Load optimization scheduling method and system based on thermodynamic and electrical coupling model

Abstract

The invention discloses a load optimization scheduling method and system based on a thermodynamic and electrical coupling model, relates to the technical field of load scheduling, and solves the technical problem that the existing load adjustment method neglects the coupling relationship between thermodynamic characteristics and electrical parameters of devices, resulting in poor adjustment effect. The method comprises the steps of obtaining operation data of load equipment; processing the operation data to obtain initial data; extracting and fusing multi-dimensional data features of the initial data to obtain a model input vector; constructing a thermodynamic and electrical parameter coupling model; the optimal scheduling strategy is generated based on the coupling model and the optimization algorithm, and the technical problem is solved.

Description

Technical Field

[0001] This invention belongs to the field of load scheduling, specifically a load optimization scheduling method and system based on a thermodynamic and electrical coupling model. Background Technology

[0002] Currently, load regulation in power systems mainly relies on traditional methods, failing to fully consider the coupling mechanism between the thermodynamic and electrical parameters of load equipment. Variable frequency compressors, water heaters, and charging piles, among other load equipment, have significant potential for power regulation, but current technologies struggle to achieve efficient power regulation. The operating state of load equipment is closely related to power load fluctuations, but existing load regulation methods neglect the coupling relationship between the thermodynamic characteristics and electrical parameters of these devices, resulting in poor regulation performance.

[0003] Therefore, this invention proposes a load optimization scheduling method and system based on a thermodynamic and electrical coupling model to solve the above-mentioned technical problems. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a load optimization scheduling method and system based on a thermodynamic and electrical coupling model, which is used to solve the technical problem that the existing load regulation methods ignore the coupling relationship between the thermodynamic characteristics and electrical parameters of these devices, resulting in poor regulation effect.

[0005] To achieve the above objectives, a first aspect of the present invention provides a load optimization scheduling method based on a thermodynamic and electrical coupling model, comprising: Obtain operating data from the load equipment; Process the running data to obtain the initial data; Extract and fuse the multidimensional data features of the initial data to obtain the model input vector; Construct a coupled model of thermodynamic and electrical parameters; The optimal scheduling strategy is generated based on the coupling model and optimization algorithm.

[0006] In conjunction with the first aspect above, in one possible implementation, acquiring the operating data of the load equipment includes: The system collects real-time operating data of the load equipment through data acquisition devices; the operating data includes temperature, voltage, and current, etc.

[0007] It should be noted that load equipment refers to types such as variable frequency compressors (such as core components in air conditioners and refrigeration equipment), water heaters (such as electric water heaters, which rely on electrical energy to achieve heat energy conversion and storage), and charging piles (used for charging electric vehicles, which need to dynamically adjust power according to battery status and grid load). The common feature of these types of equipment is that during their operation, there is both the consumption and conversion of electrical energy (electrical dimension) and changes in thermal energy states such as temperature and pressure (thermodynamic dimension), and the two influence each other, requiring coordinated control through a coupling model. Operational data refers to the raw parameters that are acquired in real time from load equipment through data acquisition devices (sensors, smart meters), reflecting the electrical and thermodynamic states of the equipment. These parameters serve as the foundational input for subsequent data processing and coupled model calculations. They are mainly divided into two categories: first, thermodynamic data, including internal equipment temperatures (such as the cylinder temperature of a variable frequency compressor, the water temperature in a water heater tank), ambient temperature, and medium pressures (such as the water pressure inside a water heater tank), used to characterize the equipment's heat energy conversion, transfer, and storage states; second, electrical data, including voltage, current, and frequency during equipment operation, which can be further used to calculate instantaneous power, power factor, and other electrical characteristics. The document explicitly requires that the sampling frequency of operational data must meet control requirements (usually on the order of seconds or milliseconds), and that the data must undergo cleaning, verification, and synchronization alignment to ensure data reliability and timeliness.

[0008] In conjunction with the first aspect above, in one possible implementation, the processing of the running data to obtain initial data includes: Initial data is obtained by preprocessing the running data; the preprocessing includes: linear interpolation, spline interpolation, historical mean completion method and 3σ rule; The initial data from different devices are aligned by timestamps to construct vector samples within a unified time window.

[0009] It should be noted that the content attributes of the initial data and the running data are consistent; The construction of vector samples within a unified time window refers to timestamping the operating data of different load devices (such as frequency converters, water heaters, and charging piles), then constructing a unified time window (such as 1 minute), and aggregating the data of each device within the same time window into vector samples (such as "[frequency converter temperature at time t, water heater power at time t, charging pile current at time t, ...]"), to ensure that data from multiple devices can be analyzed collaboratively in the time dimension.

