Power and heat combined supply type virtual power plant optimization scheduling method based on photo-thermal power generation technology

Abstract

The invention discloses a photo-thermal power generation technology-based power and heat combined supply type virtual power plant optimization scheduling method. The method comprises the steps of constructing a multi-energy virtual power plant model comprising wind power generation, photovoltaic power generation, a photo-thermal power plant, an energy storage device and an electric boiler; utilizing Latin hypercube sampling to process wind and light output uncertainty; a double-layer optimization strategy and a hybrid dragonfly differential evolution algorithm (CT-DADE) based on a chaos theory are adopted to carry out optimization scheduling. An upper-layer optimization scheduling model takes wind and light output deviation minimization and economic benefit maximization as targets, and the output deviation is corrected through energy storage and an electric boiler; and the lower-layer optimization scheduling model optimizes the electric heating load curve by combining the time-of-use electricity price and the flexible load regulation characteristics. According to the method, a virtual power plant model including photo-thermal power generation, wind power generation, photovoltaic power generation, energy storage equipment, an electric boiler and a flexible load is constructed, uncertainty disturbance and supply-demand imbalance of new energy consumption are comprehensively considered, and stable supply and economical operation of an electric heating system are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of virtual power plant optimization scheduling, in particular to a kind of electric heat combined supply type virtual power plant optimization scheduling method based on photo-thermal power generation technology. BACKGROUND

[0002] With the increasing demand for clean energy and environmental pollution, renewable energy generation technologies such as wind power, photovoltaic power generation and photo-thermal power generation have developed rapidly. However, the intermittency and volatility of these renewable energy generation technologies have brought challenges to the stable operation and reliable power supply of the power system. Large-scale integration of renewable energy into the power grid will make it difficult to balance the supply and demand of the power system, and the phenomenon of wind and light abandonment is serious. Fluctuations may cause the frequency and voltage of the power grid to deviate, and cannot guarantee good power supply quality, which may seriously affect industrial production; intermittency will impact the power grid, and when the power grid does not have enough capacity, it will lead to a decrease in reactive power compensation capability and instability of the power system.

[0003] Virtual power plant (VPP) is a new type of power management mode that integrates distributed energy resources together, which can effectively improve the utilization rate of renewable energy and the stability of the power system. However, the existing virtual power plant still has deficiencies in dealing with wind and light output uncertainty, load fluctuation and achieving stable supply of electric heat system. Most existing researches focus on source and load uncertainty and user response behavior, and it is difficult to effectively represent the adjustment potential of VPP in different time periods under complex dynamic environment. On the other hand, most of the current real-time scheduling methods use fixed optimization window, which cannot flexibly adjust the size of the rolling window according to the fluctuation characteristics of frequency regulation demand, limiting the further optimization ability of VPP. SUMMARY

[0004] To solve the above technical problems, the present application relates to an electric heat combined supply type virtual power plant optimization scheduling method based on photo-thermal power generation technology, which aims to improve the consumption capacity of virtual power plant for renewable energy, reduce operation cost, improve economic benefit and reduce heat discharge. This method builds a virtual power plant model containing photo-thermal power generation, wind power generation, photovoltaic power generation, energy storage devices, electric boilers and flexible loads, and comprehensively considers the uncertainty disturbance of new energy consumption and supply-demand imbalance, to achieve stable supply and economic operation of electric heat system.

[0005] The technical scheme adopted by the present application is as follows: The electric heat combined supply type virtual power plant optimization scheduling method based on photo-thermal power generation technology comprises the following steps: Step 1: build a multi-energy virtual power plant model containing wind power generation, photovoltaic power generation, photo-thermal power station, energy storage device and electric boiler; Step 2: use Latin hypercube sampling to process wind and light output uncertainty; Step 3: Optimal scheduling is performed by using a double-layer optimization strategy and a hybrid dragonfly differential evolution algorithm (CT-DADE) based on chaos theory: The upper-layer optimal scheduling model aims to minimize the wind-solar output deviation and maximize the economic benefits, and corrects the output deviation through energy storage and electric boilers; The lower-layer optimal scheduling model optimizes the electric-heating load curve by combining time-of-use electricity prices and the regulation characteristics of flexible loads.

[0006] In step 1, the multi-energy virtual power plant model includes: (1) Wind power generation model: Let be the historical data of wind power output ultra-short-term prediction error, , ,…, be the prediction error historical data of the 1st to the th sample, which together constitute the sample space of wind power ultra-short-term prediction error data; is the sample size; let the probability density function of the random variable be , then the probability density function of the wind power output ultra-short-term prediction error is : (1); In formula (1), is the bandwidth; represents the Gaussian kernel function; denotes the i th observation value in the sample space; (2); In formula (2), represents the Gaussian kernel function; denotes the decay exponential term of the Gaussian kernel function.

