A safe and economic based electric heating comprehensive energy control method

Abstract

The application discloses an energy system control field and discloses a safe and economic electric-thermal comprehensive energy control method, which comprises the following steps: predicting renewable energy and multi-element load in a comprehensive energy system through a pre-trained SA-PSO-BP neural network, and controlling the comprehensive energy system according to the predicted structure of the renewable energy and the multi-element load; the training process of the SA-PSO-BP neural network comprises the following steps: pre-processing and screening related features to obtain optimal features and constructing a feature training set; constructing an objective function of a BP neural network and adding power network constraints and heat network constraints of optimal scheduling; training the BP neural network through the feature training set, iteratively updating the weight and threshold of the BP neural network by using a simulated annealing optimization particle swarm SA-PSO algorithm in the training process, and obtaining the SA-PSO-BP neural network; and the application adjusts the weight in the objective function so that the control of the comprehensive energy system considers safety and economy.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of energy system control, and particularly relates to a safe and economic electric-thermal comprehensive energy control method. BACKGROUND

[0002] With the continuous development of technology and economy, various fields of social production and life have put forward greater demand for energy, and under the continuous consumption of fossil energy, energy and environmental problems have become increasingly prominent. In order to meet the demand for energy and environmental protection, it has become the strategic choice of each country to add low-carbon, clean and green renewable energy to the existing energy system. However, the randomness and intermittency of wind power and photovoltaic renewable energy greatly hinder the consumption of wind power and photovoltaic in the comprehensive energy system, so the prediction of renewable energy generation is very important.

[0003] After the comprehensive energy system of heterogeneous energy such as electricity, heat and gas is proposed and widely applied, the barriers between different energies are broken, the mutual support between energies is enhanced, and the stability of each energy supply is effectively improved. However, the current research on electric-thermal comprehensive energy system is mostly in economic optimization and improvement of renewable energy penetration, and the safety research is not sufficient. Due to the existence of some uncertain factors such as wind power, photovoltaic output and multi-load demand in the electric-thermal interconnected comprehensive energy system, the system may have safety accidents such as voltage out-of-limit. SUMMARY

[0004] The purpose of the present application is to provide a safe and economic electric-thermal comprehensive energy control method, which ensures the safe operation of the system while optimizing the economic dispatch.

[0005] To achieve the above purpose, the technical scheme adopted by the present application is:

[0006] The present application provides a safe and economic electric-thermal comprehensive energy control method, which includes:

[0007] The SA-PSO-BP neural network is used to predict the renewable energy and multi-load in the comprehensive energy system, and the comprehensive energy system is controlled according to the prediction structure of the renewable energy and multi-load;

[0008] The training process of the SA-PSO-BP neural network includes:

[0009] The related characteristics of wind power, photovoltaic and electric-thermal load in the comprehensive energy system are collected, the related characteristics are pretreated and selected to obtain the optimal characteristics and construct the feature training set;

[0010] According to the preferred features and the output power of the comprehensive energy system, a target function of the BP neural network is constructed, and power network constraints and heat network constraints of the optimized scheduling are added; the BP neural network is trained through a feature training set, and in the training process, the weights and thresholds of the BP neural network are iteratively updated by using a simulated annealing optimization particle swarm SA-PSO algorithm, so as to obtain a SA-PSO-BP neural network.

[0011] Preferably, the method for pre-processing the related features comprises:

[0012] The abnormal data in the related features of the wind power photovoltaic and electric heating load are removed by the 3δ principle, and the expression formulae are respectively:

[0013]

[0014]

[0015]

[0016] In the formula, p i represents the ith sample value of the same feature attribute, represents the sample mean, δ represents the reference standard value, n is the sample number, p e is the residual error; when the residual error p e of the related feature value is greater than 3δ, the related feature value is removed;

[0017] The missing data in the related features of the wind power photovoltaic and electric heating load is filled by the Lagrange interpolation method, and the expression formula is:

[0018]

[0019] In the formula, x i represents the time of the ith+1 value point, y i represents the feature value of the ith+1 value point; x j represents the feature value of the jth value point; and L(x) represents the feature interpolation corresponding to the given time x.