[0010] In conjunction with the first aspect above, in one possible implementation, the extraction and fusion of multidimensional data features from the initial data to obtain the model input vector includes: Electrical feature extraction: The electrical features corresponding to the equipment are calculated based on the initial data; among which, the electrical features include: instantaneous power, power factor, and harmonic content; Thermodynamic feature extraction: Based on thermodynamic equations, key thermodynamic features are extracted; among them, thermodynamic features include: heat flow, thermal efficiency, and rate of change of temperature difference; Multi-source data fusion: Principal component analysis is used to reduce the dimensionality of electrical and thermodynamic features, retaining principal components with a cumulative variance contribution rate of ≥90%, and eliminating linear redundancy between features; Kalman filtering is used for dynamic estimation and fusion; The fused electrical and thermodynamic features are concatenated into a multi-dimensional input vector to obtain the model input vector.

[0011] In conjunction with the first aspect above, in one possible implementation, the construction of the coupled thermodynamic and electrical parameter model includes: Based on the thermodynamic equations and electrical equations, nonlinear coupling terms are introduced to form a coupled model. The thermodynamic equation is specifically as follows: Where m is the mass of the working medium, and c p It is the specific heat capacity, Q in It is the input heat, Q loss It represents the heat loss, and dT / dt is the rate of temperature change. The electrical equations are specifically as follows: Where U and I are voltage and current respectively, and η elec It is electrical efficiency, P elec It is the electrical input power of the equipment; Nonlinear coupling terms: The interaction between thermodynamic and electrical parameters is quantified by the equation f(T,P)=0; Forming a coupled model by combining the two: ; The effect of temperature on electrical efficiency: η elec =η0×(1 α(T T0); where η0 is the reference electrical efficiency, α is the temperature influence coefficient, and T0 is the reference temperature; The effect of electrical power on heat loss: Q loss =k×(T Tenv)×(1+βP elec ); where k is the thermal resistance coefficient, Tenv is the ambient temperature, and β is the power influence coefficient, ensuring that the model reflects the true nonlinear relationship of the device's "thermal-electric" coupling.

[0012] It should be noted that, using experimental or historical operating data of the load equipment (such as temperature change curves of water heaters at different power levels, and current curves of charging piles at different SOC levels), the key parameters of the model—including the working medium mass m and specific heat capacity c—are determined by employing least squares fitting or maximum likelihood estimation methods. p Parameters such as thermal resistance coefficient k and temperature influence coefficient α are used to ensure that the parameters match the actual characteristics of the equipment.

[0013] In conjunction with the first aspect above, in one possible implementation, the generation of the optimal scheduling strategy based on the coupled model and optimization algorithm includes: By training an artificial intelligence model with historical data, a predictive model can be obtained. The predicted data is obtained by making predictions based on the initial data using a prediction model; The optimal scheduling strategy is obtained by optimizing the predicted data based on the optimization algorithm.

[0014] It should be noted that the predicted data refers to the key data output by the trained artificial intelligence model (such as convolutional neural network or deep belief network) during the process of generating candidate scheduling strategies based on the coupled model and optimization algorithm. This artificial intelligence model first takes the multi-dimensional feature vector fused at the initial time (including electrical characteristics such as instantaneous power and thermodynamic characteristics such as temperature difference rate of change of frequency converter, water heater, and charging pile, i.e., standard input data) as input, and uses the actual state of the equipment at the predicted time (such as equipment temperature, charging current, battery SOC, etc. in the next 5-15 minutes, i.e., standard output data) as labels, and uses the coupled model (including thermodynamic equation, electrical equation, and nonlinear coupling terms) as physical constraints to remove abnormal data before training. Then, the initial data is input into the model, and the final output is data covering the electrical state (such as predicted power and predicted current) and thermodynamic state (such as predicted water temperature and predicted SOC) of the three types of core load equipment in the next 5-15 minutes. This provides key future equipment state information for subsequent multi-objective optimization and candidate scheduling strategy generation through optimization algorithms such as genetic algorithm and particle swarm optimization.

[0015] In conjunction with the first aspect above, in one possible implementation, training the artificial intelligence model using historical data includes: The initial data at the initial moment is integrated into the input data, and the labels of the manually set prediction moments are integrated into the output data. Using the coupled model as a physical constraint, outliers in the standard input and standard output data are removed to obtain the standard input and standard output data. An artificial intelligence model is trained based on standard input data and standard output data to obtain a prediction model; wherein, the artificial intelligence model includes: convolutional neural network or deep belief network.

