An energy consumption and carbon emission intelligent management and control method and system based on multi-source data fusion

Abstract

The application discloses an energy consumption and carbon emission intelligent management and control method and system based on multi-source data fusion, and belongs to the technical field of carbon emission analysis. The method comprises the following steps: collecting human flow density, temperature and humidity, air conditioning equipment running state and power grid carbon emission factor data in real time; calculating the human flow density and outdoor meteorological change rate to obtain a target temperature compensation value; combining the compensation value, the equipment state and the carbon emission factor to construct a comprehensive carbon emission and energy efficiency target function, solving by using an improved multi-objective particle swarm optimization algorithm, and outputting an optimal chilled water temperature set value and an optimal fan rotating speed; based on the optimal chilled water temperature set value, using a model-free adaptive control algorithm to calculate a compressor operating frequency control amount, and issuing the optimal fan rotating speed for execution. The application effectively overcomes the response lag of the heat load, realizes the double-target collaborative optimization of energy efficiency and carbon emission under the dynamic working condition, eliminates the execution deviation, and significantly reduces the invalid energy consumption and carbon emission.

Description

Technical Field

[0001] This invention relates to the field of carbon emission analysis technology, specifically to an intelligent management and control method and system for energy consumption and carbon emissions based on multi-source data fusion. Background Technology

[0002] Among various building energy-consuming facilities, the environmental temperature and humidity control system, as a key component in maintaining indoor thermal comfort, accounts for a large proportion of total energy consumption. However, the thermal environment inside a building is not static, but is influenced by the dual coupling of indoor occupant activity patterns and external climate fluctuations. This complex dynamic interaction makes the load demand of environmental control facilities highly nonlinear and time-varying, placing extremely high demands on the response speed and foresight of control strategies.

[0003] Currently, existing environmental control strategies largely rely on preset real-time thresholds for passive responses, failing to effectively capture and predict the dynamic lag characteristics of load caused by population movement and weather fluctuations. This easily leads to excessive cooling or heating, resulting in severe energy losses and increased carbon emissions. Furthermore, at the global optimization decision-making level, existing comprehensive evaluation models often over-rely on actual heat load observation data that is difficult to obtain accurately in real time. When integrating multi-dimensional indicators such as energy consumption, carbon emissions, and environmental deviations, they neglect the consistency of physical dimensions, leading to significant black-box effects and numerical distortion risks in the decision function. Finally, at the underlying execution level, conventional tracking control algorithms lack adaptability in the face of complex and variable operating conditions, struggling to overcome nonlinear deviations at the equipment execution layer, resulting in low overall system efficiency and difficulty in guaranteeing control accuracy. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for intelligent management and control of energy consumption and carbon emissions based on multi-source data fusion, so as to solve the problems mentioned in the background art.

[0005] To address the aforementioned technical problems, this invention provides a method for intelligent management and control of energy consumption and carbon emissions based on multi-source data fusion, comprising:

[0006] The infrared array sensors deployed at the entrances and exits of the area collect the sequence data of pedestrian flow density within the target control area at a set sampling frequency.

[0007] Human density sequence data refers to a data sequence of continuously collected data on the change of the number of people in a target control area over time, used to characterize the real-time changes of indoor human body heat dissipation, a dynamic heat source.

[0008] Indoor and outdoor temperature and humidity sequence data of the target control area are collected synchronously by temperature and humidity sensors deployed indoors and outdoors.

[0009] Indoor and outdoor temperature and humidity series data refers to the data series of temperature and relative humidity changes of indoor and outdoor environments over time. It is used to reflect the differences between indoor and outdoor thermal and humidity environments and the baseline state of heat transfer load of building envelope.

[0010] Collect operational status data of air conditioning equipment within the target control area. This data includes the operating frequency of the compressor and blower, as well as the chilled water temperature.

[0011] The operating frequency of the compressor and the blower refers to the time series data of the AC frequency currently output by the air conditioner inverter. This parameter directly determines the cooling / heating output power of the compressor and the air volume of the blower, and is a core indicator reflecting the actual energy consumption status of the air conditioning equipment.

[0012] Real-time carbon emission factor sequence data of the power grid is obtained from the power grid dispatch center through a communication interface, and the real-time carbon emission factor sequence data of the power grid is stored in a time-series database aligned with timestamps. Specifically, this includes:

[0013] Each source data is accompanied by a uniform absolute timestamp identifier.

[0014] Since different data sources have different sampling frequencies or communication delays, the data source with the highest sampling frequency is used as the reference time axis. The low-frequency acquired power grid carbon emission factor data is resampled using forward padding or linear interpolation algorithms so that all sequences have corresponding parameter values ​​at the same time node.

[0015] Multidimensional data with uniformly formatted timestamps are batch-written into the time-series database as structured entries of "timestamp-parameter key-parameter value". This ensures strict alignment of data across all dimensions on the time slices, avoiding computational distortions caused by time misalignment during multi-source data fusion.

[0016] The rate of change of pedestrian density is obtained by performing first-order difference calculation on the pedestrian density sequence data. Specifically:

[0017] Let the current pedestrian density sequence data be P1 and the previous data be P0. Then the current pedestrian density change rate is: P1 - P0.

[0018] This first-order difference operation extracts the instantaneous incremental trend of population flow, thus removing the influence of the static population base.

[0019] The outdoor meteorological change rate is obtained by calculating the slope of the outdoor temperature series in the indoor and outdoor temperature and humidity series data. Specifically:

[0020] Set a time window containing a fixed number of historical sampling points, extract the outdoor temperature at the last moment of the window and the outdoor temperature at the beginning moment, calculate the difference and divide it by the window time span, and use the average slope of this difference as the rate of change of outdoor weather at the current moment.

[0021] Compared to single-step difference, the sliding window slope can effectively filter out high-frequency meteorological noise and extract the short-to-medium-term evolution trend of outdoor temperature.

[0022] The target temperature compensation value was calculated based on the rate of change in pedestrian density and the rate of change in outdoor weather conditions. The formula is as follows:

[0023] ;

[0024] In the formula, The preset thermal conversion coefficient for population density is expressed in degrees Celsius per person per sampling period. It is used to convert changes in population density into corresponding temperature change equivalents.

[0025] The outdoor temperature heat conversion coefficient is a preset value, dimensionless; it is used to convert changes in outdoor temperature into the temperature equivalent of their impact on indoor temperatures.

[0026] For the first The rate of change of pedestrian density at each sampling point, expressed in units of people per sampling cycle, reflects the potential thermal disturbance caused by pedestrian movement.