[0020] Preferably, the method for screening the related features to obtain the preferred features comprises:

[0021] The Pearson correlation coefficient is used to estimate the correlation between the related features, and the expression formula is:

[0022]

[0023] In the formula, X represents a feature value vector, Y represents an actual value vector of wind power generation or photovoltaic power generation or electric load or heat load demand, and ρ XYrepresents the degree of association between X and Y; Cov(X,Y) represents the covariance of X and Y, σ X and σ Y respectively represent the standard deviation of X and Y;

[0024] According to the degree of association ρ XY Preferred features are selected from the correlation characteristics of wind power, photovoltaic and electric heating load.

[0025] Preferably, the method for constructing the objective function of the BP neural network according to the preferred features and the output power of the integrated energy system comprises:

[0026] The absolute deviation of the voltage of each node of the integrated energy system at each time period and f1 are normalized to F1;

[0027] The purchase and sale of electricity cost C E , gas purchase cost C GAS , equipment operation cost C OP and wind and light penalty cost C GWP are integrated into economic cost f2 and normalized to F2;

[0028] The expression formula of the objective function of the BP neural network is constructed as:

[0029]

[0030] In the formula, λ1 and λ2 are the weights of F1 and F2 respectively; f1 max is the maximum value of the sum of the absolute deviation of the voltage of each node in the integrated energy system; is the maximum cost value of the output of each device in the integrated energy system; T is the total working period of the integrated energy system; N V is the number of electric nodes; is the difference between the voltage of the i-th electric node at the t-th time period and the safety boundary.

[0031] Preferably, the calculation process of the difference between the voltage of the electric node and the safety boundary comprises:

[0032]

[0033] In the formula, represents the voltage per unit of the i-th electric node at the t-th time period, V max represents the upper limit of the voltage per unit, V min represents the lower limit of the voltage per unit.

[0034] Preferably, the voltage per unit of the electric node is corrected under the influence of the heat network by using the Newton-Raphson method The calculation method comprises:

[0035] The relative injection power of the electric node is expressed as:

[0036]

[0037] P i and Q i represent the active power and reactive power injected by the i-th electric node, respectively, P chp,i and Q chp,i represent the active power and reactive power of CHP units in the i-th electric node, respectively, P es,i and Q es,i represent the active power and reactive power of batteries in the i-th electric node, respectively, P wd,i and Q wd,i represent the active power and reactive power of wind turbines in the i-th electric node, respectively, P pv,i and Q pv,i represent the active power and reactive power of photovoltaic in the i-th electric node, respectively, P eb,i and Q eb,i represent the active power and reactive power of electric boilers in the i-th electric node, respectively, P load,i and Q load,i represent the active power and reactive power of electric loads in the i-th electric node, respectively.

[0038] The power error equation of the electric node is calculated, and the formula is:

[0039]

[0040] In the formula, P is and Q is are the active power and reactive power set for the i-th electric node; V i and V j are the voltages injected into the i-th electric node and the j-th electric node, respectively; G ij , B ij and θ ij are the conductance, susceptance and phase angle difference between the i-th electric node and the j-th electric node, respectively.

[0041] The modified equation based on the simplified Newton-Raphson method is:

[0042]

[0043] According to the modified equation, the phase angle correction amount Δθ and the phase angle correction amount ΔV of the electric node are calculated, and the phase angle and voltage of the electric node are repeatedly corrected. When ΔP i and ΔQ i are both less than ε, the correction is stopped to obtain the final phase angle and final voltage of the electric node; ε represents the allowed error of the node power imbalance.

[0044] Preferably, the purchase and sale electricity cost C EThe expression formula of C is:

[0045]

[0046] In the formula, t represents the time period of electricity purchase and sale, and t represents the electricity price of electricity purchase and sale at t time period, respectively. P represents the electricity power of electricity purchase and sale, which is positive when purchasing electricity and negative when selling electricity.

[0047] Preferably, the gas purchase cost C GAS is calculated by the formula:

[0048]

[0049] In the formula, N chp represents the number of CHP units, w gas represents the price of unit electricity power generated by the CHP unit, and P represents the electricity power generated by the i-th CHP unit.

[0050] Preferably, the equipment operation cost C OP is calculated by the formula:

[0051]

[0052] In the formula, N wd , N pv , N es , N hs , and N eb represent the number of wind turbine generators, photovoltaic generators, electricity storage devices, heat storage devices, and electric boilers, respectively, O wd , O pv , O es , O chp , O hs , and O eb represent the operation cost coefficients of wind power generation, photovoltaic power generation, electricity storage devices, CHP units, heat storage devices, and electric boilers, respectively, and P represent the electricity power generated by wind power generation, photovoltaic power generation, electricity storage devices, and CHP units, respectively, and Q represent the heat power generated by heat storage devices and electric boilers, respectively.