[0016] It should be noted that removing outliers from standard input and output data refers to using the "physical laws governing the interaction between heat and electricity" as the basis for judging the reasonableness of the data, filtering out outliers in the standard input and output data that violate these laws. "Standard input data" refers to multi-dimensional core feature vectors, such as instantaneous power, thermal efficiency, and rate of temperature change, covering the key electrical and thermodynamic characteristics of variable frequency compressors, water heaters, and charging piles. "Standard output data" refers to the equipment status labels corresponding to the prediction time (such as the water heater temperature and charging pile current in the next 5 minutes). The physical constraints of the coupled model define the "reasonable range" of the data—for example, according to the thermodynamic equation, when the water heater power is 1000W, the data within 5 minutes... The maximum temperature rise of the water inside the clock is approximately 2.35℃. If the input data "water heater power 1000W" corresponds to the output data "water temperature rises 5℃ in 5 minutes", it is judged as abnormal data (violating the coupling relationship between power and temperature change). For example, according to electrical equations and grid constraints, the charging pile current should not exceed 15A. If the output data shows "charging pile current 20A", it is judged as abnormal because it exceeds the physical boundary of safe operation of the equipment. Removing these abnormal data is to ensure that the dataset for subsequent training of artificial intelligence models (such as convolutional neural networks and deep belief networks) conforms to the actual operating rules of the equipment, avoids the model learning false "thermal-electrical correlation", and thus ensures the physical feasibility of the prediction results. The integration of manually set prediction time tags into output data is based on scheduling needs (such as planning power regulation strategies 5-15 minutes in advance). The system manually specifies the future time nodes to be predicted and organizes the actual operating status of the equipment at these nodes (i.e., "tags") into structured output data for training artificial intelligence models. The "human-set prediction time" needs to be tailored to the scenario requirements and is typically set to 5, 10, or 15 minutes in the future (e.g., for load regulation 10 minutes before peak power generation, the prediction time is set to "current time + 10 minutes"). "Tags" refer to key parameters (covering electrical and thermodynamic dimensions) collected from the equipment's historical operating logs at these prediction times that reflect the equipment's true state—for example, for a water heater, the tags could be "current time + 5 minutes water temperature" or "current time + 10 minutes thermal efficiency." For charging piles, the labels can be "current time + 8 minutes of charging current" or "current time + 15 minutes of battery SOC"; for variable frequency compressors, the labels can be "current time + 7 minutes of cooling power" or "current time + 12 minutes of air outlet temperature"; "integrating into output data" means structuring these scattered labels according to the logical structure of "predicted time - device type - parameter type" (e.g., constructing a vector form of "time t+5: water heater temperature 55℃, charging pile current 12A, variable frequency compressor power 1.8kW"), and matching them one-to-one with the standard input data of the corresponding initial time (e.g., the fused feature vector of time t), forming "input-output" training sample pairs, ensuring that the artificial intelligence model can learn the mapping relationship of "current device characteristics → future specified time state", and ultimately meet the requirement of "generating scheduling basis based on prediction model" in the document.

[0017] In conjunction with the first aspect above, in one possible implementation, the step of optimizing the predicted data based on an optimization algorithm to obtain the optimal scheduling strategy includes: The goal is to minimize power regulation accuracy, response time, and energy consumption cost. The prediction results output by the prediction model are used as the optimization benchmark; At the same time, the physical constraint boundaries for optimization are clearly defined by combining thermodynamic and electrical parameter coupling models: equipment safety operation constraints, including the maximum / minimum power and safe temperature range of each piece of equipment; The optimal scheduling strategy is obtained by iteratively searching for the optimal scheduling strategy by taking the power adjustment of each load device as the optimization object through optimization algorithms; among them, optimization algorithms include: genetic algorithm, particle swarm optimization or multi-objective optimization.