[0027] For the first The outdoor weather change rate at each sampling point, expressed in degrees Celsius per sampling period, reflects the thermal disturbance caused by outdoor climate fluctuations.

[0028] The preset heat load response attenuation coefficient determines the rate at which the influence of historical changes on the current compensation value decays, reflecting the lag characteristics of building heat load.

[0029] The sampling point number corresponding to the current time. This refers to the serial number of the historical sampling point; This represents the number of sampling points included in the time series window. It is a natural constant.

[0030] By integrating historical rates of change in pedestrian density and outdoor weather patterns, and using an exponential decay function to weight the changes at different historical moments, a feedforward compensation value with temperature dimensions is calculated.

[0031] This target temperature compensation value can reflect the dynamic trend of future indoor heat load changes in advance, thereby overcoming the dynamic lag of the air conditioning system in responding to changes in heat load.

[0032] With the goals of minimizing air conditioning system energy consumption and minimizing carbon emissions, combined with target temperature compensation values... A comprehensive carbon emission and energy efficiency objective function is constructed using air conditioning equipment operating status data and real-time carbon emission factor sequence data from the power grid.

[0033] ;

[0034] In the formula, The value of the comprehensive objective function is used to evaluate the overall control effect under the current combination of decision variables; the smaller the value, the better.

[0035] , , These are preset weighting coefficients, used to balance the importance of energy consumption, carbon emissions, and temperature deviation in the overall objective.

[0036] The predicted step size represents the number of steps within the time span considered for future optimization. For the first The energy consumption value of the compressor within a predicted step size. For the first The energy consumption value of the blower within a predicted step size.

[0037] This is a preset baseline energy consumption value used to normalize energy consumption items into dimensionless proportional values. For the first The real-time carbon emission factor of the power grid, with a predicted step size, reflects the carbon emission intensity per unit of electricity generated by the power grid at that moment.

[0038] This is a preset baseline carbon emission value used to normalize carbon emission items. The preset baseline indoor temperature represents the desired comfort level.

[0039] This represents the actual indoor temperature, reflecting the current real-time indoor thermal environment. This is a preset reference temperature deviation value, used to normalize the absolute temperature deviation into a relative proportional value.

[0040] By introducing a baseline value, the energy consumption, carbon emissions, and temperature deviation within the future prediction step are normalized to dimensionless values ​​and then summed using weighted averages.

[0041] The integrated carbon emission and energy efficiency objective function unifies multi-objective problems with different physical dimensions into a single function, avoiding black-box approach and providing a clear fitness evaluation standard for optimization algorithms.

[0042] Establish an explicit polynomial mathematical mapping relationship between the two decision variables, chilled water temperature setpoint and blower speed, and compressor energy consumption.

[0043] With this explicit model, the optimization algorithm can directly predict equipment energy consumption based on decision variables, avoiding the black box problem caused by unmeasurable variables.

[0044] Specifically, this involves establishing an energy consumption model for air conditioning equipment:

[0045] ;

[0046] ;

[0047] In the formula, For the first The setpoint for the chilled water temperature of each prediction step is one of the decision variables for the optimization algorithm. For the first The fan speed of each prediction step is another decision variable for the optimization algorithm.

[0048] The preset equipment energy consumption fitting coefficient is obtained by fitting historical operating data and reflects the inherent energy consumption characteristics of a specific air conditioning device.

[0049] Initialize the particle population of the improved multi-objective particle swarm optimization algorithm, where each particle represents a set of decision variables including chilled water temperature setpoint and blower speed.

[0050] Calculate the fitness value of each particle in the current iteration, and update the particle velocity and position by introducing adaptive inertia weights to generate a new generation of particle population. Specifically:

[0051] First, the new velocity vector of each particle in the current iteration is calculated based on the particle velocity update formula.

[0052] Subsequently, the new velocity vector is superimposed on the current position according to the position update formula to obtain the new position coordinates of each particle.

[0053] Next, a boundary constraint check is performed on the new position. If the particle exceeds the preset upper and lower limits of the decision variable, it is forcibly pulled back to the boundary value or subjected to reflection processing.

[0054] Finally, all the updated particles that meet the boundary conditions are recombined to form a new generation of particle population.

[0055] Guided by adaptive inertia weights and random numbers, this process enables the population to continuously explore and develop within the solution space, gradually approaching the Pareto optimal frontier, thus providing a continuously evolving solution space foundation for optimizing the comprehensive objective function.

[0056] The update formula is as follows:

[0057] ;

[0058] ;

[0059] In the formula, For the first The particle in the first The speed of each iteration indicates the direction and step size of the next move. For the first The particle in the first The speed of the next iteration represents the inertial direction of the previous step.

[0060] For the first The particle in the first The position of the next iteration represents the value of the current set of decision variables (chilled water temperature and fan speed).

[0061] The adaptive inertia weights decrease linearly with the number of iterations, being larger in the early stages to enhance global search and smaller in the later stages to enhance local exploration.

[0062] and The learning factor controls the step size ratio at which a particle moves closer to its historical best position and its global best position, respectively. and The random number between [0,1] is used to increase the randomness of the algorithm's search and help it escape local optima.

[0063] For the first The individual historical best position of each particle records the optimal combination of decision variables that the particle has found. This represents the global historical optimal position, recording the optimal combination of decision variables found so far for the entire particle swarm.

[0064] By combining adaptive inertia weights, individual cognitive terms, and social cognitive terms, the speed and direction of the particle's next movement in the solution space are calculated, thereby guiding the population to achieve a balance between global search and local development, and efficiently finding the optimal chilled water temperature and fan speed.

[0065] When the maximum number of iterations is reached, the set of non-dominated solutions in the external archive is output, and a fuzzy decision method is used to select a compromise solution from the set of non-dominated solutions as the optimal chilled water temperature setpoint and the optimal fan speed.

[0066] The set of non-dominated solutions in the external archive refers to the set of non-dominated particle positions selected through Pareto dominance relations during the multi-objective optimization iteration process.

[0067] Specifically, for any solution in the archive, there is no other solution in the current population and the archive that can simultaneously outperform it in the three objectives of energy consumption, carbon emissions, and temperature deviation.

[0068] The role of external archives is to preserve all globally optimal compromise solutions discovered during the iteration process, prevent excellent solutions from being discarded during iteration, provide candidate space for selecting compromise solutions that take into account both energy efficiency and carbon emissions using fuzzy decision-making methods, and ensure that the final output control setpoint has multi-objective cooperative optimality.

[0069] The system collects the actual chilled water temperature of the air conditioning equipment and calculates the system output error between the optimal chilled water temperature setpoint and the actual chilled water temperature. The system output error directly reflects the degree of dynamic deviation of the lower-level actuators when tracking the upper-level optimization decisions.