[0007] (2) Photovoltaic power generation model: The normal distribution is used to represent the photovoltaic power generation output prediction error, and the specific probability density function is represented as: (3); In formula (3), denotes the probability density function; denotes the photovoltaic output prediction error; is the photovoltaic output prediction value at t time, is the error of the photovoltaic output prediction error at t time; is the photovoltaic output prediction error at tThe percentage of the predicted value relative to the predicted value of photovoltaic power output at any given time.

[0008] Actual photovoltaic output for: (4); (3) Energy storage device model: Thermal energy storage devices and electrical energy storage devices have similar overall operating characteristics and very similar constraints. X This refers to energy storage devices and thermal storage devices, whose state characteristics are characterized by State of Charge (SOC), and the mathematical expression is: (5); In equation (5), Indicates the energy storage device at time The state of charge; Indicates the energy storage device at time t The initial state of charge; The time indicated t The charging power; Indicates time t The discharge power; Indicates charging efficiency; Indicates discharge efficiency; Indicates the time interval for power conversion.

[0009] (4) Energy storage device model: 1) Flexible electrical load model: The following can reduce the cost of electricity load compensation: (6); In equation (6), This indicates that electricity load compensation costs can be reduced; Reduce fixed costs to reduce load; The unit capacity load reduction compensation coefficient; for t Time-of-use load reduction; This represents the total number of time periods included in the entire optimized scheduling cycle; It indicates a specific moment in time.

[0010] The output constraints and compensation costs of transferable electrical loads are as follows: (7); (8); In the formula: , They are respectively t Load power that can be transferred before and after time-based scheduling; The total cost of dispatching transferable loads is compensated. The unit transfer compensation coefficient; The t Period transfer load; And Respectively represent the charge-discharge time constant of the device.

[0011] 2) Thermal flexible load model based on comfort constraints: The mathematical expression of the flexible thermal load model is: (9); In formula (9), The t Heat supply load at time t; And Respectively represent t The indoor and outdoor temperature at time t; The heat capacity per unit heating area; Indicates the heating area; Indicates the time interval; Indicates t The indoor temperature at time t-1; Indicates the heat loss coefficient per unit area.

[0012] Indoor temperature is an important indicator affecting flexible thermal load. The range of consumer thermal comfort is measured by the change of indoor temperature, which ultimately limits the participation of flexible thermal load. The consumer thermal comfort model is as follows: (10); In formula (10), And The upper and lower adjustment range of indoor temperature, respectively; The middle value of the set temperature range; The change value of indoor temperature; The maximum set variable of indoor temperature; Indicates t The indoor temperature at time t; Indicates t The indoor temperature at time t-1.

[0013] Based on the Weber-Fechner law, a thermal load compensation price model is constructed, and based on this, a calculation method for thermal load adjustment cost is set: (11); (12); Where: Indicates the compensation price of unit power thermal load; The thermal load flexible demand response compensation coefficient; The thermal load sales price; This represents the constant offset of the compensation model; This represents the compensation cost for flexible demand response to heat load. For VPP in t The amount of heat load adjustment at any given time; Indicates the number of user categories; This indicates the number of individuals in each user category.

[0014] (5) Solar thermal power plant model: 1) Thermal power balance constraint: Ignoring heat loss, the mathematical expression for the heat power balance relationship inside a concentrated solar power (CSP) plant is: (13); in: for t The heat power that is constantly transferred from the heat collection system to the heat transfer fluid; for t The heat power that the thermal storage system releases to the heat transfer fluid at all times; for t The heat power absorbed by the heat storage system from the heat transfer fluid at any time; for t The heat power generated by the heat transfer fluid is constantly supplied to the power generation system.

[0015] 2) Constraints on Concentrated Solar Power Conversion: (14); in: The efficiency of converting light energy into heat energy; This represents the total area of ​​the solar mirror field; for t Predicted values ​​of solar radiation intensity at any given time; for t The heat that is not converted by the heat collection system at any given time.

[0016] 3) Constraints in the thermal storage process: The mathematical expression for thermal storage capacity constraints is: (15); in: for t The thermal energy stored in the thermal storage device at all times; , These are the minimum and maximum values ​​of thermal energy stored in the thermal storage device, respectively.

[0017] Constraints on heat storage devices and heat release power: (16); (17); (18); wherein: is t the thermal power supplied by the thermal storage system to the heat user at time t; , are t the charging and discharging power of the thermal storage device at time t, and the charging and discharging cannot be performed simultaneously; , represent the maximum charging and discharging power of the thermal storage device, respectively; , are the charging and discharging efficiencies of the thermal storage device, respectively.