[0018] This invention defines multi-dimensional optimization objectives, namely, minimizing power regulation deviation (ensuring the deviation between the actual power of the optimized equipment and the predicted power output by the prediction model (such as the total predicted power of the charging pile and the predicted heating power of the water heater in the next 5-15 minutes) is controlled within a reasonable range to avoid load fluctuations caused by insufficient regulation accuracy), minimizing response time (shortening the time from the generation of the optimization strategy to the execution of the regulation command by the equipment to meet the real-time scheduling requirements in the document, such as ensuring that the response time does not exceed 10 seconds), and minimizing energy consumption cost (reducing the total power consumption of load equipment such as frequency converters, water heaters, and charging piles, such as reducing the ineffective operation of high-energy-consuming equipment through reasonable power allocation). The core optimization directions are: first, the results output by the prediction model; and second, the results of the prediction model. The optimization benchmark, whose prediction results cover the electrical state (e.g., predicted current, predicted power) and thermodynamic state (e.g., predicted water temperature, predicted battery SOC) of each load device within a specific future time period, is the core basis for the "expected future equipment state" in the optimization process, ensuring that the optimization strategy does not deviate from actual operational needs. Simultaneously, it strictly combines a thermodynamic and electrical parameter coupling model (including thermodynamic equations, electrical equations, and nonlinear coupling terms) to clearly define the physical constraint boundaries of the optimization. Specifically, the equipment safety operation constraints include the maximum / minimum power limits for each type of load device (e.g., the maximum power of a single charging pile does not exceed 3.5kW, and the minimum operating power is not less than 1kW; the maximum heating power of a water heater does not exceed 3kW, and the minimum heat preservation power is not less than 500W) and safe temperature... The optimization process involves setting specific temperature ranges (e.g., water heater tank temperature needs to be maintained between 30℃ and 75℃, charging pile battery temperature needs to be controlled between 0℃ and 45℃, and inverter compressor outlet temperature needs to be within safe thresholds) to avoid optimization results violating the physical operating laws of the equipment. Finally, the optimization targets are the power adjustment amounts of each load device (i.e., the power adjustment values ​​set for the inverter compressor, water heater, and charging pile respectively, such as adjusting the single-pile power of the charging pile from the predicted 2.5kW to 2.2kW, and the water heater power from the predicted 1.9kW to 1.7kW). Iterative optimization is then performed using optimization algorithms (Genetic Algorithm (GA), Particle Swarm Optimization (PSO), or multi-objective optimization algorithms). For example, when using PSO, the power adjustment amounts of each device are encoded as particle dimensions within the physical constraint boundaries. The particle position (i.e., power adjustment amount) is iteratively adjusted using a velocity-position update formula. After each iteration, the objective function value (a comprehensive score of power adjustment deviation, response time, and energy consumption cost) corresponding to the adjustment amount is calculated, and it is verified whether all constraints are met (such as total power not exceeding the grid's allocable limit and temperature not exceeding the safe range). If not, a forced correction is made. When the iteration converges (e.g., no better objective function value is found after 5 consecutive iterations), the Pareto optimal solution that takes into account multiple objectives is selected, and finally, a solution containing precise power adjustment commands for each device (e.g., "charging pile current limit to 10A (power 2.2kW), water heater power reduced to 1.7kW, frequency converter compressor power adjusted to 1.9kW") and corresponding objective function values ​​(e.g., power adjustment deviation 3).The optimal scheduling strategy (2% reduction in efficiency, 8-second response time, and 5% reduction in energy consumption cost) provides a basis for selecting the final execution strategy based on real-time grid status and operational priorities.

[0019] A second aspect of the present invention provides a load optimization scheduling system based on a thermodynamic and electrical coupling model, comprising: a data acquisition module, a data analysis module, and an optimization scheduling module; Data acquisition module: used to acquire operating data of load equipment; Data analysis module: used to process operational data to obtain initial data; extract and fuse multidimensional data features of the initial data to obtain model input vector; construct a coupled model of thermodynamic and electrical parameters; Optimized scheduling module: Generates the optimal scheduling strategy based on the coupled model and optimization algorithm.

[0020] In conjunction with the first aspect above, in one possible implementation, the data analysis module is communicatively and / or electrically connected to the data acquisition module and the optimization scheduling module, respectively.

[0021] Compared with the prior art, the beneficial effects of the present invention are: This invention constructs a coupled model of thermodynamic and electrical parameters, transforming load regulation from "blindly adjusting power in a single dimension" to "precise control through coordinated thermo-electrical regulation," significantly improving regulation accuracy and avoiding the regulation deviations caused by neglecting coupling relationships in existing technologies. Addressing the problem of "difficulty in efficiently regulating equipment such as variable frequency compressors, water heaters, and charging piles," this invention collects thermodynamic and electrical data from multiple devices, preprocesses and fuses them to form a unified model input vector, then relies on an artificial intelligence model to predict future device states. Finally, using power regulation as the optimization target, it achieves multi-objective optimization through genetic algorithms, particle swarm optimization, etc., forming a unified scheduling strategy covering multiple devices, thus solving the problems of existing technologies. This invention addresses the limitations of existing technologies in coordinating different types of loads, improving system regulation flexibility and efficiency. To address the issue of "disconnection between load regulation and equipment operating status and grid fluctuations," this invention uses a coupled model as a physical constraint. During data preprocessing, it eliminates abnormal data that violates thermo-electrical laws. In the optimization phase, it ensures that the scheduling strategy does not exceed equipment safety and grid capacity limits. Simultaneously, it uses a predictive model to anticipate future load conditions, transforming regulation from "passive response" to "active planning." This effectively reduces the impact of grid load fluctuations on regulation performance, ultimately achieving a comprehensive effect of improved load regulation accuracy, faster response speed, and lower energy costs, thoroughly improving the poor regulation performance of existing technologies. Attached Figure Description