[0070] The system output error serves as the core driving signal for the model-free adaptive control algorithm.

[0071] On the one hand, its absolute value is input into the dynamic step size factor calculation formula, which is used to dynamically adjust the convergence step size of the control algorithm.

[0072] On the other hand, this error participates in the online estimation of pseudo-partial derivatives and the recursive calculation of control quantities. Through a closed-loop negative feedback mechanism, the operating frequency control quantity of the compressor is continuously corrected until the error is eliminated, thereby ensuring the accuracy of the underlying execution tracking.

[0073] Based on the magnitude of the error between the set reference value and the actual output value, the control step size is dynamically adjusted nonlinearly using an exponential function. When the error is large, the step size is increased to accelerate the convergence speed; when the error is small, the step size is decreased to suppress overshoot, thereby improving tracking accuracy and stability under varying operating conditions.

[0074] Specifically, a model-free adaptive control algorithm is adopted, and the control input is dynamically updated based on the system output error using an online pseudo-partial derivative estimation method, and the dynamic step size factor is calculated:

[0075] ;

[0076] In the formula, For the first The dynamic step size factor at any given time directly participates in the calculation of the control quantity and dynamically affects the control strength. The initial step size factor is the baseline value, which limits the maximum adjustment range of the dynamic step size.

[0077] The error sensitivity coefficient determines the sensitivity of the step size factor to changes in error and the steepness of the response curve.

[0078] For the first The reference value set at any given time is the optimal chilled water temperature setpoint issued by the optimization algorithm. For the first The actual chilled water temperature at any given moment is collected and fed back in real time by sensors.

[0079] In the absence of a precise mathematical model of the air conditioning system, the pseudo-partial derivatives of the system are estimated in real time online using only historical input and output data and a recursive algorithm with a penalty term. This reflects the local dynamic sensitivity of the system and provides a core basis for calculating the actual control quantity.

[0080] Calculate the pseudo-partial derivative estimate for the next time step:

[0081] ;

[0082] In the formula, For the first The pseudo-partial derivative estimate at time t reflects the local gradient or dynamic sensitivity of the system at that time. For the first The change in control input at any given time is the change in compressor frequency applied at the previous time.

[0083] ;

[0084] In the formula, For the first The change in system output at any given time reflects the temperature difference change in the actual system response. The preset weighting factors control the convergence speed and stability of gradient estimation.

[0085] , These are preset penalty factors used to prevent the denominator from being zero and to limit excessive drift of the estimated value when the input changes drastically. Calculate the compressor's operating frequency control value for the next moment:

[0086] ;

[0087] In the formula, For the first The compressor operating frequency control value at any given time.

[0088] When there is no previous reference value in the initial value case, assign reasonable preset initial constants with physical meaning to the relevant recursive parameters.

[0089] Specifically, the initial control input change is set. Initial system output change .

[0090] For the pseudo-partial derivative estimate, set its initial value. (in (It is a small, non-zero positive real number, calibrated based on the system's historical operating experience).

[0091] Initial control quantity Set to the current factory default start frequency of the air conditioner compressor or the steady-state frequency of the previous control cycle.

[0092] By assigning initial values ​​as described above, the recursive algorithm is guaranteed to have a definite numerical basis and stable mathematical boundaries when it is executed for the first time. Subsequently, it can automatically iterate and correct the pseudo-partial derivatives and control quantities based on the actual input and output data, smoothly transition to the normal adaptive control state, and avoid computational collapse or control divergence caused by missing initial values.

[0093] The optimal fan speed is used as the operating frequency control quantity of the blower. The operating frequency control quantities of the compressor and blower are then converted into digital communication commands and sent to the air conditioning equipment for execution, thereby achieving synchronous management of energy consumption and carbon emissions.

[0094] The present invention also provides an intelligent energy consumption and carbon emission control system based on multi-source data fusion, including a data acquisition and sensing module, a dynamic load calculation module, an optimization decision-making module, and a closed-loop execution module.

[0095] The data acquisition and sensing module is used to collect real-time data on population density, indoor and outdoor temperature and humidity, air conditioning equipment operation status, and real-time carbon emission factor data of the power grid within the target control area.

[0096] The dynamic load calculation module is used to calculate the rate of change of pedestrian density based on pedestrian density sequence data, calculate the rate of change of outdoor weather based on indoor and outdoor temperature and humidity sequence data, and calculate the target temperature compensation value based on the rate of change of pedestrian density and the rate of change of outdoor weather.

[0097] The optimization decision module is used to construct a comprehensive carbon emission and energy efficiency objective function by combining the target temperature compensation value, air conditioning equipment operating status data and real-time carbon emission factor sequence data of the power grid. An improved multi-objective particle swarm optimization algorithm is used to solve the comprehensive carbon emission and energy efficiency objective function, and output the optimal chilled water temperature setpoint and the optimal fan speed.

[0098] The closed-loop execution module is used to send the optimal chilled water temperature setpoint as a set reference value to the air conditioning controller, and uses a model-free adaptive control algorithm to calculate the compressor operating frequency control quantity to eliminate execution deviation. At the same time, it sends the optimal fan speed directly to the blower inverter for execution.

[0099] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0100] Existing technologies mostly employ passive control based on real-time thresholds, which struggles to address load lag caused by changes in population and weather. This technical solution extracts the rate of change in population density and the rate of change in outdoor weather, introduces a heat conversion coefficient to unify both into a target temperature compensation value with temperature dimensions, and performs feedforward compensation, effectively overcoming the problem of advanced or delayed regulation caused by lag in heat load response.

[0101] Existing integrated objective functions rely on actual heat load data that cannot be measured in real time and have inconsistent physical dimensions. This technical solution constructs an objective function that includes an explicit equipment energy consumption polynomial model, and introduces benchmark values ​​for energy consumption, carbon emissions, and temperature deviation, normalizes them to dimensionless values, and then performs a weighted summation. By unifying dimensions and normalizing the processing, the black-box nature of multi-objective functions and the curse of numerical values ​​are avoided.

[0102] Existing underlying control algorithms lack sufficient tracking accuracy under varying operating conditions. This technical solution introduces a step size factor that dynamically adjusts with error into the model-free adaptive control algorithm to calculate the compressor's operating frequency control quantity. This achieves the dual-objective optimal balance between energy efficiency and carbon emissions in the air conditioning system under dynamic operating conditions, eliminates nonlinear deviations at the equipment execution layer, significantly reduces ineffective energy consumption and carbon emissions, and ensures control accuracy. Attached Figure Description

[0103] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0104] Figure 1 This is a flowchart illustrating an intelligent energy consumption and carbon emission control method based on multi-source data fusion according to the present invention.

