An optimized scheduling method for electric drive heat sources based on the thermal inertia of buildings

Abstract

The present invention relates to the technical field of energy scheduling, and discloses an optimized scheduling method for electric-driven heat sources based on the thermal inertia of buildings, including: constructing a heat load demand prediction model and a cost-benefit analysis model; predicting the initial heat energy required by the building at the current moment based on the heat load demand prediction model; using a machine learning algorithm to predict and evaluate the thermal inertia of the current building, and optimizing and adjusting the initial heat energy; evaluating the costs of meeting the adjusted required heat energy during different electricity price periods, and dynamically adjusting the scheduling strategy of the electric-driven heat source in combination with the evaluation results. The present invention can collect and analyze real-time climate data, thermal inertia data of buildings, and electricity price data, and use advanced algorithms and models to predict energy demand, optimize energy distribution strategies, and achieve the optimal utilization of energy. In this way, not only can the energy utilization efficiency be improved, the energy consumption cost be reduced, but also greenhouse gas emissions can be reduced, contributing to environmental protection.

Description

Technical Field

[0001] The present invention relates to the technical field of energy scheduling, and more specifically, to an optimized scheduling method for electric-driven heat sources based on the thermal inertia of buildings. Background Art

[0002] Globally, with the rapid economic development and continuous population growth, the demand for energy is also constantly rising. This trend has exerted great pressure on the environment and also promoted the global attention to energy conservation, emission reduction, and sustainable development. Against this background, building energy conservation is particularly important because buildings are not only places for human life and work but also major energy consumers. According to statistics, buildings consume approximately 40% of the global energy and generate nearly 20% of greenhouse gas emissions. Therefore, improving building energy efficiency and reducing energy consumption are of crucial significance for achieving the sustainable development of global energy and the environment.

[0003] In the field of building energy conservation, the optimization of energy management systems is the key. Traditional building energy management systems often rely on empirical energy distribution strategies, which are usually formulated based on historical data and fixed rules. However, this method lacks the ability to dynamically respond to real-time climate conditions, building thermal inertia, and electricity price fluctuations. Changes in real-time climate conditions, such as temperature, humidity, and wind speed, directly affect the heat load demand of buildings. The thermal inertia of buildings, that is, the resistance of buildings to temperature changes, also affects energy demand. In addition, the frequent and drastic fluctuations in electricity prices make energy costs an uncertain factor. The changes in these factors all require the energy management system to be adjusted and optimized in real time to achieve the optimal allocation and use of energy.

[0004] Therefore, in order to improve building energy efficiency, reduce energy consumption, and reduce operating costs, while taking into account the requirements of environmental protection, it is necessary to develop a dynamic energy management method that can respond in real time to changes in climate conditions, building thermal inertia, and electricity price fluctuations. Summary of the Invention

[0005] In view of the problems in the related art, the present invention proposes an optimized scheduling method for electric-driven heat sources based on the thermal inertia of buildings to overcome the above-mentioned technical problems existing in the existing related technologies.

[0006] To this end, the specific technical solutions adopted by the present invention are as follows:

[0007] An optimized scheduling method for electric-driven heat sources based on the thermal inertia of buildings, comprising the following steps:

[0008] S1. Construct a heat load demand prediction model using historical climate condition data and the thermal energy data of the building, and construct a cost-benefit analysis model in combination with electricity price data and the operating cost of the electric-driven heat source;

[0009] S2. Based on the heat load demand prediction model, combine the real-time climate condition data and the building thermal energy usage data at the previous moment to predict the initial thermal energy required by the building at the current moment;

[0010] S3. Use machine learning algorithms to predict and evaluate the thermal inertia of the current building, and optimize and adjust the initial thermal energy according to the thermal inertia of the building to obtain the adjusted required thermal energy;

[0011] S4. Evaluate the costs of meeting the adjusted required thermal energy during different electricity price periods through a cost-benefit analysis model, and dynamically adjust the scheduling strategy of the electric drive heat source based on the fluctuation range of the indoor temperature and the evaluation results.

[0012] Preferably, constructing the heat load demand prediction model using the historical climate condition data and the building thermal energy data, and constructing the cost-benefit analysis model in combination with the electricity price data and the operating cost of the electric drive heat source includes the following steps:

[0013] S11. Based on the historical database, respectively obtain the climate condition data of the building location, the corresponding thermal energy usage data, the electricity price data in the same period, and the operating cost data of the electric drive heat source, and preprocess the obtained data;

[0014] S12. Construct a heat load demand prediction model, and use the preprocessed historical climate conditions and building thermal energy usage data for training and verification to obtain the trained heat load demand prediction model;

[0015] S13. Construct a cost-benefit analysis model, and use the preprocessed historical electricity price data and the operating cost data of the electric drive heat source for training and verification to obtain the trained cost-benefit analysis model.

[0016] Preferably, the construction of the cost-benefit analysis model includes:

[0017] Define costs and benefits: Determine that the costs include electricity charges, equipment depreciation, and maintenance costs, and the benefits include energy conservation and environmental benefits.

[0018] Establish a cost function: Establish a cost function based on the electricity price, thermal energy demand, and equipment efficiency.

[0019] Establish a benefit function: Establish a benefit function based on the indoor temperature fluctuation range and the heat source efficiency.

[0020] Preferably, predicting the initial thermal energy required by the building at the current moment based on the heat load demand prediction model, combining the real-time climate condition data and the building thermal energy usage data at the previous moment includes the following steps:

[0021] S21. Collect the real-time climate condition data of the area where the current building is located and the thermal energy usage data of the building at the previous moment.

[0022] S22. Clean the collected data, select the features related to the heat load demand based on the cleaned data, and standardize the selected feature data.

[0023] S23. Based on the standardized real-time climate condition data and the building's thermal energy usage data at the previous moment, use the heat load demand prediction model to predict the initial thermal energy required by the building at the current moment.