[0022] 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.

[0023] Figure 1 This is a schematic diagram of the method steps in an embodiment of the present invention; Figure 2 This is a schematic diagram of the system modules in an embodiment of the present invention. Detailed Implementation

[0024] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. 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.

[0025] Please see Figure 1 The first aspect of this invention provides a load optimization scheduling method based on a thermodynamic and electrical coupling model, comprising: Obtain operating data from the load equipment; Process the running data to obtain the initial data; Extract and fuse the multidimensional data features of the initial data to obtain the model input vector; Construct a coupled model of thermodynamic and electrical parameters; The optimal scheduling strategy is generated based on the coupling model and optimization algorithm.

[0026] Obtain operational data from the load equipment, including: The system collects real-time operating data of the load equipment through data acquisition devices; the operating data includes temperature, voltage, and current, etc.

[0027] The running data is processed to obtain initial data, including: Initial data is obtained by preprocessing the running data; the preprocessing includes: linear interpolation, spline interpolation, historical mean completion method and 3σ rule; The initial data from different devices are aligned by timestamps to construct vector samples within a unified time window.

[0028] Extract and fuse the multidimensional data features of the initial data to obtain the model input vector, including: Electrical feature extraction: The electrical features corresponding to the equipment are calculated based on the initial data; among which, the electrical features include: instantaneous power, power factor, and harmonic content; Thermodynamic feature extraction: Based on thermodynamic equations, key thermodynamic features are extracted; among them, thermodynamic features include: heat flow, thermal efficiency, and rate of change of temperature difference; Multi-source data fusion: Principal component analysis is used to reduce the dimensionality of electrical and thermodynamic features, retaining principal components with a cumulative variance contribution rate of ≥90%, and eliminating linear redundancy between features; Kalman filtering is used for dynamic estimation and fusion; The fused electrical and thermodynamic features are concatenated into a multi-dimensional input vector to obtain the model input vector.

[0029] Constructing a coupled model of thermodynamic and electrical parameters, including: Based on the thermodynamic equations and electrical equations, nonlinear coupling terms are introduced to form a coupled model. The thermodynamic equation is specifically as follows: Where m is the mass of the working medium, and c p It is the specific heat capacity, Q in It is the input heat, Q loss It represents the heat loss, and dT / dt is the rate of temperature change. The electrical equations are specifically as follows: Where U and I are voltage and current respectively, and η elec It is electrical efficiency, P elec It is the electrical input power of the equipment; Nonlinear coupling terms: The interaction between thermodynamic and electrical parameters is quantified by the equation f(T,P)=0; Forming a coupled model by combining the two: ; The effect of temperature on electrical efficiency: η elec =η0×(1 α(T T0); where η0 is the reference electrical efficiency, α is the temperature influence coefficient, and T0 is the reference temperature; The effect of electrical power on heat loss: Q loss =k×(T Tenv)×(1+βP elec ); where k is the thermal resistance coefficient, Tenv is the ambient temperature, and β is the power influence coefficient, ensuring that the model reflects the true nonlinear relationship of the device's "thermal-electric" coupling.

[0030] The optimal scheduling strategy is generated based on the coupling model and optimization algorithm, including: By training an artificial intelligence model with historical data, a predictive model can be obtained. The predicted data is obtained by making predictions based on the initial data using a prediction model; The optimal scheduling strategy is obtained by optimizing the predicted data based on the optimization algorithm.

[0031] Training artificial intelligence models using historical data includes: The initial data at the initial moment is integrated into the input data, and the labels of the manually set prediction moments are integrated into the output data. Using the coupled model as a physical constraint, outliers in the standard input and standard output data are removed to obtain the standard input and standard output data. An artificial intelligence model is trained based on standard input data and standard output data to obtain a prediction model; wherein, the artificial intelligence model includes: convolutional neural network or deep belief network.