[0105] Figure 2 This is a schematic diagram comparing the cumulative carbon emissions of the present invention and existing technologies over 24 hours.

[0106] Figure 3 This is a schematic diagram comparing the deviation curves of indoor temperature control between the present invention and existing technologies. Detailed Implementation

[0107] 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.

[0108] Please see Figure 1 This invention provides a method for intelligent management and control of energy consumption and carbon emissions based on multi-source data fusion, comprising:

[0109] The infrared array sensors deployed at the entrances and exits of the area collect the sequence data of pedestrian flow density within the target control area at a set sampling frequency.

[0110] Human density sequence data refers to a data sequence of continuously collected data on the change of the number of people in a target control area over time, used to characterize the real-time changes of indoor human body heat dissipation, a dynamic heat source.

[0111] Indoor and outdoor temperature and humidity sequence data of the target control area are collected synchronously by temperature and humidity sensors deployed indoors and outdoors.

[0112] Indoor and outdoor temperature and humidity series data refers to the data series of temperature and relative humidity changes of indoor and outdoor environments over time. It is used to reflect the differences between indoor and outdoor thermal and humidity environments and the baseline state of heat transfer load of building envelope.

[0113] Collect operational status data of air conditioning equipment within the target control area. This data includes the operating frequency of the compressor and blower, as well as the chilled water temperature.

[0114] The operating frequency of the compressor and the blower refers to the time series data of the AC frequency currently output by the air conditioner inverter. This parameter directly determines the cooling / heating output power of the compressor and the air volume of the blower, and is a core indicator reflecting the actual energy consumption status of the air conditioning equipment.

[0115] Real-time carbon emission factor sequence data of the power grid is obtained from the power grid dispatch center through a communication interface, and the real-time carbon emission factor sequence data of the power grid is stored in a time-series database aligned with timestamps. Specifically, this includes:

[0116] Each source data is accompanied by a uniform absolute timestamp identifier.

[0117] Since different data sources have different sampling frequencies or communication delays, the data source with the highest sampling frequency is used as the reference time axis. The low-frequency acquired power grid carbon emission factor data is resampled using forward padding or linear interpolation algorithms so that all sequences have corresponding parameter values ​​at the same time node.

[0118] Multidimensional data with uniformly formatted timestamps are batch-written into a time-series database (such as InfluxDB) as structured entries of "timestamp-parameter key-parameter value". Strict alignment of data across dimensions across time slices is ensured to avoid computational distortions caused by time misalignment during multi-source data fusion.

[0119] The rate of change of pedestrian density is obtained by performing first-order difference calculation on the pedestrian density sequence data. Specifically:

[0120] Let the current pedestrian density sequence data be P1 and the previous data be P0. Then the current pedestrian density change rate is: P1 - P0.

[0121] This first-order difference operation extracts the instantaneous incremental trend of population flow, thus removing the influence of the static population base.

[0122] The outdoor meteorological change rate is obtained by calculating the slope of the outdoor temperature series in the indoor and outdoor temperature and humidity series data. Specifically:

[0123] Set a time window containing a fixed number of historical sampling points (e.g., the past 3 sampling periods), extract the outdoor temperature at the last moment of the window and the outdoor temperature at the beginning moment, calculate the difference and divide it by the window time span, and use the average slope of this difference as the rate of change of outdoor weather at the current moment.

[0124] Compared to single-step difference, the sliding window slope can effectively filter out high-frequency meteorological noise and extract the short-to-medium-term evolution trend of outdoor temperature.

[0125] The target temperature compensation value was calculated based on the rate of change in pedestrian density and the rate of change in outdoor weather conditions. The formula is as follows:

[0126] ;

[0127] In the formula, The preset thermal conversion coefficient for population density is expressed in degrees Celsius per person per sampling period. It is used to convert changes in population density into corresponding temperature change equivalents.

[0128] The outdoor temperature heat conversion coefficient is a preset value, dimensionless; it is used to convert changes in outdoor temperature into the temperature equivalent of their impact on indoor temperatures.

[0129] For the first The rate of change of pedestrian density at each sampling point, expressed in units of people per sampling cycle, reflects the potential thermal disturbance caused by pedestrian movement.

[0130] For the first The outdoor weather change rate at each sampling point, expressed in degrees Celsius per sampling period, reflects the thermal disturbance caused by outdoor climate fluctuations.

[0131] The preset heat load response attenuation coefficient determines the rate at which the influence of historical changes on the current compensation value decays, reflecting the lag characteristics of building heat load.

[0132] The sampling point number corresponding to the current time. This refers to the serial number of the historical sampling point; This represents the number of sampling points included in the time series window. It is a natural constant.

[0133] By integrating historical rates of change in pedestrian density and outdoor weather patterns, and using an exponential decay function to weight the changes at different historical moments, a feedforward compensation value with temperature dimensions is calculated.

[0134] This target temperature compensation value can reflect the dynamic trend of future indoor heat load changes in advance, thereby overcoming the dynamic lag of the air conditioning system in responding to changes in heat load.

[0135] With the goals of minimizing air conditioning system energy consumption and minimizing carbon emissions, combined with target temperature compensation values... A comprehensive carbon emission and energy efficiency objective function is constructed using air conditioning equipment operating status data and real-time carbon emission factor sequence data from the power grid.

[0136] ;

[0137] In the formula, The value of the comprehensive objective function is used to evaluate the overall control effect under the current combination of decision variables; the smaller the value, the better.

[0138] , , These are preset weighting coefficients, used to balance the importance of energy consumption, carbon emissions, and temperature deviation in the overall objective.

[0139] The predicted step size represents the number of steps within the time span considered for future optimization. For the first The energy consumption value of the compressor within a predicted step size. For the first The energy consumption value of the blower within a predicted step size.

[0140] This is a preset baseline energy consumption value used to normalize energy consumption items into dimensionless proportional values. For the first The real-time carbon emission factor of the power grid, with a predicted step size, reflects the carbon emission intensity per unit of electricity generated by the power grid at that moment.

[0141] This is a preset baseline carbon emission value used to normalize carbon emission items. The preset baseline indoor temperature represents the desired comfort level.

[0142] This represents the actual indoor temperature, reflecting the current real-time indoor thermal environment. This is a preset reference temperature deviation value, used to normalize the absolute temperature deviation into a relative proportional value.

[0143] By introducing a baseline value, the energy consumption, carbon emissions, and temperature deviation within the future prediction step are normalized to dimensionless values ​​and then summed using weighted averages.