[0024] Preferably, the steps of predicting and evaluating the thermal inertia of the current building using a machine learning algorithm and optimizing and adjusting the initial thermal energy according to the thermal inertia of the building to obtain the adjusted required thermal energy include the following steps:

[0025] S31. Build a thermal inertia prediction model based on historical data and optimize the thermal inertia prediction model using the quantum genetic algorithm.

[0026] S32. Use the optimized thermal inertia prediction model to predict the thermal inertia data of the current building in combination with the characteristic parameters of the current building.

[0027] S33. Based on the immune optimization algorithm, solve the optimal thermal energy required by the current building in combination with the thermal inertia data, real-time climate data, and electricity price data of the current building.

[0028] S34. Optimize and adjust the initial thermal energy according to the optimal thermal energy required by the current building to obtain the required thermal energy of the current building.

[0029] Preferably, the steps of building a thermal inertia prediction model based on historical data and optimizing the thermal inertia prediction model using the quantum genetic algorithm include the following steps:

[0030] S311. Collect the historical characteristic parameters of the building and the corresponding thermal inertia data, and perform preprocessing. Among them, the characteristic parameters include building material data, envelope structure data, and building orientation data.

[0031] S312. Build a grey neural network model and optimize and adjust the grey neural network model using the quantum genetic algorithm to obtain an optimized grey neural network model.

[0032] S313. Train the optimized grey neural network model using the preprocessed historical characteristic parameters of the building and the corresponding thermal inertia data to obtain a trained grey neural network model.

[0033] Preferably, the steps of constructing the grey neural network model and optimizing and adjusting the grey neural network model by using the quantum genetic algorithm to obtain the optimized grey neural network model are as follows:

[0034] S3121. Design the hierarchical structure of the grey neural network, the number of neurons in each layer, the number and size of the hidden layers according to the number of input and output variables, and construct the grey neural network model;

[0035] S3122. Randomly initialize the parameters of the grey neural network, including weights and thresholds, and determine the population size, the number of qubits and the rotation angle of the quantum genetic algorithm;

[0036] S3123. Encode the parameters of the grey neural network using qubits, and define the mean square error as the fitness function, where the probability amplitude of each qubit represents a potential value of a parameter;

[0037] S3124. Randomly generate an initial population, where each individual in the population represents a set of parameters of the grey neural network, and the qubit encoding of each individual represents a set of parameter values;

[0038] S3125. Based on the fitness function, evaluate the fitness of each individual in the initial population, and record the fitness value and the corresponding parameters of each individual;

[0039] S3126. Select excellent individuals for reproduction according to the fitness values, perform crossover operations on the selected individuals to generate new individuals, and perform quantum rotation gate operations on the new individuals to achieve mutation;

[0040] S3127. Evaluate the fitness of each individual in the new population, and update the optimal individual and the best fitness value;

[0041] S3128. Determine whether the preset number of iterations or the fitness threshold is reached. If not, return to step S3126; if so, execute the next step;

[0042] S3129. Obtain the optimal parameters based on the final population, and reconstruct the grey neural network model using the optimal parameters obtained by the quantum genetic algorithm to obtain the optimized grey neural network model.

[0043] Preferably, the expression for performing quantum rotation gate operations on new individuals to achieve mutation is:

[0044]

[0045] where α i and β i respectively represent the two complex coefficients of the original qubit state; θ iRepresents the rotation angle associated with the i-th qubit; α i ′ and β i ′ respectively represent the two complex coefficients of the new qubit state after quantum gate operation.

[0046] Preferably, the method for solving the optimal thermal energy required by the current building based on the immune optimization algorithm, combined with the thermal inertia data, real-time climate data, and electricity price data of the current building, includes the following steps:

[0047] S331. Obtain the thermal inertia data, real-time climate data, and electricity price data of the current building, and perform cleaning, standardization, and normalization processing;

[0048] S332. Based on the thermal inertia, real-time climate, and electricity price data of the building, define minimizing the energy consumption cost as the optimization objective, and use the comfort requirements inside the building, energy usage limitations, and electricity price fluctuations as constraint conditions;

[0049] S333. Encode the thermal energy supply plan as antibodies, each antibody representing a thermal energy distribution strategy, and design an antibody affinity evaluation function for evaluating the pros and cons of the thermal energy distribution strategy according to the optimization objective and constraint conditions;

[0050] S334. Design a clone mutation operator for generating new antibodies and a concentration inhibition operator for maintaining population diversity, and select antibodies with high affinity and low concentration;

[0051] S335. Randomly generate a preset number of antibodies to form an initial population;

[0052] S336. Clone and mutate the antibodies in the population to generate new antibodies;

[0053] S337. Use the antibody affinity evaluation function to calculate the affinity of the newly generated antibodies, and select a new generation of population according to the affinity and concentration;

[0054] S338. Determine whether the predetermined number of iterations, affinity threshold, or time limit is reached. If not, return to step S336. If so, output the antibody with the highest affinity, obtain the optimal thermal energy distribution strategy, and use the result of the optimal thermal energy distribution strategy as the optimal thermal energy required by the current building.

[0055] Preferably, the expression of the antibody affinity evaluation function is:

[0056] F = ω1·(C·ΣE i ) + ω2·[f(T avg , H avg )] + ω3·max[0, (E max - ∑E i )]

[0057] In the formula, F represents the value of the antibody affinity evaluation function; ω1, ω2, and ω3 respectively represent the weights of the energy consumption cost, the comfort index, and the constraint conditions; C represents the unit energy cost; E i represents the energy consumption in the i-th time period; f(T avg ,H avg ) represents the comfort index; T avg represents the average temperature; H avg represents the average humidity; max(0, E max -∑E i ) represents the penalty term, and E max represents the upper limit of energy use.