[0032] The optimal scheduling strategy is obtained by optimizing the predicted data based on the optimization algorithm, including: The goal is to minimize power regulation accuracy, response time, and energy consumption cost. The prediction results output by the prediction model are used as the optimization benchmark; At the same time, the physical constraint boundaries for optimization are clearly defined by combining thermodynamic and electrical parameter coupling models: equipment safety operation constraints, including the maximum / minimum power and safe temperature range of each piece of equipment; The optimal scheduling strategy is obtained by iteratively searching for the optimal scheduling strategy by taking the power adjustment of each load device as the optimization object through optimization algorithms; among them, optimization algorithms include: genetic algorithm, particle swarm optimization or multi-objective optimization.

[0033] For example, taking a load dispatching scenario in a residential community's public area (including 2 inverter air conditioners, 1 100L electric water heater, and 8 AC charging piles) as an example, the peak electricity consumption period is from 18:00 to 19:00 every day. The real-time power allocation limit for this area by the power grid is 25kW. At 18:00, the actual measured total power has reached 27.2kW (exceeding the limit by 8.8%). The optimal dispatching strategy is generated using the method of this invention, and the specific process is as follows: Step 1: Obtain the operating data of the load equipment; 1. Data acquisition equipment deployment: Install temperature sensors (to measure indoor temperature and air outlet temperature) and smart meters for variable frequency air conditioners; install water temperature sensors, pressure sensors, and smart meters for water heaters; install current sensors, SOC sensors, and smart meters for each charging pile, with sampling frequencies set to the second level (air conditioners and charging piles) and the 10-second level (water heaters).

[0034] Data acquisition will begin at 18:00. Thermodynamic data: Indoor temperature 28℃ (air conditioner target 26℃), water heater temperature 52℃ (target 60℃), average SOC of charging pile battery 35% (target 80%), ambient temperature 29℃, water heater tank pressure 0.35MPa; Electrical data: Air conditioner: 220V, 9.8A (single unit power 2156W, total power 4312W); Water heater: 220V, 13.2A (power 2904W); Charging pile: 220V, 12.5A (single unit power 2750W, total power 22000W).

[0035] Step 2: Process the running data to obtain initial data; 1. Preprocessing steps: Missing value imputation: The current data of air conditioner No. 2 at 18:00:03 is missing. Using the "linear interpolation method", based on 9.8A at 18:00:02 and 9.7A at 18:00:04, it is imputed to 9.75A. Outlier removal: At 18:00:05, the current of charging pile No. 7 suddenly increased to 25A (normal range 8-15A). It was identified as an outlier by the "3σ rule" (mean current 12.3A, standard deviation 1.8A, outlier threshold > 12.3 + 3 × 1.8 = 17.7A) and replaced with the mean current of 12.4A for the 5 seconds before and after. 2. Data synchronization and alignment: Based on the central clock of the community, a unified timestamp is applied to the data of all devices to construct a vector sample of "1-minute time window". The sample of the window from 18:00 to 18:01 is: [28℃ (air conditioner room temperature), 52℃ (water heater water temperature), 35% (charging pile SOC), 4312W (total power of air conditioner), 2904W (power of water heater), 22000W (total power of charging pile)].

[0036] Step 3: Extract and fuse the multidimensional data features of the initial data to obtain the model input vector; 1. Feature extraction: Electrical characteristics: Air conditioner instantaneous power P = 220 × 9.75 × 0.92 ≈ 1978W (power factor 0.92), water heater power factor 0.95, charging pile third harmonic emission rate 1.8%; Thermodynamic characteristics: Heat flow rate of water heater Q = 100 × 4.2 × (52 - 29) = 9660 kJ / h (specific heat capacity of water 4.2 kJ / (kg·℃)), air conditioning temperature difference change rate dT / dt = (28 - 26) / 60 ≈ 0.033℃ / s, thermal efficiency of water heater η th =2904×0.92 / 2904≈0.92 (heat loss 8%) 2. Data Fusion: Principal component analysis dimensionality reduction: Dimensionality reduction was performed on the 11 original features, and the three principal components with a cumulative variance contribution rate of 91% were retained (PC1: total power, PC2: temperature deviation, PC3: charging SOC). Kalman filtering: Dynamically fuses and smooths the power of charging piles with large fluctuations, stabilizing the total power at 21800W; 3. Construct the input vector: The final model input vector x=[21800,4312,2904,0.033,0.92,35%] (which represents the power of the charging pile / air conditioner / water heater, the air conditioner temperature difference change rate, the water heater thermal efficiency, and the charging pile SOC, respectively).