[0144] The integrated carbon emission and energy efficiency objective function unifies multi-objective problems with different physical dimensions into a single function, avoiding black-box approach and providing a clear fitness evaluation standard for optimization algorithms.

[0145] Establish an explicit polynomial mathematical mapping relationship between the two decision variables, chilled water temperature setpoint and blower speed, and compressor energy consumption.

[0146] With this explicit model, the optimization algorithm can directly predict equipment energy consumption based on decision variables, avoiding the black box problem caused by unmeasurable variables.

[0147] Specifically, this involves establishing an energy consumption model for air conditioning equipment:

[0148] ;

[0149] ;

[0150] In the formula, For the first The setpoint for the chilled water temperature of each prediction step is one of the decision variables for the optimization algorithm. For the first The fan speed of each prediction step is another decision variable for the optimization algorithm.

[0151] The preset equipment energy consumption fitting coefficient is obtained by fitting historical operating data and reflects the inherent energy consumption characteristics of a specific air conditioning device.

[0152] Initialize the particle population of the improved multi-objective particle swarm optimization algorithm, where each particle represents a set of decision variables including chilled water temperature setpoint and blower speed.

[0153] Calculate the fitness value of each particle in the current iteration, and update the particle velocity and position by introducing adaptive inertia weights to generate a new generation of particle population. Specifically:

[0154] First, the new velocity vector of each particle in the current iteration is calculated based on the particle velocity update formula.

[0155] Subsequently, the new velocity vector is superimposed on the current position according to the position update formula to obtain the new position coordinates of each particle.

[0156] Next, a boundary constraint check is performed on the new position. If the particle exceeds the preset upper and lower limits of the decision variables (chilled water temperature setpoint and blower speed), it is forcibly pulled back to the boundary value or subjected to reflection processing.

[0157] Finally, all the updated particles that meet the boundary conditions are recombined to form a new generation of particle population.

[0158] Guided by adaptive inertia weights and random numbers, this process enables the population to continuously explore and develop within the solution space, gradually approaching the Pareto optimal frontier, thus providing a continuously evolving solution space foundation for optimizing the comprehensive objective function.

[0159] The update formula is as follows:

[0160] ;

[0161] ;

[0162] In the formula, For the first The particle in the first The speed of each iteration indicates the direction and step size of the next move. For the first The particle in the first The speed of the next iteration represents the inertial direction of the previous step.

[0163] For the first The particle in the first The position of the next iteration represents the value of the current set of decision variables (chilled water temperature and fan speed).

[0164] The adaptive inertia weights decrease linearly with the number of iterations, being larger in the early stages to enhance global search and smaller in the later stages to enhance local exploration.

[0165] and The learning factor controls the step size ratio at which a particle moves closer to its historical best position and its global best position, respectively. and The random number between [0,1] is used to increase the randomness of the algorithm's search and help it escape local optima.

[0166] For the first The individual historical best position of each particle records the optimal combination of decision variables that the particle has found. This represents the global historical optimal position, recording the optimal combination of decision variables found so far for the entire particle swarm.

[0167] By combining adaptive inertia weights, individual cognitive terms, and social cognitive terms, the speed and direction of the particle's next movement in the solution space are calculated, thereby guiding the population to achieve a balance between global search and local development, and efficiently finding the optimal chilled water temperature and fan speed.

[0168] When the maximum number of iterations is reached, the set of non-dominated solutions in the external archive is output, and a fuzzy decision method is used to select a compromise solution from the set of non-dominated solutions as the optimal chilled water temperature setpoint and the optimal fan speed.

[0169] The set of non-dominated solutions in the external archive refers to the set of non-dominated particle positions selected through Pareto dominance relations during the multi-objective optimization iteration process.

[0170] Specifically, for any solution in the archive (i.e. a set of chilled water temperature setpoints and blower speed), there is no other solution in the current population and archive that can simultaneously outperform it in the three objectives of energy consumption, carbon emissions, and temperature deviation.

[0171] The role of external archives is to preserve all globally optimal compromise solutions discovered during the iteration process, prevent excellent solutions from being discarded during iteration, provide candidate space for selecting compromise solutions that take into account both energy efficiency and carbon emissions using fuzzy decision-making methods, and ensure that the final output control setpoint has multi-objective cooperative optimality.

[0172] The system collects the actual chilled water temperature of the air conditioning equipment and calculates the system output error between the optimal chilled water temperature setpoint and the actual chilled water temperature. The system output error directly reflects the degree of dynamic deviation of the lower-level actuators when tracking the upper-level optimization decisions.

[0173] The system output error serves as the core driving signal for the model-free adaptive control algorithm.

[0174] On the one hand, its absolute value is input into the dynamic step size factor calculation formula, which is used to dynamically adjust the convergence step size of the control algorithm.

[0175] On the other hand, this error participates in the online estimation of pseudo-partial derivatives and the recursive calculation of control quantities. Through a closed-loop negative feedback mechanism, the operating frequency control quantity of the compressor is continuously corrected until the error is eliminated, thereby ensuring the accuracy of the underlying execution tracking.

[0176] Based on the magnitude of the error between the set reference value and the actual output value, the control step size is dynamically adjusted nonlinearly using an exponential function. When the error is large, the step size is increased to accelerate the convergence speed; when the error is small, the step size is decreased to suppress overshoot, thereby improving tracking accuracy and stability under varying operating conditions.

[0177] Specifically, a model-free adaptive control algorithm is adopted, and the control input is dynamically updated based on the system output error using an online pseudo-partial derivative estimation method, and the dynamic step size factor is calculated:

[0178] ;

[0179] In the formula, For the first The dynamic step size factor at any given time directly participates in the calculation of the control quantity and dynamically affects the control strength. The initial step size factor is the baseline value, which limits the maximum adjustment range of the dynamic step size.

[0180] The error sensitivity coefficient determines the sensitivity of the step size factor to changes in error and the steepness of the response curve.

[0181] For the first The reference value set at any given time is the optimal chilled water temperature setpoint issued by the optimization algorithm. For the first The actual chilled water temperature at any given moment is collected and fed back in real time by sensors.

[0182] In the absence of a precise mathematical model of the air conditioning system, the pseudo-partial derivatives of the system are estimated in real time online using only historical input (control quantity) and output (temperature) data and a recursive algorithm with a penalty term. This reflects the local dynamic sensitivity of the system and provides a core basis for calculating the actual control quantity.

[0183] Calculate the pseudo-partial derivative estimate for the next time step:

[0184] ;

[0185] In the formula, For the first The pseudo-partial derivative estimate at time t reflects the local gradient or dynamic sensitivity of the system at that time. For the first The change in control input at any given time is the change in compressor frequency applied at the previous time.