[0058] Compared with the prior art, the present invention provides an optimized scheduling method for electric drive heat sources based on the thermal inertia of buildings, and has the following beneficial effects:

[0059] (1) The present invention can collect and analyze real-time climate data, thermal inertia data of buildings, and electricity price data, and use advanced algorithms and models to predict energy demand, optimize energy distribution strategies, and achieve optimal utilization of energy. In this way, not only can the energy utilization efficiency be improved, the energy consumption cost be reduced, but also greenhouse gas emissions can be reduced, contributing to environmental protection.

[0060] (2) The present invention can significantly reduce the heating cost within the allowable range of room temperature. Through accurate prediction and real-time adjustment, it can maximize the use of low-price electricity periods for heating while ensuring the indoor temperature comfort, and reduce energy demand by releasing the thermal inertia of the building during peak electricity price periods, thereby achieving the optimization of energy consumption. This refined energy management helps to reduce unnecessary energy waste, improve energy utilization efficiency, and thus reduce the overall operating cost.

[0061] (3) The present invention can not only significantly improve the energy utilization efficiency through accurate heat load prediction and dynamic thermal energy optimization adjustment, but also respond to changes in climate conditions and electricity price fluctuations in real time, improve the flexibility and adaptability of energy management. At the same time, it can combine cost-benefit analysis to minimize the energy consumption cost, thereby effectively reducing the energy consumption cost of buildings, improving energy utilization efficiency, enhancing the environmental adaptability and intelligent management level of buildings, and is of great significance for promoting the sustainable development of the building industry.

[0062] (4) By using the quantum genetic algorithm to optimize the model, the prediction accuracy and generalization ability of the model can be improved, ensuring accurate prediction results under different environments and conditions. In addition, by using the immune optimization algorithm to solve the optimal thermal energy required by the current building in combination with the thermal inertia data, real-time climate data, and electricity price data, complex constraint conditions and multi-objective optimization problems can be effectively handled, ensuring that under the constraints of meeting the internal comfort of the building, energy use limitations, and electricity price fluctuations, the thermal energy distribution strategy with the lowest energy consumption cost is found, thereby not only improving energy utilization efficiency, but also reducing operating costs, while enhancing the robustness and adaptability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The drawings described herein are used to provide a further understanding of the present application and form a part of the present application. The illustrative embodiments and descriptions thereof of the present application are used to explain the present application and do not constitute an improper limitation of the present application. In the drawings:

[0064] Figure 1 is a flowchart of an optimized scheduling method for an electric-driven heat source based on building thermal inertia according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0065] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0066] According to an embodiment of the present invention, an optimized scheduling method for an electric-driven heat source based on building thermal inertia is provided.

[0067] The present invention will be further described below in conjunction with the drawings and specific embodiments. As Figure 1 shown, according to an embodiment of the present invention, an optimized scheduling method for an electric-driven heat source based on building thermal inertia is provided, including the following steps:

[0068] S1. Construct a heat load demand prediction model by using historical climate condition data and the thermal energy data of the building, and construct a cost-benefit analysis model in combination with electricity price data and the operating cost of the electric-driven heat source;

[0069] Among them, the step of constructing a heat load demand prediction model by using historical climate condition data and the thermal energy data of the building, and constructing a cost-benefit analysis model in combination with electricity price data and the operating cost of the electric-driven heat source includes the following steps:

[0070] S11. Based on the historical database, obtain the climate condition data (including temperature, humidity, wind speed, etc.) of the area where the building is located, the corresponding heat energy usage data (including indoor temperature records, heat energy consumption, etc.), the electricity price data of the same period (different time periods, such as changes in electricity prices during peak hours and valley hours) and the operating cost data of the electric drive heat source (including electricity charges, equipment depreciation, maintenance costs, etc.), and pre-process the obtained data;

[0071] Data preprocessing includes:

[0072] Data cleaning: Clean the collected data and remove invalid or abnormal data.

[0073] Data standardization: Standardize data from different sources into a unified format to ensure data consistency and comparability.

[0074] Feature engineering: Extracting features from raw data that are useful for predictive models.

[0075] S12, constructing a heat load demand prediction model, and using the pre-processed historical climate conditions and building heat energy usage data for training and verification to obtain a trained heat load demand prediction model;

[0076] Specifically, the heat load demand prediction model in this embodiment can be any one of linear regression, time series analysis, and machine learning models (such as random forest, gradient boosting machine, etc.).

[0077] S13. Construct a cost-benefit analysis model, and use the pre-processed historical electricity price data and the operating cost data of the electric drive heat source for training and verification to obtain a trained cost-benefit analysis model.

[0078] Specifically, the construction of the cost-benefit analysis model includes:

[0079] Define costs and benefits: Determine costs including electricity, equipment depreciation and maintenance costs, and benefits including energy savings and environmental benefits.

[0080] Establish cost function: Establish cost function based on electricity price, heat demand and equipment efficiency.

[0081] The cost function Y can include electricity costs, equipment depreciation and maintenance costs. Assuming E is the heat demand, p is the electricity price, t is the time, D is the equipment depreciation cost, M is the maintenance cost, and η is the equipment efficiency, the cost function can be expressed as:

[0082] Y(t)=[E(t) / η]×p(t)+D+M;

[0083] Establish benefit function: Establish benefit function based on indoor temperature fluctuation range and heat source efficiency.

[0084] The benefit function Z can include energy conservation and environmental benefits. Assume that S is the amount of energy conserved, S(t) = a × [E max - E(t)], V is the value of environmental benefits, V(t) = b × [T set - T actual (t)] 2 , T set is the set indoor temperature, and T actual is the actual indoor temperature. Then the benefit function can be expressed as:

[0085] Z(t) = a × [E max - E(t)] + b × [T set - T actual (t)] 2 + V(t);

[0086] In the formula, a is the value coefficient of energy conservation, b is the value coefficient of environmental benefits, and T actual (t) is the actual indoor temperature at time t.