[0053] Preferably, the penalty cost C GWP for abandoned wind and light is calculated by the formula:

[0054]

[0055] In the formula, α wd and α pvPenalty coefficients of abandoned wind and light respectively, and Predicted values of wind power and photovoltaic power respectively.

[0056] Preferably, the power network constraints added in the optimal scheduling include: active power balance constraints of the power network, bus voltage constraints, branch transmission power constraints and power network branch power loss constraints; the heat network constraints added in the optimal scheduling include heat network power balance constraints and heat network pipeline heat loss constraints.

[0057] Preferably, the method for obtaining the SA-PSO-BP neural network includes: iteratively updating the weights and thresholds of the BP neural network by using the simulated annealing optimization particle swarm (SA-PSO) algorithm.

[0058] Initializing the weights and thresholds of the BP neural network; taking the length of the weights and thresholds in the BP neural network as the dimension of the particle swarm, taking the weights and thresholds as the position of the particles, initializing the weight w, the learning rate c1 and c2, the position x and the speed v of the particle swarm, and the temperature T and the annealing coefficient K of the simulated annealing;

[0059] Taking the prediction error in the neural network training process as the fitness F of the particle swarm, giving the particles a random disturbance to obtain new particles x new If the new fitness F x is less than or equal to the existing F , the new F is accepted as the optimal fitness value;

[0060] If F and exp(-(F-F) / TK)≤rand() are established, the new F is accepted as the optimal fitness value, if F and exp(-(F-F) / TK)≤rand() are established, the existing F x is kept as the optimal fitness value; exp() represents the exponential operation with the natural logarithm base e as the base; rand() represents a random function for generating random numbers;

[0061] Iteratively updating the weight w, the learning rate c1 and c2, the position x and the speed v of the particle swarm, and the fitness F of the particle swarm; when the iteration number reaches a predetermined value, outputting the global optimal solution F g and the corresponding BP neural network, and taking the trained BP neural network as the SA-PSO-BP neural network.

[0062] Compared with the prior art, the present application has the following advantages:

[0063] The application collects the related characteristics of wind power, photovoltaic and electric heating load in the integrated energy system, pre-processes and screens the related characteristics to obtain preferred characteristics and construct a feature training set, trains the BP neural network through the feature training set, predicts the renewable energy and multi-element load in the integrated energy system through the SA-PSO-BP neural network pre-trained, controls the integrated energy system according to the prediction structure of the renewable energy and multi-element load, and the safety and stability of system operation are improved.

[0064] The application constructs the objective function of the BP neural network according to the preferred characteristics and the output power of the integrated energy system, adds the power network constraint and the heat network constraint of the optimized scheduling, trains the BP neural network through the feature training set, iteratively updates the weight and threshold of the BP neural network by using the SA-PSO algorithm in the training process, and obtains the SA-PSO-BP neural network; the application adjusts the weight in the objective function to make the control of the integrated energy system consider safety and economy. BRIEF DESCRIPTION OF DRAWINGS

[0065] Figure 1 It is a flow chart of the electric-heat integrated energy control method based on safety and economy provided by the embodiment of the application;

[0066] Figure 2 It is a flow chart of the renewable energy and multi-element load prediction based on the SA-PSO-BP neural network;

[0067] Figure 3 It is a topology chart of the safety and economy integrated design method of the electric-heat integrated energy system;

[0068] Figure 4 It is a flow chart of the objective optimization function based on the SA-PSO algorithm;

[0069] Figure 5 It is a renewable energy and multi-element load prediction chart based on the SA-PSO-BP neural network;

[0070] Figure 6 It is a power grid scheduling chart based on the safety and economy integrated method;

[0071] Figure 7 It is a heat grid scheduling chart based on the safety and economy integrated method;

[0072] Figure 8 It is a system electric node voltage chart based on the safety and economy integrated method. DETAILED DESCRIPTION

[0073] The application will be further described below with reference to the drawings. The following examples are only used to more clearly illustrate the technical solutions of the application, and cannot be used to limit the protection scope of the application.