[0186] ;

[0187] In the formula, For the first The change in system output at any given time reflects the temperature difference change in the actual system response. The preset weighting factors control the convergence speed and stability of gradient estimation.

[0188] , These are preset penalty factors used to prevent the denominator from being zero and to limit excessive drift of the estimated value when the input changes drastically. Calculate the compressor's operating frequency control value for the next moment:

[0189] ;

[0190] In the formula, For the first The compressor operating frequency control value at any given time.

[0191] In the case of initial values ​​(i.e.) When there is no previous reference value at any time or during a system cold start, assign reasonable preset initial constants with physical meaning to the relevant recursive parameters.

[0192] Specifically, the initial control input change is set. Initial system output change .

[0193] For the pseudo-partial derivative estimate, set its initial value. (in (It is a small, non-zero positive real number, calibrated based on the system's historical operating experience).

[0194] Initial control quantity Set to the current factory default start frequency of the air conditioner compressor or the steady-state frequency of the previous control cycle.

[0195] By assigning initial values ​​as described above, the recursive algorithm is guaranteed to have a definite numerical basis and stable mathematical boundaries when it is executed for the first time. Subsequently, it can automatically iterate and correct the pseudo-partial derivatives and control quantities based on the actual input and output data, smoothly transition to the normal adaptive control state, and avoid computational collapse or control divergence caused by missing initial values.

[0196] The optimal fan speed is used as the operating frequency control quantity of the blower. The operating frequency control quantities of the compressor and blower are then converted into digital communication commands and sent to the air conditioning equipment for execution, thereby achieving synchronous management of energy consumption and carbon emissions.

[0197] The present invention also provides an intelligent energy consumption and carbon emission control system based on multi-source data fusion, including a data acquisition and sensing module, a dynamic load calculation module, an optimization decision-making module, and a closed-loop execution module.

[0198] The data acquisition and sensing module is used to collect real-time data on population density, indoor and outdoor temperature and humidity, air conditioning equipment operation status, and real-time carbon emission factor data of the power grid within the target control area.

[0199] The dynamic load calculation module is used to calculate the rate of change of pedestrian density based on pedestrian density sequence data, calculate the rate of change of outdoor weather based on indoor and outdoor temperature and humidity sequence data, and calculate the target temperature compensation value based on the rate of change of pedestrian density and the rate of change of outdoor weather.

[0200] The optimization decision module is used to construct a comprehensive carbon emission and energy efficiency objective function by combining the target temperature compensation value, air conditioning equipment operating status data and real-time carbon emission factor sequence data of the power grid. An improved multi-objective particle swarm optimization algorithm is used to solve the comprehensive carbon emission and energy efficiency objective function, and output the optimal chilled water temperature setpoint and the optimal fan speed.

[0201] The closed-loop execution module is used to send the optimal chilled water temperature setpoint as a set reference value to the air conditioning controller, and uses a model-free adaptive control algorithm to calculate the compressor operating frequency control quantity to eliminate execution deviation. At the same time, it sends the optimal fan speed directly to the blower inverter for execution.

[0202] Example 1: A typical factory building with a floor area of ​​approximately 2000 square meters was selected as the target control area. A comparative test experiment was conducted for 24 consecutive hours under typical summer workday conditions. The experimental platform configuration is as follows:

[0203] Infrared array sensors are deployed at the main entrances and exits of the building, with a sampling frequency of once every 5 minutes, to collect real-time indoor pedestrian density sequence data.

[0204] Temperature and humidity sensors are installed in the indoor production area and on the exterior of the building to collect indoor and outdoor temperature and humidity data simultaneously.

[0205] The air conditioning system uses a centrifugal chiller unit with variable frequency fan coil units. The compressor operates at a frequency of 25Hz to 100Hz, the fan speed is adjustable from 15Hz to 50Hz, and the chilled water supply temperature is adjustable from 6℃ to 12℃.

[0206] Real-time carbon emission factor data is obtained from the regional power grid dispatch center through a communication interface, with a data update cycle of 15 minutes.

[0207] The specific values ​​for each preset parameter in this embodiment are set as follows:

[0208] Human flow density heat conversion coefficient The value is taken as 0.025℃ / person / sampling cycle, representing the outdoor temperature heat conversion coefficient. The value is 0.15, representing the heat load response attenuation coefficient. The value is 0.3, and the time series window length N is 12 sampling points;

[0209] Overall objective function weight coefficients , , The values ​​are taken as 0.4, 0.35, and 0.25 respectively, the prediction step size M is 6 steps, and the baseline energy consumption value is... The value is set at 100 kWh, the baseline carbon emission value. The value is 58 kg CO2, and the reference indoor temperature is... The value is 26℃, which is the reference temperature deviation value. The value is 2℃;

[0210] Equipment energy consumption fitting coefficient to The values ​​were obtained by fitting the historical operating data of the air conditioning unit using the least squares method, and the specific values ​​were as follows: , , , , , ;

[0211] The improved multi-objective particle swarm optimization algorithm has a population size of 40, a maximum number of iterations of 100, and a learning factor of [missing information]. and The average value is 1.494, the initial value of the adaptive inertia weight is 0.9, and the final value is 0.4;

[0212] Initial step size factor baseline value in model-free adaptive control algorithm The value is 0.6, which is the error sensitivity coefficient. The value is 1.2, and the weighting factor is... The value is 0.5, which is the penalty factor. and All values ​​are set to 0.01, initial values ​​of pseudo-partial derivatives. The value is 0.8.

[0213] The comparison adopts the existing conventional air conditioning control method, namely a passive PID control strategy based on a fixed temperature threshold.

[0214] Specifically, the chilled water supply temperature is set to 7°C, the blower operating frequency is set to 35Hz, and the compressor operating frequency is adjusted in real time by a conventional PID controller based on the deviation between the measured indoor temperature and the set temperature of 26°C. Feedforward prediction of population density and outdoor weather is not introduced, dynamic changes in grid carbon emission factors are not considered, and multi-objective optimization is not performed.

[0215] During the experimental testing period, the outdoor temperature exhibited a typical summer diurnal variation pattern, with the lowest temperature around 15.4℃ in the early morning and the highest temperature around 32.4℃ around 2 PM.

[0216] The indoor population density is consistent with the characteristics of industrial production and work schedules, with a low level of 20 to 40 people at night, a high level of 60 to 120 people from 9:00 am to 6:00 pm, and a brief drop from 12:00 pm to 1:00 pm as people go out to eat.