[0087] In this embodiment, the relationship between the heat load demand prediction model and the cost - benefit analysis model is as follows:

[0088] The heat load demand prediction model and the cost - benefit analysis model are interdependent. The heat load prediction provides an estimate of energy demand, while the cost - benefit analysis provides an economic evaluation under different energy - use strategies. For example, if the prediction model predicts a low heat energy demand during a certain period, the cost - benefit analysis model will recommend reducing energy use during this period to reduce costs. Conversely, if the prediction model predicts a high heat energy demand, the cost - benefit analysis model will recommend increasing energy reserves during low - electricity - price periods to reduce long - term costs.

[0089] S2. Based on the heat load demand prediction model, combined with real - time climate condition data and the building's heat energy usage data at the previous moment, predict the initial heat energy required by the building at the current moment;

[0090] Among them, the prediction of the initial heat energy required by the building at the current moment based on the heat load demand prediction model, combined with real - time climate condition data and the building's heat energy usage data at the previous moment, includes the following steps:

[0091] S21. Collect real - time climate condition data of the current building's location and the building's heat energy usage data at the previous moment;

[0092] S22. Clean the collected data, select features related to heat load demand based on the cleaned data, and perform standardization processing on the selected feature data;

[0093] S23. Based on the real-time climate condition data after standardization processing and the building thermal energy usage data at the previous moment, use the heat load demand prediction model to predict the initial thermal energy required by the building at the current moment.

[0094] S3. Use machine learning algorithms to predict and evaluate the thermal inertia of the current building, and optimize and adjust the initial thermal energy according to the thermal inertia of the building to obtain the adjusted required thermal energy.

[0095] Specifically, the thermal inertia of a building refers to the ability of the building to resist temperature changes, that is, its thermal stability when absorbing or releasing heat. Thermal inertia has an important impact on the energy efficiency and indoor comfort of the building. The following are some important factors related to the thermal inertia of the building:

[0096] Building materials: The thermal conductivity, specific heat capacity, and density of building materials are key factors affecting thermal inertia. For example, materials such as concrete, bricks, and stones have higher thermal inertia, while wood and lightweight materials have lower thermal inertia.

[0097] Envelope structure: The thickness of the walls, roof, and floor of the building, the design of the insulation layer, and the construction quality will all affect its thermal inertia.

[0098] Building orientation: The orientation of the building and the shading design of the windows can affect the absorption of solar radiation, thereby affecting thermal inertia.

[0099] Among them, the step of using machine learning algorithms to predict and evaluate the thermal inertia of the current building, and optimizing and adjusting the initial thermal energy according to the thermal inertia of the building to obtain the adjusted required thermal energy includes the following steps:

[0100] S31. Build a thermal inertia prediction model based on historical data, and use the quantum genetic algorithm to optimize the thermal inertia prediction model.

[0101] Specifically, the step of building a thermal inertia prediction model based on historical data and using the quantum genetic algorithm to optimize the thermal inertia prediction model includes the following steps:

[0102] S311. Collect the historical characteristic parameters of the building and the corresponding thermal inertia data, and perform preprocessing. Among them, the characteristic parameters include building material data, envelope structure data, and building orientation data.

[0103] S312. Build a grey neural network model, and use the quantum genetic algorithm to optimize and adjust the grey neural network model to obtain an optimized grey neural network model. The step of building a grey neural network model and using the quantum genetic algorithm to optimize and adjust the grey neural network model to obtain an optimized grey neural network model includes the following steps:

[0104] S3121. Design the hierarchical structure of the grey neural network, the number of neurons in each layer, the number and size of the hidden layers according to the number of input and output variables, and construct a grey neural network model;

[0105] A grey neural network (GNN) is a model that combines grey system theory and neural network theory, used to handle problems with partially known information and partially unknown information (i.e., grey information). When designing the hierarchical structure of a grey neural network, the number of input and output variables, the complexity of the problem, and the characteristics of the data need to be considered. The following is a simplified step to guide how to design the hierarchical structure of a grey neural network:

[0106] 1) Determine the input and output variables

[0107] · Input variables: According to the specific situation of the problem, determine which variables are used as the input of the network. For example, in this embodiment, these variables are the real-time climate condition data of the area where the current building is located and the thermal energy usage data of the building at the previous moment.

[0108] · Output variables: Similarly, determine the variables that the network needs to predict or output. For example, in this embodiment, these variables are the initial thermal energy required by the building at the current moment.

[0109] 2) Design the network hierarchical structure

[0110] · Input layer: The number of neurons in the input layer is the same as the number of input variables.

[0111] · Hidden layer: The number and size of the hidden layers depend on the complexity of the problem. Generally speaking, one or two hidden layers can handle most problems. The number of neurons in the hidden layer is usually between the input layer and the output layer.

[0112] · Output layer: The number of neurons in the output layer is the same as the number of output variables.

[0113] 3) Determine the size of the hidden layer

[0114] · Empirical formula: An empirical formula can be used to estimate the number of neurons in the hidden layer, such as where N h is the number of neurons in the hidden layer, N i is the number of neurons in the input layer, N o is the number of neurons in the output layer, and m is a constant between 0 and 1.

[0115] · Trial-and-error method: By experimenting with different hidden layer sizes, select the network structure that can provide the best performance.

[0116] 4) Construct a grey neural network model

[0117] · Grey generator: Introduce a grey generator, such as an accumulative generating operator (AGO) or a decrement generating operator (AGO), into the network to process grey information.

[0118] · Connection weights: Initialize the connection weights and thresholds of the network.

[0119] · Train the network: Use the training data to train the network and adjust the weights and thresholds until the network performance meets the requirements.