[0037] Step 4: Construct a coupled model of thermodynamic and electrical parameters;

[0038] 1. Solve the core equations simultaneously: Thermodynamic equation (water heater): 100×4.2×dT / dt=2904×0.92-8×(T-29) (thermal resistance coefficient 8W / ℃), calculated to dT / dt≈0.58℃ / min (water temperature rises by 0.58℃ per minute, it takes 13.8 minutes to go from 52℃ to 60℃); Electrical equation (charging pile): P elec =220×12.4×0.9≈2491W (charging efficiency 0.9), which deviates from the measured 2750W by 9.4%; Nonlinear coupling term: The effect of charging pile temperature on efficiency is 0.896 (battery temperature 32℃, temperature influence coefficient 0.0018 / ℃), after correction P elec =2480W, deviation reduced to 2.5%; 2. Parameter calibration: Using historical data (water heater power-temperature curve) from the past 7 days at 18:00, the thermal resistance coefficient k=7.8W / ℃ was corrected by least squares fitting, and the model verification R^2=0.93 (meets the standard).

[0039] Step 5: Generate the optimal scheduling strategy based on the coupled model and optimization algorithm; 1. Train the prediction model and obtain the prediction data: Training a deep belief network: Using the "input vector - future 10-minute state" data from the past 3 months as samples (input is x, labels are temperature and power for the next 10 minutes), a coupled model is used to remove abnormal samples such as "water heater with 1000W power but a 5℃ temperature increase". After training, the model predicts the following data at 18:10: air conditioner room temperature 26.5℃, water heater water temperature 58℃, and total charging pile power 23000W (total power 23000+4312+2904=30216W, exceeding the 25kW limit). 2. Optimize the algorithm through iterative optimization: Objectives and constraints: Minimize power deviation (≤5%), response time (≤10 seconds), and energy consumption cost. Constraints are: total power ≤25kW, water heater temperature ≥55℃, and single charging pile power ≤3.5kW. Particle swarm optimization: The power adjustment of 8 charging piles + 2 air conditioners + 1 water heater is encoded as 11-dimensional particles. After 30 iterations, the process converges, and the "balancing strategy" in the Pareto optimal solution is selected: the total power of the charging piles is reduced to 18000W (2250W per pile, 10.2A current), the total power of the air conditioners is reduced to 3800W (1900W per unit), and the power of the water heater is maintained at 2904W. 3. Optimal Strategy Output: The final scheduling strategy is: "Air conditioner single unit power 1900W (temperature control 26.5℃), water heater power 2904W (reaches 60℃ in 13.8 minutes), charging pile single pile current limit 10.2A (power 2250W)", total power 18000+3800+2904=24704W (≤25kW), power deviation 3.9%, response time 7 seconds, energy consumption cost reduction 6%.

[0040] See Figure 2 The second aspect of the present invention provides a load optimization scheduling system based on a thermodynamic and electrical coupling model, including: a data acquisition module, a data analysis module and an optimization scheduling module; Data acquisition module: used to acquire operating data of load equipment; Data analysis module: used to process operational data to obtain initial data; extract and fuse multidimensional data features of the initial data to obtain model input vector; construct a coupled model of thermodynamic and electrical parameters; Optimized scheduling module: Generates the optimal scheduling strategy based on the coupled model and optimization algorithm.

[0041] Some of the data in the above formula are calculated by removing dimensions and taking their numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0042] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A load optimization scheduling method based on a thermodynamic and electrical coupling model, characterized in that, include: Obtain operating data from the load equipment; Process the running data to obtain the initial data; Extract and fuse the multidimensional data features of the initial data to obtain the model input vector; Construct a coupled model of thermodynamic and electrical parameters; The optimal scheduling strategy is generated based on the coupling model and optimization algorithm.

2. The load optimization scheduling method based on a thermodynamic and electrical coupling model according to claim 1, characterized in that, The acquisition of operating data of the load equipment includes: The system collects real-time operating data of the load equipment using data acquisition devices.

3. The load optimization scheduling method based on a thermodynamic and electrical coupling model according to claim 1, characterized in that, The process of processing the running data to obtain initial data includes: Initial data is obtained by preprocessing the running data; the preprocessing includes: linear interpolation, spline interpolation, historical mean completion method and 3σ rule; The initial data from different devices are aligned by timestamps to construct vector samples within a unified time window.

