning system energy consumption model building module is used to build an air conditioning system energy consumption model applicable to subway vehicle scenarios; The multi-objective optimization objective function construction module is used to construct a multi-objective optimization objective function based on the passenger comfort PMV model and the air conditioning system energy consumption model, comprehensively considering the air conditioning system energy consumption, passenger comfort PMV, air conditioning equipment operating losses, and passenger load fluctuation characteristics. It introduces an air conditioning system energy consumption correction factor, a comfort penalty factor, and a passenger load fluctuation coefficient. : ； in: The total energy consumption of the air conditioning system is obtained based on the aforementioned air conditioning system energy consumption model. These represent air conditioning operating parameters, cabin environment parameters, external climate parameters, and cabin load data, respectively. The passenger comfort level is obtained based on the PMV model for passenger comfort in the carriage. This is the preset optimal comfort value; This refers to the operating loss indicators of air conditioning equipment. , For compressor frequency, This refers to the fan speed. This refers to the frequency of the water pump's variable frequency drive (VFD). ; , and These are the weights for total energy consumption of the air conditioning system, passenger comfort in the carriage, and operating losses of the air conditioning equipment and fluctuations in the carriage load. This is the energy consumption correction factor for the air conditioning system, relative to the outdoor temperature. Positive correlation ; As a comfort penalty factor; ; The coefficient for fluctuation of the carriage load. ; This is the difference between the load on the carriage and the rated load of the carriage. This represents the maximum load on the carriage. The solution module is used to solve the multi-objective optimization objective function based on the currently collected real-time air conditioning operating parameters, carriage environmental parameters, external climate parameters and carriage load data, in conjunction with the pre-constructed digital twin model of the subway vehicle air conditioning system, with the optimal value of the air conditioning operating parameters as the decision variable, and using the gradient descent algorithm to iteratively solve the multi-objective optimization objective function to obtain the optimal value of the air conditioning operating parameters. The on-demand heat calculation model is used to determine the current operating condition of the air conditioner. If it is in cooling condition, the on-demand cooling capacity calculation model is used to obtain the on-demand cooling capacity; if it is in heating condition, the on-demand heat calculation model is used to obtain the on-demand heat. The control command issuance and execution module is used to dynamically set the target temperature of the air conditioning system according to the on-demand cooling capacity or on-demand heating capacity, and to convert the optimal value of the air conditioning operating parameters into control commands and issue them to the air conditioning system.

Images