[0217] The real-time carbon emission factor of the power grid fluctuates with the daily power structure adjustment. During the daytime when the photovoltaic output is larger, it drops to about 0.45 kgCO2 / kWh. During the morning and evening peak hours, the proportion of thermal power increases, and the carbon emission factor rises to about 0.71 kgCO2 / kWh.

[0218] Under the above operating conditions, the cumulative carbon emissions of the comparative scheme running for 24 hours (selecting the complete time period from 0:00 to 24:00 on the same day) are approximately 906.4 kg CO2.

[0219] Please see Figure 2 This figure shows a comparison of the cumulative carbon emissions of the present invention and the prior art during a 24-hour testing period. The horizontal axis represents time in hours; the vertical axis represents cumulative carbon emissions in kgCO2; the dashed line represents the cumulative carbon emission change curve of the prior art solution, and the solid line represents the cumulative carbon emission change curve of the present invention solution.

[0220] As can be seen from the figure, the overall trend of the carbon emission curve is basically consistent with that of the energy consumption curve. However, due to the intraday fluctuations in the carbon emission factor of the power grid, the gap between the two curves does not widen uniformly, but rather becomes more pronounced during the morning and evening peak hours when the carbon emission factor is higher.

[0221] This phenomenon confirms the effectiveness of the carbon emission sensing and optimization mechanism in this invention:

[0222] When the power grid carbon emission factor is high, the weight of the carbon emission item in the comprehensive objective function increases, and the optimization algorithm will actively select a more energy-efficient combination of operating parameters, thereby achieving a greater carbon reduction effect during periods of high carbon emission intensity.

[0223] At the end of the test period, the cumulative carbon emissions of the existing technical solution were approximately 906.4 kg CO2, while the cumulative carbon emissions of the present invention were approximately 657.4 kg CO2, resulting in a cumulative reduction of approximately 248.9 kg CO2 in carbon emissions. The overall carbon reduction rate reached approximately 27.5%, which is slightly higher than the energy saving rate, demonstrating the additional carbon reduction benefits brought about by the dynamic adaptation of the grid carbon emission factor.

[0224] Under the above operating conditions, the average deviation between the indoor temperature and the benchmark setting value was approximately 0.608℃ during 24 hours of operation of the comparative scheme (selecting the cross-day period from 8:00 on the current day to 8:00 on the next day). The largest deviation occurred during the period when people entered in the morning and returned at noon (14:00), reaching approximately 1.394℃. The temperature fluctuation was large, and there was a relatively obvious response lag.

[0225] Please see Figure 3 This figure shows a comparison curve of the absolute value of the indoor temperature deviation between the present invention and the prior art over time. The horizontal axis in the figure represents time in hours, and the vertical axis represents the absolute value of the indoor temperature deviation relative to the reference set value in °C. The dashed line represents the temperature deviation curve of the prior art, and the solid line represents the temperature deviation curve of the present invention. The two horizontal dotted lines in the figure correspond to the average deviation levels of the two schemes, respectively.

[0226] As can be seen from the figure, the temperature deviation of the existing technology fluctuates significantly. During periods of rapid load change, such as the morning rush hour when people enter, the lunch hour when people return, and the evening rush hour when people leave, the deviation value shows a significant peak, with the maximum deviation reaching approximately 1.394℃. This reflects the inherent defects of conventional PID control in responding to rapid disturbances, such as lag and large overshoot.

[0227] In contrast, the temperature deviation curve of the present invention is at a lower level overall, with a significantly narrower fluctuation range. Even during periods of sudden load changes, the deviation can be controlled within a small range, with a maximum deviation of approximately 0.441°C throughout the day.

[0228] According to statistics, the average temperature deviation of the existing technology throughout the day is about 0.608℃, while the average temperature deviation of the present invention throughout the day is about 0.216℃, representing a reduction of about 64.5%.

[0229] The results fully demonstrate that the feedforward compensation mechanism constructed by the present invention through the rate of change of pedestrian density and the rate of change of outdoor weather can effectively predict heat load disturbances and perform temperature compensation in advance. Combined with the high-precision tracking capability of the underlying modelless adaptive control, it can significantly improve the control accuracy and stability of the indoor thermal environment, and ensure indoor thermal comfort while achieving energy saving and carbon reduction.

[0230] The following conclusions can be drawn from the comparative tests of the above comparative examples and the embodiments of the present invention:

[0231] Regarding carbon emission indicators, this invention achieves a carbon reduction effect of approximately 27.5%, which is slightly higher than the energy saving rate. This is due to the introduction of a real-time grid carbon emission factor into the comprehensive objective function, enabling the system to have carbon emission sensing capabilities. During periods of high grid carbon intensity, the system can proactively adjust its operating strategy to prioritize energy saving, thereby obtaining additional carbon reduction gains and achieving synergistic optimization of energy consumption and carbon emissions.

[0232] In terms of control accuracy, this invention reduces the average indoor temperature deviation from approximately 0.608℃ to approximately 0.216℃, a reduction of over 60%, and the maximum deviation is also significantly narrowed. The combination of feedforward compensation and adaptive closed-loop control solves both the problem of anticipating load disturbances and the problem of accurately tracking nonlinear deviations at the execution layer, achieving a balance between comfort and economy.

[0233] In summary, the energy consumption and carbon emission intelligent management method and system based on multi-source data fusion proposed in this invention, through the synergistic effect of a three-layer architecture of dynamic load feedforward prediction, multi-objective optimization decision-making, and adaptive closed-loop execution, significantly reduces the operating energy consumption and carbon emissions of the air conditioning system while ensuring indoor thermal comfort, demonstrating clear technological advancement and practical application value.

[0234] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0235] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent management and control of energy consumption and carbon emissions based on multi-source data fusion, characterized in that: The method includes: S100: Real-time collection of population density sequence data, indoor and outdoor temperature and humidity sequence data, air conditioning equipment operation status data, and real-time carbon emission factor sequence data of the power grid within the target control area; S200: Calculate the rate of change of pedestrian density based on pedestrian density sequence data, calculate the rate of change of outdoor weather based on indoor and outdoor temperature and humidity sequence data, and calculate the target temperature compensation value based on the rate of change of pedestrian density and the rate of change of outdoor weather. S300: Combine the target temperature compensation value, air conditioning equipment operation status data and real-time carbon emission factor sequence data of the power grid to construct and solve the comprehensive carbon emission and energy efficiency objective function, and output the optimal chilled water temperature setpoint and the optimal fan speed. S400 calculates the compressor's operating frequency control value based on the optimal chilled water temperature setpoint, and simultaneously uses the optimal fan speed as the blower's operating frequency control value, sending it to the air conditioning equipment to perform synchronous management of energy consumption and carbon emissions.