[0120] S3122. Randomly initialize the parameters of the grey neural network, including weights and thresholds, and determine the parameters such as the population size, the number of qubits, and the rotation angle of the quantum genetic algorithm;

[0121] When designing a grey neural network and using the quantum genetic algorithm (QGA) for parameter optimization, it is necessary to initialize the weights and thresholds of the network and determine the relevant parameters of the QGA. The following is a simplified step to guide how to set these parameters:

[0122] 1) Randomly initialize the parameters of the grey neural network

[0123] · Weights and thresholds: Usually, the weights and thresholds of the neural network are randomly assigned small values during initialization, and these values are usually between -1 and 1. For example, a uniform distribution U(-1, 1) can be used to initialize the weights.

[0124] 2) Determine the parameters of the quantum genetic algorithm

[0125] · Population size: The population size P usually depends on the complexity of the problem. A common starting point is to select a smaller population, such as P = 20 to P = 50.

[0126] · Number of qubits: The number of qubits N q should match the number of parameters of the grey neural network. For example, if the network has N w weights and N b thresholds, then

[0127] N q = N w + N b .

[0128] · Rotation angle: The rotation angle θ controls the intensity of the quantum rotation gate operation. A common practice is to use a smaller angle, such as θ ∈ [0.01, 0.1] radians.

[0129] S3123. Use qubits to encode the parameters of the gray neural network, and define the mean squared error as the fitness function. Here, the probability amplitude of each qubit represents a potential value of a parameter.

[0130] In the quantum genetic algorithm (QGA), when using qubits to encode the parameters of the gray neural network, the probability amplitude of each qubit can represent a potential value of a parameter. In this case, the mean squared error (MSE) is defined as the fitness function to evaluate the performance of the network parameters.

[0131] Suppose there is a gray neural network with parameters including weights w i and thresholds b j , where i = 1, …, N w and j = 1, …, N b . The network has N training samples, and y k is the true output of the k-th sample, and

[0132] is the output predicted by the network. i or b j can be represented by the probability amplitude α i or β j of a qubit, where α i , β j are complex numbers satisfying ∣α i ∣ 2 + ∣β j ∣ 2 = 1.

[0133] The expression of the fitness function can be defined as:

[0134]

[0135] where:

[0136] · F(α, β) is the fitness function, which depends on the probability amplitudes α and β of all qubits.

[0137] · y k is the true output of the k-th sample.

[0138] · is the network predicted output of the k-th sample, which can be calculated from the parameters encoded by qubits.

[0139] Specifically for the parameters encoded by qubits, the network output can be expressed as:

[0140]

[0141] where f is the activation function of the network, and x ki is the value of the i-th input in the k-th sample.

[0142] In practical applications, since the probability amplitudes of qubits are complex numbers, we need to convert them into real parameters w i and b j to calculate the network output. This usually involves a quantum measurement process, in which the probability amplitudes are converted into actual values.

[0143] S3124. Randomly generate an initial population, where each individual in the population represents a set of parameters of the grey neural network, and the qubit encoding of each individual represents a set of parameter values;

[0144] S3125. Based on the fitness function, evaluate the fitness of each individual in the initial population, and record the fitness value and the corresponding parameters of each individual;

[0145] S3126. Select excellent individuals for reproduction according to the fitness values (methods such as roulette wheel selection and tournament selection can be used), perform crossover operations on the selected individuals to generate new individuals, and perform quantum rotation gate operations on the new individuals to achieve mutation;

[0146] The expression for performing quantum rotation gate operations on the new individuals to achieve mutation is:

[0147]

[0148] In the formula, α i and β i respectively represent the two complex coefficients of the original qubit state; θ i represents the rotation angle associated with the i-th qubit; α i ′ and β i ′ respectively represent the two complex coefficients of the new qubit state after the quantum gate operation.

[0149] S3127. Evaluate the fitness of each individual in the new population, and update the optimal individual and the best fitness value;

[0150] S3128. Determine whether the preset number of iterations (for example: the preset number of iterations is 100 to 500 times) or the fitness threshold (for example: the fitness threshold is 98% of the expected minimum cost) is reached. If not, return to step S3126; if so, execute the next step;

[0151] S3129. Obtain the optimal parameters based on the final population, and reconstruct the grey neural network model using the optimal parameters obtained by the quantum genetic algorithm to obtain an optimized grey neural network model.

[0152] S313. Train the optimized grey neural network model using the historical characteristic parameters of the preprocessed building and the corresponding thermal inertia data to obtain the trained grey neural network model.

[0153] S32. Use the optimized thermal inertia prediction model to predict the thermal inertia data of the current building in combination with the characteristic parameters of the current building;

[0154] S33. Based on the immune optimization algorithm, solve for the optimal thermal energy required for the current building in combination with the thermal inertia data, real-time climate data, and electricity price data of the current building;

[0155] Specifically, the process of solving for the optimal thermal energy required for the current building based on the immune optimization algorithm in combination with the thermal inertia data, real-time climate data, and electricity price data of the current building includes the following steps:

[0156] S331. Obtain the thermal inertia data, real-time climate data, and electricity price data of the current building, and perform cleaning, standardization, and normalization processing;

[0157] S332. Based on the thermal inertia, real-time climate, and electricity price data of the building, define minimizing the energy consumption cost as the optimization objective, and use the comfort requirements inside the building, energy usage restrictions, and electricity price fluctuations as the constraint conditions;

[0158] Specifically, in the case of the thermal inertia, real-time climate, and electricity price data of the building, define the following optimization objective and constraint conditions:

[0159] Optimization objective: Minimize the energy consumption cost

[0160] The optimization objective can be expressed as minimizing the total energy consumption cost, which is usually related to the electricity bill expenditure. Assume E t represents the energy consumption (such as electricity) at time t, p t represents the electricity price at time t, and T is the total length of time considered (e.g., the number of hours in a day or the number of days in a week), then the optimization objective can be expressed as:

[0161]

[0162] Constraint conditions:

[0163] 1) Comfort requirements inside the building:

[0164] Assume T min and T max are respectively the lowest and highest temperatures allowed inside the building, and T i (t) is the temperature of the i-th room at time t, then the comfort constraint can be expressed as:

[0165]

[0166] 2) Energy usage limit:

[0167] Assume E max is the maximum energy consumption allowed at any time t, then the energy usage limit can be expressed as:

[0168]

[0169] 3) Electricity price fluctuation:

[0170] Electricity price fluctuation is usually not controlled by us, but we need to consider it in the optimization process. This condition is usually not directly expressed in a mathematical form, but by using the actual electricity price data p t to affect the optimization objective.