[0074] As shown in the figure, a safe and economic based electric heat comprehensive energy control method comprises: Figures 1 to 4

[0075] The renewable energy and multi-element load in the comprehensive energy system are predicted through the SA-PSO-BP neural network trained in advance, and the comprehensive energy system is controlled according to the predicted structure of the renewable energy and multi-element load.

[0076] The training process of the SA-PSO-BP neural network comprises:

[0077] The related characteristics of wind power photovoltaic and electric heat load in the comprehensive energy system are collected, the related characteristics are preprocessed and screened to obtain the preferred characteristics and construct a feature training set.

[0078] The method for preprocessing the related characteristics comprises:

[0079] The abnormal data in the related characteristics of wind power photovoltaic and electric heat load are removed through the 3delta principle, and the expression formulae are respectively:

[0080]

[0081]

[0082]

[0083] In the formula, p i represents the i th sample value of the same characteristic attribute, represents the sample mean, delta represents the reference standard value, n is the sample number, and p e is the residual error; when the residual error p e of the related characteristic value is greater than 3delta, the related characteristic value is removed.

[0084] The missing data in the related characteristics of wind power photovoltaic and electric heat load are filled through the Lagrange interpolation method, and the expression formula is:

[0085]

[0086] In the formula, x i represents the time of the i+1 th value point, y i represents the characteristic value of the i+1 th value point; x j represents the characteristic value of the j th value point; and L(x) represents the characteristic interpolation corresponding to the given time x.

[0087] ​The method for screening the related features to obtain the preferred features comprises:

[0088] The Pearson correlation coefficient is used to estimate the correlation between the related features, and the expression formula is:

[0089]

[0090] In the formula, X represents the feature value vector, Y represents the actual value vector of wind power generation or photovoltaic power generation or electrical load or thermal load demand, and p XY represents the correlation degree between X and Y; Cov(X, Y) represents the covariance of X and Y, and X and Y respectively represent the standard deviations of X and Y; the preferred features are selected from the related features of wind power, photovoltaic power and electrical and thermal loads according to the correlation degree p XY .

[0091] The method for constructing the objective function of the BP neural network according to the preferred features and the output power of the integrated energy system comprises:

[0092] The absolute deviation of the voltage of each node of the integrated energy system at each time period and f1 are normalized to F1.

[0093] The purchase and sale of electricity cost C E , the purchase of gas cost C GAS , the equipment operation cost C OP and the penalty cost C GWP of abandoned wind and light are integrated into the economic cost f2 and normalized to F2.

[0094] The expression formula of the objective function of the BP neural network is constructed as:

[0095]

[0096] In the formula, λ1 and λ2 are the weights of F1 and F2 respectively; f1 max is the maximum value of the sum of the absolute deviations of the voltage of each node of the integrated energy system; is the maximum cost value of the output of each device of the integrated energy system; T is the total working period of the integrated energy system; N V is the number of electrical nodes; is the difference between the voltage of the i-th electrical node at the t-th time period and the safety boundary.

[0097] The calculation process of the difference between the voltage of the electrical node and the safety boundary comprises:

[0098]

[0099] In the formula, represents the voltage standard value of the i-th electrical node at the t-th time period, and V maxV represents the upper limit of the voltage per unit value. min This indicates the lower limit of the voltage per unit value.

[0100] The per-unit voltage value of the electrical node was corrected using the Newton-Raphson method under the influence of the heating network. The calculation methods include:

[0101] The relative injected power at the electrical node is expressed by the following formula:

[0102]

[0103] In the formula, P i and Q i Let P represent the active power and reactive power injected at the i-th electrical node, respectively. chp,i and Q chp,i P represents the active power and reactive power of the CHP unit in the i-th electrical node, respectively. es,i and Q es,i Let P represent the active power and reactive power of the battery in the i-th electrical node, respectively. wd,i and Q wd,i Let P represent the active power and reactive power of the wind turbine in the i-th electrical node, respectively. pv,i and Q pv,i Let P represent the active power and reactive power of the photovoltaic system in the i-th electrical node, respectively. eb,i and Q eb,i Let P represent the active power and reactive power of the electric boiler in the i-th electrical node, respectively. load,i and Q load,i Let represent the active power and reactive power of the electrical load in the i-th electrical node, respectively;

[0104] The equation for calculating the power error at electrical nodes is expressed as follows:

[0105]

[0106] In the formula, P is Q is The active and reactive power settings for electrical node i; V i and V j G represents the voltage injected into electrical nodes i and j, respectively; ij B ij and θ ij These are the conductance, susceptance, and phase angle difference between electrical nodes i and j, respectively.