2. The intelligent energy consumption and carbon emission control method based on multi-source data fusion according to claim 1, characterized in that: In S100, pedestrian density sequence data is obtained by infrared array sensors deployed at the entrances and exits of the area at a set sampling frequency; Indoor and outdoor temperature and humidity sequence data are obtained synchronously through temperature and humidity sensors deployed indoors and outdoors; The operating status data of the air conditioning equipment includes the operating frequency of the compressor and the blower, as well as the chilled water temperature; Real-time carbon emission factor sequence data of the power grid is obtained from the power grid dispatch center in real time through the communication interface, and the real-time carbon emission factor sequence data of the power grid is stored in the time series database aligned with the timestamp.

3. The intelligent energy consumption and carbon emission control method based on multi-source data fusion according to claim 2, characterized in that: S200 includes: S201. The rate of change of pedestrian density is obtained by performing first-order difference calculation on the pedestrian density sequence data. S202. Calculate the outdoor temperature series slope in the indoor and outdoor temperature and humidity series data to obtain the outdoor meteorological change rate. S203. The target temperature compensation value is calculated based on the rate of change of pedestrian density and the rate of change of outdoor weather conditions. .

4. The intelligent energy consumption and carbon emission control method based on multi-source data fusion according to claim 3, characterized in that: In S203, the target temperature compensation value is calculated. The formula is as follows: ； In the formula, The preset heat conversion coefficient for pedestrian density; The preset outdoor temperature heat conversion coefficient; For the first Rate of change in pedestrian density at each sampling point; For the first Outdoor weather change rate at each sampling point; The preset heat load response attenuation coefficient; The sampling point number corresponding to the current time. This refers to the serial number of the historical sampling point; This represents the number of sampling points included in the time series window. It is a natural constant.

5. The intelligent energy consumption and carbon emission control method based on multi-source data fusion according to claim 3, characterized in that: The S300 includes: S301, combined with target temperature compensation value A comprehensive carbon emission and energy efficiency objective function is constructed using air conditioning equipment operating status data and real-time carbon emission factor sequence data from the power grid. ； In the formula, The value of the comprehensive objective function; , , These are preset weighting coefficients; To predict the step size; For the first The energy consumption value of the compressor within each predicted step; For the first Energy consumption value of the blower within a predicted step size; This is the preset baseline energy consumption value; For the first Real-time carbon emission factor of the power grid with a predicted step size; This is a preset baseline carbon emission value; Preset baseline indoor temperature; This refers to the actual indoor temperature. This is the preset reference temperature deviation value; The energy consumption model for air conditioning equipment is as follows: ； ； In the formula, For the first The chilled water temperature setpoint for each predicted step size; For the first The predicted step size of the blower speed; The preset equipment energy consumption fitting coefficient; S302. Initialize the particle population of the improved multi-objective particle swarm optimization algorithm, where each particle represents a set of decision variables including the chilled water temperature setpoint and the blower speed. S303. Calculate the fitness value of each particle in the current iteration, and update the velocity and position of the particles by introducing adaptive inertia weights to generate a new generation of particle population; S304. When the maximum number of iterations is reached, the set of non-dominated solutions in the external archive is output, and a fuzzy decision method is used to select a compromise solution from the set of non-dominated solutions as the optimal chilled water temperature setpoint and the optimal fan speed.

6. The intelligent energy consumption and carbon emission control method based on multi-source data fusion according to claim 5, characterized in that: In S303, the update formulas for particle velocity and position are as follows: ； ； In the formula, For the first The particle in the first The speed of each iteration; For the first The particle in the first The position of the next iteration; An adaptive inertia weight that decreases linearly with the number of iterations; and For learning factors; and A random number between [0, 1]; For the first The individual historical best position of each particle; This is the best position in the global history.

7. The intelligent energy consumption and carbon emission control method based on multi-source data fusion according to claim 5, characterized in that: The S400 includes: S401. Collect the actual chilled water temperature of the air conditioning equipment and calculate the system output error between the optimal chilled water temperature setpoint and the actual chilled water temperature. S402. A model-free adaptive control algorithm is adopted. Based on the system output error, the control input is dynamically updated using the pseudo-partial derivative online estimation method. The dynamic step size factor and the estimated value of the pseudo-partial derivative at the next moment are calculated, thereby obtaining the compressor's operating frequency control quantity at the next moment. S403: The optimal fan speed is used as the operating frequency control quantity of the blower. The operating frequency control quantities of the compressor and blower are then converted into digital communication commands and sent to the air conditioning equipment for execution, thereby achieving synchronous management of energy consumption and carbon emissions.

8. The intelligent energy consumption and carbon emission control method based on multi-source data fusion according to claim 7, characterized in that: In S402, the formula for calculating the dynamic step size factor is: ； In the formula, For the first The dynamic step size factor at any given time; This is the initial step size factor baseline value; This is the error sensitivity coefficient; For the first Reference values ​​for setting the time; For the first The actual chilled water temperature at that moment; The pseudo-partial derivative online estimation formula and control quantity calculation formula for the model-free adaptive control algorithm are as follows: ； ； In the formula, For the first The pseudo-partial derivative estimate at time t; For the first The change in control input at any given time; ; These are preset weighting factors; , These are the preset penalty factors; For the first The compressor operating frequency control value at any given time.

9. An intelligent energy consumption and carbon emission control system based on multi-source data fusion, characterized in that: The system, applied to the intelligent energy consumption and carbon emission control method based on multi-source data fusion as described in claim 1, comprises: The data acquisition and sensing module is used to collect real-time data on population density, indoor and outdoor temperature and humidity, air conditioning equipment operation status, and real-time carbon emission factor data of the power grid within the target control area. The dynamic load calculation module is used to calculate the rate of change of pedestrian density based on pedestrian density sequence data, calculate the rate of change of outdoor weather based on indoor and outdoor temperature and humidity sequence data, and calculate the target temperature compensation value based on the rate of change of pedestrian density and the rate of change of outdoor weather. The optimization decision module is used to construct a comprehensive carbon emission and energy efficiency objective function by combining the target temperature compensation value, air conditioning equipment operation status data and real-time carbon emission factor sequence data of the power grid. An improved multi-objective particle swarm optimization algorithm is used to solve the comprehensive carbon emission and energy efficiency objective function, and output the optimal chilled water temperature setpoint and the optimal fan speed. The closed-loop execution module is used to send the optimal chilled water temperature setpoint as a set reference value to the air conditioning controller, and uses a model-free adaptive control algorithm to calculate the compressor operating frequency control quantity to eliminate execution deviation. At the same time, it sends the optimal fan speed directly to the blower inverter for execution.

Images