[0171] 4) Thermal inertia constraint:

[0172] Assume τ is the thermal inertia parameter of the building, and ΔT(t) is the temperature difference between indoors and outdoors at time t, then the thermal inertia constraint can be expressed as:

[0173]

[0174] This means that the indoor temperature change cannot exceed the range allowed by the building's thermal inertia.

[0175] 5) Other constraints:

[0176] According to the actual situation, there may be other constraints in this embodiment, such as the operation limit of equipment, maintenance plan, etc.

[0177] S333. Encode the heat energy supply plan into antibodies, each antibody representing a heat energy distribution strategy, and design an antibody affinity evaluation function for evaluating the pros and cons of the heat energy distribution strategy according to the optimization objective and constraint conditions;

[0178] The expression of the antibody affinity evaluation function is:

[0179] F = ω1·(C·ΣE i ) + ω2·[f(T avg , H avg )] + ω3·max[0, (E max - ∑E i )]

[0180] In the formula, F represents the value of the antibody affinity evaluation function; ω1, ω2, and ω3 respectively represent the weights of energy consumption cost, comfort index, and constraint conditions; C represents the unit energy cost; E i represents the energy consumption in the i-th time period; f(T avg , H avg ) represents the comfort index; Tavg represents the average temperature; H avg represents the average humidity; max(0, E max -∑E i ) represents the penalty term, E max represents the upper limit of energy usage.

[0181] S334. A clone mutation operator designed to generate new antibodies, such as swapping certain elements in the encoding, and a concentration inhibition operator designed to maintain population diversity, selecting antibodies with high affinity and low concentration;

[0182] S335. Randomly generate a preset number of antibodies to form an initial population;

[0183] S336. Clone and mutate the antibodies in the population to produce new antibodies;

[0184] S337. Use the antibody affinity evaluation function to calculate the affinity of the newly generated antibodies, and select the new generation of population according to the affinity and concentration;

[0185] Specifically, the clone mutation operator designed to generate new antibodies and the concentration inhibition operator designed to maintain population diversity are key steps in the Immune Algorithm (IA). The immune algorithm is a search algorithm that simulates the biological immune system and is used to solve optimization problems. The following are the steps on how to design these operators:

[0186] 1) Clone mutation operator

[0187] The clone mutation operator is used to generate new antibodies by swapping certain elements in the encoding to simulate the antibody mutation process in the biological immune system.

[0188] Selection: Select antibodies with high affinity from the population for cloning.

[0189] Cloning: Clone the selected antibodies to produce multiple copies.

[0190] Mutation: Mutate the cloned antibodies. The mutation can be random or targeted. For example, certain elements in the encoding can be randomly selected for swapping, or specific strategies can be used to guide the mutation.

[0191] Re-evaluation: Re-evaluate the mutated antibodies to determine their affinity.

[0192] 2) Concentration inhibition operator

[0193] The concentration inhibition operator is used to maintain population diversity and prevent the algorithm from converging to a local optimal solution prematurely.

[0194] Calculation of concentration: For each antibody in the population, calculate its concentration, i.e., the number of antibodies similar to it.

[0195] Selection for inhibition: Select antibodies with high concentration for inhibition. This can be achieved by random selection or based on a certain strategy (such as selecting the antibody with the highest concentration).

[0196] Replacement: Replace the inhibited antibodies with new antibodies, which can be randomly generated or generated by other operators (such as crossover and mutation).

[0197] 3) Select antibodies with high affinity and low concentration

[0198] When selecting a new population, antibodies with high affinity and low concentration should be selected to ensure the diversity and search ability of the population.

[0199] Affinity calculation: Calculate the affinity of each antibody, which is usually done by evaluating the performance of the antibody in the problem domain.

[0200] Concentration calculation: Calculate the concentration of each antibody.

[0201] Selection: Select antibodies with high affinity and low concentration to form a new population.

[0202] In practical applications, the specific implementation of these operators may be more complex and needs to be adjusted according to the characteristics of the specific problem. The immune algorithm can effectively search for the optimal solution in a complex and dynamic environment by simulating the mechanism of the biological immune system.

[0203] S338. Determine whether the predetermined number of iterations (e.g., the number of iterations is 1000 - 1500 times), affinity threshold (e.g., the affinity threshold is 90% of the optimal solution), or time limit (e.g., 2 - 5 hours) is reached. If not, return to step S336; if so, output the antibody with the highest affinity, obtain the optimal heat energy distribution strategy, and use the result of the optimal heat energy distribution strategy as the optimal heat energy required by the current building.

[0204] S34. Optimally adjust the initial heat energy according to the optimal heat energy required by the current building to obtain the required heat energy of the current building.

[0205] Specifically, replace the initial heat energy with the optimal heat energy required by the current building to obtain the required heat energy of the current building.

[0206] S4. Evaluate the cost of meeting the adjusted required heat energy during different electricity price periods through a cost - benefit analysis model, and dynamically adjust the scheduling strategy of the electric - driven heat source based on the fluctuation range of the indoor temperature.