[0107] The simplified correct equation based on the Newton-Rapson method is as follows:

[0108]

[0109] According to the modified equation, the phase angle correction amount Δθ and the phase angle correction amount ΔV of the electric node are calculated, and the phase angle and voltage of the electric node are repeatedly corrected. When ΔP i and ΔQ i are both less than ε, the correction is stopped to obtain the final phase angle and the final voltage of the electric node; ε represents the allowed error of the node power imbalance.

[0110] The expression formula of the purchase and sale electricity cost C E is as follows:

[0111]

[0112] In the formula, P and V respectively represent the electricity price of the purchase and sale electricity at t period, represents the electricity power of the purchase and sale electricity, which is positive when purchasing electricity and negative when selling electricity.

[0113] The calculation formula of the gas purchase cost C GAS is as follows:

[0114]

[0115] In the formula, N chp represents the number of CHP units, w gas represents the price of the unit electricity power generated by the CHP unit, represents the electricity power generated by the i-th CHP unit.

[0116] The calculation formula of the equipment operation cost C OP is as follows:

[0117]

[0118] In the formula, N wd , N pv , N es , N hs and N eb respectively represent the number of wind power generators, photovoltaic power generators, electricity storage devices, heat storage devices and electric boilers, O wd , O pv , O es , O chp , O hs and O eb respectively represent the operation cost coefficients of the wind power generation, photovoltaic power generation, electricity storage device, CHP unit, heat storage device and electric boiler, and respectively represent the electricity power generated by the wind power generation, photovoltaic power generation, electricity storage device and CHP unit, and respectively represent the heat power emitted by the heat storage device and electric boiler.

[0119] Penalty cost of curtailment of wind power and solar power C GWP The calculation formula is as follows:

[0120]

[0121] In the formula, α wd and α pv respectively represent penalty coefficients of curtailment of wind power and solar power, and respectively represent predicted values of wind power and solar power.

[0122] The power network constraint of optimal scheduling is added to the objective function, including active power balance constraint of the power network, voltage constraint of the electric node, branch transmission power constraint, power network branch power loss constraint and other power network constraints.

[0123] (1) The expression formula of the active power balance constraint of the power network is as follows:

[0124]

[0125] In the formula, N eb and N el respectively represent the number of electric boilers and electric loads, and respectively represent charging and discharging power of the electric storage device, η c and η d respectively represent charging and discharging efficiencies, represents electric power consumed by the electric boiler, represents electric power consumed by the electric load, represents power loss of the i-th branch of the power grid at the t period.

[0126] (2) Branch transmission power constraint

[0127] In order to ensure that the voltage of the electric node of the system does not exceed the safety limit, the value of the safety weight λ1 is set to be much larger than the value of the cost weight λ2 in the embodiment, that is, the voltage is ensured to be in the safety domain [V min ,V max ] for economic scheduling.

[0128] (3) Power network branch power loss constraint

[0129]

[0130] In the formula, and respectively represent the minimum and maximum values of the active power flow of the branch l. P l tPij represents the active power flow of branch j, Gij represents the power of line between node i and j. ij and B ij respectively represent the real part and imaginary part of the element in the i-th row and j-th column of the node admittance matrix. θ i represent the voltage phase angle of node i, θ ij represent the voltage phase angle difference between node i and j. i and θ j .

[0131] (4) Power network branch power loss constraint

[0132]

[0133] In the formula, P i t and are the active and reactive power injected into the i-th electric node; U0 and R i respectively represent the reference voltage of the system and the resistance of the branch connected to node i.