4. The load optimization scheduling method based on a thermodynamic and electrical coupling model according to claim 1, characterized in that, The multidimensional data features of the initial data are extracted and fused to obtain the model input vector. include: Electrical feature extraction: The electrical features corresponding to the equipment are calculated based on the initial data; among which, the electrical features include: instantaneous power, power factor, and harmonic content; Thermodynamic feature extraction: Based on thermodynamic equations, key thermodynamic features are extracted; among them, thermodynamic features include: heat flow, thermal efficiency, and rate of change of temperature difference; Multi-source data fusion: Principal component analysis is used to reduce the dimensionality of electrical and thermodynamic features, retaining principal components with a cumulative variance contribution rate of ≥90%, and eliminating linear redundancy between features; Kalman filtering is used for dynamic estimation and fusion; The fused electrical and thermodynamic features are concatenated into a multi-dimensional input vector to obtain the model input vector.

5. The load optimization scheduling method based on a thermodynamic and electrical coupling model according to claim 1, characterized in that, The construction of the coupled thermodynamic and electrical parameter model includes: Based on the thermodynamic equations and electrical equations, nonlinear coupling terms are introduced to form a coupled model. The thermodynamic equation is specifically as follows: Where m is the mass of the working medium, and c p It is the specific heat capacity, Q in It is the input heat, Q loss It represents the heat loss, and dT / dt is the rate of temperature change. The electrical equations are specifically as follows: Where U and I are voltage and current respectively, and η elec It is electrical efficiency, P elec It is the electrical input power of the equipment; Nonlinear coupling terms: The interaction between thermodynamic and electrical parameters is quantified by the equation f(T,P)=0; Forming a coupled model by combining the two: ; The effect of temperature on electrical efficiency: η elec =η0×(1 α(T T0); where η0 is the reference electrical efficiency, α is the temperature influence coefficient, and T0 is the reference temperature; The effect of electrical power on heat loss: Q loss =k×(T Tenv)×(1+βP elec ); where k is the thermal resistance coefficient, Tenv is the ambient temperature, and β is the power influence coefficient.

6. The load optimization scheduling method based on a thermodynamic and electrical coupling model according to claim 1, characterized in that, The generation of the optimal scheduling strategy based on the coupling model and optimization algorithm includes: By training an artificial intelligence model with historical data, a predictive model can be obtained. The predicted data is obtained by making predictions based on the initial data using a prediction model; The optimal scheduling strategy is obtained by optimizing the predicted data based on the optimization algorithm.

7. The load optimization scheduling method based on a thermodynamic and electrical coupling model according to claim 6, characterized in that, The training of the artificial intelligence model using historical data includes: The initial data at the initial moment is integrated into the input data, and the labels of the manually set prediction moments are integrated into the output data. Using the coupled model as a physical constraint, outliers in the standard input and standard output data are removed to obtain the standard input and standard output data. An artificial intelligence model is trained based on standard input data and standard output data to obtain a prediction model; wherein, the artificial intelligence model includes: convolutional neural network or deep belief network.

8. The load optimization scheduling method based on a thermodynamic and electrical coupling model according to claim 6, characterized in that, The optimization of the predicted data based on the optimization algorithm to obtain the optimal scheduling strategy includes: The goal is to minimize power regulation accuracy, response time, and energy consumption cost. The prediction results output by the prediction model are used as the optimization benchmark; At the same time, the physical constraint boundaries for optimization are clearly defined by combining thermodynamic and electrical parameter coupling models: equipment safety operation constraints, including the maximum / minimum power and safe temperature range of each piece of equipment; The optimal scheduling strategy is obtained by iteratively searching for the optimal scheduling strategy by taking the power adjustment of each load device as the optimization object through optimization algorithms; among them, optimization algorithms include: genetic algorithm, particle swarm optimization or multi-objective optimization.

9. A load optimization scheduling system based on a thermodynamic and electrical coupling model, executing the load optimization scheduling method based on a thermodynamic and electrical coupling model as described in any one of claims 1-8, characterized in that, include: Data acquisition module, data analysis module, and optimized scheduling module; Data acquisition module: used to acquire operating data of load equipment; Data analysis module: used to process operational data to obtain initial data; extract and fuse multidimensional data features of the initial data to obtain model input vector; construct a coupled model of thermodynamic and electrical parameters; Optimized scheduling module: Generates the optimal scheduling strategy based on the coupled model and optimization algorithm.

10. The load optimization scheduling method based on a thermodynamic and electrical coupling model according to claim 1, characterized in that, The data analysis module is communicatively and / or electrically connected to the data acquisition module and the optimization scheduling module, respectively.

Images