[0207] For example, within the comfortable range acceptable to the user, the room temperature is allowed to fluctuate within a range of ±1°C. The cost of meeting the adjusted required thermal energy during different electricity price periods is evaluated through a cost-benefit analysis model. Heating is advanced during periods of low electricity prices to increase the indoor temperature, and the thermal inertia of the building is utilized to store thermal energy. During periods of high electricity prices, the operation of the heat source is reduced to lower the indoor temperature, and the thermal inertia of the building is utilized to release thermal energy, so as to reduce the operation cost.

[0208] In summary, by means of the above technical solutions of the present invention, the present invention can collect and analyze real-time climate data, thermal inertia data of buildings, and electricity price data, and use advanced algorithms and models to predict energy demand, optimize energy distribution strategies, and achieve the optimal utilization of energy. In this way, not only can the energy utilization efficiency be improved, the energy consumption cost be reduced, but also greenhouse gas emissions can be reduced, contributing to environmental protection.

[0209] In addition, the present invention can significantly reduce the heating cost within the allowable range of room temperature. Through accurate prediction and real-time adjustment, heating can be maximally carried out during periods of low electricity prices while ensuring the comfort of the indoor temperature, and the energy demand can be reduced by releasing the thermal inertia of the building during peak electricity price periods, thereby achieving the optimization of energy consumption. This refined energy management helps to reduce unnecessary energy waste, improve energy utilization efficiency, and thus reduce the overall operation cost.

[0210] In addition, the present invention can not only significantly improve the energy utilization efficiency through accurate heat load prediction and dynamic thermal energy optimization adjustment, but also respond to changes in climate conditions and electricity price fluctuations in real time, improve the flexibility and adaptability of energy management. At the same time, it can combine cost-benefit analysis to minimize the energy consumption cost, thereby effectively reducing the energy consumption cost of buildings, improving energy utilization efficiency, enhancing the environmental adaptability and intelligent management level of buildings, and being of great significance for promoting the sustainable development of the building industry.

[0211] In addition, the present invention optimizes the model by using the quantum genetic algorithm, thereby improving the prediction accuracy and generalization ability of the model, ensuring accurate prediction results can be provided under different environments and conditions. In addition, by using the immune optimization algorithm to solve the optimal thermal energy required by the current building in combination with thermal inertia data, real-time climate data, and electricity price data, complex constraint conditions and multi-objective optimization problems can be effectively handled, ensuring that the lowest energy consumption cost thermal energy distribution strategy is found under the constraint conditions such as meeting the internal comfort of the building, energy use limitations, and electricity price fluctuations. Thus, not only can the energy utilization efficiency be improved, but also the operation cost can be reduced, while enhancing the robustness and adaptability of the system.

[0212] The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification. Those of ordinary skill in the art can understand that all or part of the steps in implementing the above-described embodiment methods can be completed by instructing relevant hardware through a program. The program can be stored in a computer-readable storage medium. When the program is executed, it includes the steps described in the above methods. The storage medium, such as: ROM / RAM, magnetic disk, optical disk, etc.

[0213] The above-described embodiments only express several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the appended claims.

Claims

1. A method for optimizing the scheduling of electric drive heat sources based on thermal inertia of buildings, characterized in that: The following steps are involved: S1. Use historical climate data and building thermal data to build a heat load demand forecasting model, and combine electricity price data and the operating cost of electric drive heat sources to build a cost-benefit analysis model; S2. Based on the heat load demand prediction model, the initial heat energy required by the building at the current moment is predicted by combining the real-time climate condition data and the building heat energy usage data at the previous moment; S3. Use machine learning algorithms to predict and evaluate the thermal inertia of the current building, and optimize and adjust the initial thermal energy according to the thermal inertia of the building to obtain the adjusted required thermal energy; The method of using a machine learning algorithm to predict and evaluate the thermal inertia of the current building, and optimizing and adjusting the initial thermal energy according to the thermal inertia of the building to obtain the adjusted required thermal energy includes the following steps: S31. Build a thermal inertia prediction model based on historical data and optimize the thermal inertia prediction model using quantum genetic algorithm; S32, using the optimized thermal inertia prediction model and combining characteristic parameters of the current building to predict thermal inertia data of the current building; S33, based on immune optimization algorithm, combined with the thermal inertia data of the current building, real-time climate data and electricity price data, solve the optimal thermal energy required by the current building; S34, optimizing and adjusting the initial thermal energy according to the optimal thermal energy required by the current building to obtain the thermal energy required by the current building; S4. Use a cost-benefit analysis model to evaluate the cost of meeting the required thermal energy after adjustment during different electricity price periods. Based on the fluctuation range of indoor temperature, dynamically adjust the scheduling strategy of the electric drive heat source in combination with the evaluation results.

2. The method for optimizing the scheduling of electric drive heat sources based on thermal inertia of buildings according to claim 1 is characterized in that: The method of using historical climate data and building thermal energy data to construct a heat load demand forecasting model, and combining electricity price data and the operating cost of an electric drive heat source to construct a cost-benefit analysis model includes the following steps: S11. Based on the historical database, obtain the climate condition data of the area where the building is located, the corresponding heat energy usage data, the electricity price data of the same period, and the operating cost data of the electric drive heat source, and pre-process the obtained data; S12, constructing a heat load demand prediction model, and using the pre-processed historical climate conditions and building heat energy usage data for training and verification to obtain a trained heat load demand prediction model; S13. Construct a cost-benefit analysis model, and use the pre-processed historical electricity price data and the operating cost data of the electric drive heat source for training and verification to obtain a trained cost-benefit analysis model.

3. The method for optimizing the scheduling of electric drive heat sources based on thermal inertia of buildings according to claim 1 is characterized in that: The construction of the cost-benefit analysis model includes: Define costs and benefits: Determine costs including electricity, equipment depreciation and maintenance costs, and benefits including energy savings and environmental benefits; Establish cost function: Establish cost function based on electricity price, heat demand and equipment efficiency; Establish benefit function: Establish benefit function based on indoor temperature fluctuation range and heat source efficiency.