[0134] (5) Other constraints of the power network

[0135]

[0136] In the formula, and respectively represent the upper limit and lower limit of the output of the wind turbine generator set, and respectively represent the upper and lower limits of the output of the photovoltaic generator set, and respectively represent the upper limit and lower limit of the power output of the CHP unit, and respectively represent the upper limit and lower limit of the ramping of the CHP unit, represents the storage capacity, α hs is the heat loss rate of the thermal storage tank, and respectively represent the charging and discharging power, η c and η d respectively represent the charging and discharging efficiencies, and respectively represent the upper and lower limits of the storage capacity, and respectively represent the upper and lower limits of the charging power, and respectively represent the upper and lower limits of the discharging power, and respectively represent the initial storage capacity of the battery and the storage capacity after a period, indicating that the dispatch of the battery should meet the condition that the capacity returns to the initial state after a period, and respectively represent the upper and lower limits of the electric boiler power consumption, β eb represents the heat-to-power ratio of the electric boiler, P e max and P e min respectively represent the upper and lower limits of the purchase and sale of electricity.

[0137] The heat supply network constraints added in the optimization scheduling include heat supply network power balance constraints, heat supply network pipe heat loss constraints, and other heat supply network constraints.

[0138] (1) Heat supply network power balance constraints

[0139]

[0140] In the formula, N hl represents the number of heat loads, N pip is the number of heat network branches, and respectively represent the heat power generated by the CHP unit and the electric boiler, and respectively represent the heat storage and heat release states, and respectively represent the heat storage and heat release powers, represents the heat dissipation amount of the i-th heat network branch at time t.

[0141] (2) Heat supply network pipe heat loss constraints

[0142]

[0143] In the formula is the heat loss power of the i-th heat network pipe; l i is the length of the i-th heat network pipe; T i,t is the temperature of the hot water in the pipe; T0 is the ambient temperature outside the pipe; R1 is the thermal resistance of the heat network pipe; and R2 is the thermal resistance of the pipe insulation layer.

[0144] (3) Other heat supply network constraints

[0145]

[0146] In the formula, β chp represents the heat-to-power ratio, represents the price of natural gas, represents the electric power generated by the CHP, L ng represents the low heat value of natural gas, η chp represents the power generation efficiency of the CHP unit, Δt represents the unit scheduling duration, β eb represents the heat-to-power ratio of the electric boiler, αeb represents the heat loss rate of the electric boiler, represents the heat storage capacity, and hs represents the heat loss rate, and respectively represent the efficiency of heat storage and heat release, and respectively represent the power of heat storage and heat release, and respectively represent the upper and lower limits of the heat storage capacity, and respectively represent the upper and lower limits of the heat storage power, and respectively represent the upper and lower limits of the heat release power, and respectively represent the initial storage capacity of the battery and the storage capacity after a period, indicating that the scheduling of the battery should meet the condition that the capacity returns to the initial state after a period.

[0147] The BP neural network is trained through the characteristic training set, and the weight and threshold of the BP neural network are iteratively updated by using the simulated annealing optimization particle swarm SA-PSO algorithm in the training process to obtain the SA-PSO-BP neural network, and the specific method comprises:

[0148] initializing the weight and threshold of the BP neural network; taking the length of the weight and threshold in the BP neural network as the dimension of the particle swarm, taking the weight and threshold as the position of the particle, initializing the weight w, the learning rate c1 and c2, the position x and the speed v of the particle swarm, and the temperature T and the annealing coefficient K of the simulated annealing;

[0149] taking the prediction error in the neural network training process as the fitness F of the particle population, giving the particle a random disturbance to obtain a new particle x new , if the new fitness is less than or equal to the existing F x , accepting as the optimal fitness value;

[0150] if and exp(-(F-F) / TK)≤rand() are established, accepting as the optimal fitness value, if and exp(-(F-F) / TK)≤rand() are established, retaining the existing F x as the optimal fitness value; exp() represents the exponential operation with the natural logarithm base e as the base; rand() represents a random function for generating a random number;

[0151] Iterative update of the weight w, learning rate c1 and c2, position x and velocity v of the particle swarm, and fitness F of the particle population; when the number of iterations reaches a predetermined value, output the global optimal solution F g and the corresponding BP neural network, the trained BP neural network is used as the SA-PSO-BP neural network.

[0152] As shown in Figures 5 to 8 , the SA-PSO-BP neural network trained in advance is used to predict the renewable energy and multi-element load in the integrated energy system, the integrated energy system is controlled according to the prediction structure of the renewable energy and multi-element load, so that the safe and stable operation of the system is improved; by adjusting the weight in the objective function, the control of the integrated energy system takes into account the safety and economy

[0153] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0154] The present application is described with reference to flowcharts and / or block diagrams according to the methods, devices (systems), and computer program products of the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more flows and / or blocks.