4. The method for optimizing the scheduling of electric drive heat sources based on thermal inertia of buildings according to claim 1 is characterized in that: The method of predicting the initial heat energy required by the building at the current moment based on the heat load demand prediction model and combining the real-time climate condition data and the building heat energy usage data at the previous moment comprises the following steps: S21, collecting real-time climate condition data of the area where the current building is located and the thermal energy usage data of the building at the last moment; S22, cleaning the collected data, selecting features related to the heat load demand based on the cleaned data, and standardizing the selected feature data; S23. Based on the standardized real-time climate condition data and the building heat energy usage data at the previous moment, a heat load demand prediction model is used to predict the initial heat energy required by the building at the current moment.

5. The method for optimizing the scheduling of electric drive heat sources based on thermal inertia of buildings according to claim 1, characterized in that: The method of constructing a thermal inertia prediction model based on historical data and optimizing the thermal inertia prediction model using a quantum genetic algorithm comprises the following steps: S311, collecting historical characteristic parameters and corresponding thermal inertia data of the building and performing preprocessing, wherein the characteristic parameters include building material data, enclosure structure data and building orientation data; S312, constructing a grey neural network model, and optimizing and adjusting the grey neural network model using a quantum genetic algorithm to obtain an optimized grey neural network model; S313, using the pre-processed historical characteristic parameters of the building and the corresponding thermal inertia data to train the optimized grey neural network model to obtain a trained grey neural network model.

6. The method for optimizing the scheduling of electric drive heat sources based on thermal inertia of buildings according to claim 5, characterized in that: The grey neural network model is constructed and optimized by using a quantum genetic algorithm to obtain the optimized grey neural network model, which includes the following steps: S3121. According to the number of input and output variables, design the hierarchical structure of the grey neural network, the number of neurons in each layer, the number and size of hidden layers, and construct a grey neural network model; S3122. Randomly initialize the parameters of the grey neural network, including weights and thresholds, and determine the population size, number of quantum bits, and rotation angle of the quantum genetic algorithm; S3123, using quantum bits to encode the parameters of the gray neural network, and defining the mean square error as a fitness function, wherein the probability amplitude of each quantum bit represents a potential value of a parameter; S3124. Randomly generate an initial population, wherein each individual in the population represents a set of parameters of a gray neural network, and the quantum bit encoding of each individual represents a set of parameter values; S3125. Based on the fitness function, perform fitness evaluation on each individual in the initial population, and record the fitness value and corresponding parameters of each individual; S3126. Select excellent individuals for reproduction according to the fitness value, perform crossover operation on the selected individuals to generate new individuals, and perform quantum revolving door operation on the new individuals to achieve mutation; S3127, evaluate the fitness of each individual in the new population, and update the best individual and the best fitness value; S3128, determine whether the preset number of iterations or fitness threshold is reached, if not, return to step S3126, if yes, execute the next step; S3129. Obtain optimal parameters based on the final population, and reconstruct the grey neural network model using the optimal parameters obtained by the quantum genetic algorithm to obtain an optimized grey neural network model.

7. The method for optimizing the scheduling of electric drive heat sources based on thermal inertia of buildings according to claim 6 is characterized in that: The expression for performing quantum rotating door operation on the new individual to achieve mutation is: In the formula, α i and β i The two complex coefficients representing the original quantum bit state respectively; θ i represents the rotation angle associated with the i-th quantum bit; α i ′ and β i ′ represent the two complex coefficients of the new quantum bit state after the quantum gate operation.

8. The method for optimizing the scheduling of electric drive heat sources based on thermal inertia of buildings according to claim 1, characterized in that: The method of solving the optimal heat energy required by the current building based on the immune optimization algorithm and combining the thermal inertia data of the current building, the real-time climate data and the electricity price data includes the following steps: S331, obtaining thermal inertia data, real-time climate data and electricity price data of the current building, and performing cleaning, standardization and normalization processing; S332. Based on the thermal inertia of the building, real-time climate and electricity price data, define minimizing energy consumption costs as the optimization goal, and use the comfort requirements inside the building, energy usage restrictions and electricity price fluctuations as constraints; S333, encoding the heat energy supply scheme into antibodies, each antibody representing a heat energy allocation strategy, and designing an antibody affinity evaluation function for evaluating the quality of the heat energy allocation strategy according to the optimization goal and constraints; S334. Design a clonal mutation operator for generating new antibodies and a concentration suppression operator for maintaining population diversity, and select antibodies with high affinity and low concentration; S335, randomly generating a preset number of antibodies to form an initial population; S336. Clone and mutate antibodies in the population to produce new antibodies; S337, calculating the affinity of the newly generated antibodies using an antibody affinity evaluation function, and selecting a new generation population based on the affinity and concentration; S338. Determine whether the predetermined number of iterations, affinity threshold or time limit is reached. If not, return to step S336. If so, output the antibody with the highest affinity to obtain the optimal heat energy allocation strategy, and use the result of the optimal heat energy allocation strategy as the optimal heat energy required for the current building.

9. The method for optimizing the scheduling of electric drive heat sources based on thermal inertia of buildings according to claim 8, characterized in that: The expression of the antibody affinity evaluation function is: F=ω1·(C·∑E i )+ω2·[f(T avg ,H avg )]+ω3·max[0,(E max -∑E i )] In the formula, F represents the value of the antibody affinity evaluation function; ω1, ω2, and ω3 represent the weights of energy cost, comfort index, and constraint conditions, respectively; C represents unit energy cost; E i represents the energy consumption in the i-th time period; f(T avg ,H avg ) represents the comfort index; T avg represents the average temperature; H avg Indicates average humidity; max(0,E max -∑E i ) represents a penalty term; E max Indicates the upper limit of energy usage.

Images