[0155] These computer program instructions can also be stored in a computer-readable memory that can guide the computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more flows and / or blocks.

[0156] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate a computer implemented process, so that the instructions executed on the computer or other programmable data processing devices provide a process for implementing the functions specified in the flowchart Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0157] The above description is only the preferred embodiment of the present application, it should be pointed out that for those skilled in the art, without departing from the technical principles of the present application, a number of improvements and modifications can also be made, these improvements and modifications should also be considered as the protection scope of the present application.

Claims

1. A control method for an integrated electrothermal energy system, characterized in that, include: Predictions of renewable energy sources and diverse loads in an integrated energy system are made using a pre-trained SA-PSO-BP neural network. Methods for constructing the objective function of an integrated energy system include: The absolute voltage deviation and f1 of each node in the integrated energy system at each time period are normalized to F1; The cost of purchasing and selling electricity C E Gas purchase cost C GAS Equipment operating costs C OP The cost of curtailment of wind and solar power (C) GWP The total cost is f2, which is then normalized to F2. The objective function can be expressed as follows: In the formula, λ1 and λ2 are the weights of F1 and F2, respectively; f1 max This represents the maximum sum of the absolute differences in voltage deviation at each node in the integrated energy system. N represents the maximum cost of output from each device in the integrated energy system; T represents the total operating time of the integrated energy system; N represents the total operating time of the integrated energy system. V The number of electrical nodes; Let be the difference between the voltage of the i-th electrical node and the safety boundary during time period t; The difference between the electrical node voltage and the safety boundary The calculation process includes: In the formula, V represents the per-unit voltage value of the i-th electrical node during time period t. max V represents the upper limit of the voltage per unit value. min Indicates the lower limit of the voltage per unit value; The per-unit voltage value of the electrical node was corrected using the Newton-Raphson method under the influence of the heating network. The calculation methods include: The relative injected power at the electrical node is expressed by the following formula: In the formula, P i and Q i Let P represent the active power and reactive power injected at the i-th electrical node, respectively. chp,i and Q chp,i P represents the active power and reactive power of the CHP unit in the i-th electrical node, respectively. es,i and Q es,i Let P represent the active power and reactive power of the battery in the i-th electrical node, respectively. wd,i and Q wd,i Let P represent the active power and reactive power of the wind turbine in the i-th electrical node, respectively. pv,i and Q pv,i Let P represent the active power and reactive power of the photovoltaic system in the i-th electrical node, respectively. eb,i and Q eb,i Let P represent the active power and reactive power of the electric boiler in the i-th electrical node, respectively. load,i and Q load,i Let represent the active power and reactive power of the electrical load in the i-th electrical node, respectively; The equation for calculating the power error at electrical nodes is expressed as follows: In the formula, P is Q is The active and reactive power settings for electrical node i; V i and V j G represents the voltage injected into electrical nodes i and j, respectively; ij B ij and θ ij These are the conductance, susceptance, and phase angle difference between electrical nodes i and j, respectively. The simplified correct equation based on the Newton-Rapson method is as follows: The phase angle corrections Δθ and ΔV of the electrical node are calculated based on the corrected equations. The phase angle and voltage of the electrical node are then corrected repeatedly until ΔP is satisfied. i and ΔQ i Once all values ​​are less than ε, the correction is stopped to obtain the final phase angle and final voltage of the electrical node; ε represents the allowable error of the node power imbalance. The objective function is constrained by power network and thermal network for optimal scheduling. Based on the prediction results of renewable energy and multiple loads, the objective function is optimized using the simulated annealing-optimized particle swarm optimization (SA-PSO) algorithm. The integrated energy system is then controlled based on the optimized solution of the objective function. The training process of the SA-PSO-BP neural network includes: Relevant characteristics of wind power, photovoltaic power, and electrothermal loads in the integrated energy system are collected. The relevant characteristics are preprocessed and screened to obtain preferred characteristics and construct a feature training set. The topology of the BP neural network is determined based on the preferred characteristics and the output power of the integrated energy system. The BP neural network is trained using a feature training set. During the training process, the weights and thresholds of the BP neural network are iteratively updated using the simulated annealing optimized particle swarm optimization (SA-PSO) algorithm to obtain the SA-PSO-BP neural network.

2. The control method for an integrated electrothermal energy system according to claim 1, characterized in that, Methods for preprocessing relevant features include: Abnormal data in the relevant characteristics of wind power, solar power, and electrothermal loads are eliminated using the 3δ principle, expressed by the following formulas: In the formula, p i This represents the i-th sample value of the same feature attribute. δ represents the sample mean, δ represents the reference standard value, n is the number of samples, and p e The residual error; when the residual error p of the relevant eigenvalues e If the value is greater than 3δ, the relevant feature value will be removed. The missing data in the relevant characteristics of wind power, photovoltaic power, and electrothermal load are filled using Lagrange interpolation, expressed by the following formula: In the formula, x i Let y represent the time at the (i+1)th value point. i x represents the eigenvalue of the (i+1)th value point; j Let L(x) represent the feature value at the j-th point; L(x) represents the feature interpolation corresponding to a given time x.

3. The control method for an integrated electrothermal energy system according to claim 2, characterized in that, Methods for selecting preferred features by filtering relevant features include: The Pearson correlation coefficient is used to estimate the correlation between relevant features, expressed by the following formula: In the formula, X represents the eigenvalue vector, Y represents the actual value vector of wind power generation, photovoltaic power generation, electrical load, or heat load demand, and ρ XY This represents the degree of correlation between X and Y; Cov(X,Y) represents the covariance between X and Y, and σ X and σ Y Let X and Y represent the standard deviations, respectively. Based on the degree of correlation ρ XY Preferred features were selected from the relevant characteristics of wind power, photovoltaic power, and electric heating load.

4. The control method for an integrated electrothermal energy system according to claim 1, characterized in that, Equipment operating cost C OP The calculation formula is: In the formula, N wd N pv N es N hs and N eb These represent the quantities of wind turbine generators, photovoltaic generators, energy storage devices, thermal storage devices, and electric boilers, respectively. wd O pv O es O chp O hs and O eb These represent the operating cost coefficients for wind power generation, photovoltaic power generation, energy storage equipment, CHP generator sets, thermal storage equipment, and electric boilers, respectively. and These represent the electrical power generated by wind power, photovoltaic power, energy storage devices, and CHP units, respectively. and These represent the heat output of the thermal storage equipment and the electric boiler, respectively.

5. The control method for an integrated electrothermal energy system according to claim 4, characterized in that, The cost of curtailing wind and solar power (C) GWP The calculation formula is: In the formula, α wd and α pv These represent the penalty coefficients for wind curtailment and solar curtailment, respectively. and These represent the predicted values ​​for wind power and solar power, respectively.

6. The control method for an integrated electrothermal energy system according to claim 1, characterized in that, The constraints added for optimized scheduling of the power network include: active power balance constraints, node voltage constraints, branch transmission power constraints, and branch power loss constraints; the constraints added for optimized scheduling of the thermal network include thermal network power balance constraints and thermal network pipeline heat loss constraints.

7. A control method for an integrated electrothermal energy system according to claim 1 or claim 6, characterized in that, Methods for obtaining an SA-PSO-BP neural network by iteratively updating the weights and thresholds of a BP neural network using a simulated annealing-optimized particle swarm optimization (SA-PSO) algorithm include: Initialize the weights and thresholds of the BP neural network; use the lengths of the weights and thresholds in the BP neural network as the dimensions of the particle swarm, and the weights and thresholds as the positions of the particles. Initialize the weights w, learning rates c1 and c2, positions x and velocities v of the particle swarm, as well as the simulated annealing temperature T and annealing coefficient K. The prediction error during the neural network training process is used as the fitness F of the particle population. A random perturbation is applied to each particle to obtain a new particle x. new If the new fitness Less than or equal to the existing F x ,accept As the optimal fitness value; If F xnew >F x If exp(-(FF) / TK)≤rand() holds true, then accept F. xnew As the optimal fitness value, if F xnew >F x And exp(-(FF) / TK)≤rand() holds true, so retain the existing F. x As the optimal fitness value; exp() represents the exponential operation with the natural logarithm base e as the base; rand() represents the random function that generates random numbers; Iteratively update the particle swarm's weights w, learning rates c1 and c2, position x and velocity v, and the particle swarm's fitness F; when the number of iterations reaches a predetermined value, output the global optimal solution F. g The corresponding BP neural network is used as the SA-PSO-BP neural network after training.

Images