[0018] 4) CSP power generation link constraints: The mathematical expression of the power generation of CSP is: (19); wherein: represents t the power generation of the CSP unit at time t; is the start-up thermal power of the thermal-electric conversion unit; is t the start-up state variable of the power generation unit at time t; is the thermal-electric conversion efficiency of the CSP power generation unit; represents t the thermal power transferred from the heat transfer fluid to the power generation system at time t.

[0019] The output constraint of the CSP power generation unit is: (20); wherein: , are the minimum and maximum values of the output of the CSP power generation unit, respectively; , are the lower and upper rotational reserves of the unit at time t. t

[0020] In step 2, the wind and solar power output scenarios are generated based on the Latin hypercube method. After modeling the virtual power plant dispatching resources in step 1, the Latin hypercube sampling method is used. First, the cumulative probability scale [0, 1] is divided into equal intervals, then random sampling is performed on each interval, and the sampling value is obtained through inverse transformation, which can ensure global coverage with fewer sampling times. The following steps are included: Step 2.1: Let x 1, x 2…, x T be T independent random variables, and the cumulative probability distribution function is: ​ (twenty one); In equation (21), Represents the cumulative probability distribution function; This represents the empirical cumulative distribution function obtained based on kernel density estimation; express t The error value of wind and solar power output prediction at any given time.

[0021] Step 2.2: Let M represent the sampling size, and plot the cumulative probability distribution function curve. The vertical axis is divided into M equal intervals, with a width of The interval, each interval width is , n =1,2,..., M , M Indicates the number of levels in the interval division.

[0022] Step 2.3: Select the midpoint of each interval as... The sampled values, relative to the cumulative distribution function Inverse function calculation x t The sampled value, i.e. x t The m Each sample value is: (twenty two); All sampled values Form a T x M initial sampling matrix X , Represents the cumulative distribution function The inverse function maps uniformly distributed probability values ​​back to the value space of the original variable.

[0023] In step 3, a two-layer optimization scheduling objective function and constraints are set based on flexible load and time-of-use pricing, and a two-layer optimization scheduling model is adopted. In the upper-layer optimization scheduling model, based on the deviation between the VPP's day-ahead wind and solar power output forecast and the next day's actual power output plan, energy storage batteries and electric boilers are coordinated to reduce penalty costs. In the lower-layer optimization scheduling model, taking into account the flexible load regulation characteristics, load demand is adjusted through time-of-use pricing, increasing the VPP's economic benefits while potentially achieving peak shaving and valley filling benefits, and increasing the absorption of new energy. Simultaneously, based on the upper-layer optimization results, with the goal of maximizing the overall operating benefit of the VPP, the CT-DADE algorithm is used to optimize and solve the model, obtaining the operating output and maximum net benefit results of each distributed unit. Specifically: 3.1 Upper-level Optimization Scheduling Model: To mitigate the impact of wind and solar power fluctuations on the optimization scheduling strategy, the VPP (Virtual Power Utility) corrects for output deviations using energy storage batteries and electric boilers based on the actual wind and solar power outputs of the following day. Taking into account the operation and maintenance costs of distributed units and the penalty costs incurred due to wind and solar power output deviations, an upper-level optimization scheduling model is established with the objectives of maximizing VPP economic benefits and minimizing wind and solar power output deviations. The specific objective function is as follows: (twenty three); In equation (23), Indicates the first k The optimal fitness value at +1 iteration; Indicates the first k The optimal fitness value at the next iteration; This represents the relative critical value indicating the magnitude of improvement.

[0024] in: (twenty four); (25); (26); in: , respectively under time-of-use electricity pricing t The electricity purchase and sale price from the grid by the VPP at any given time; for t The penalty costs arising from deviations in output at all times; for t Electricity sales revenue in the VPP upper-layer model at any given time; for t The comprehensive operation and maintenance cost of wind and solar power generation at any given time; Indicates the virtual power plant during the time period t The total output plan for the application; Indicates time period t The actual output of photovoltaic power; Indicates time period t The actual output of wind power; Indicates the discharge power of the energy storage device; Indicates the charging power of the energy storage device; This represents the electrical energy consumed in the conversion from electricity to heat. This represents the unit operation and maintenance cost coefficient for photovoltaic systems. This represents the unit operation and maintenance cost coefficient for wind power.

[0025] 3.2: Lower-level optimized scheduling model: To reduce the fluctuation of thermal load, the lower layer model adjusts the electrical load according to the time-of-use price, and adjusts the thermal load according to the indoor temperature and user comfort index, which plays a potential role in peak shaving and valley filling. Secondly, according to the adjusted net load curve and the output of the upper layer distributed unit, the overall economic benefit of the lower layer VPP is maximized. The specific objective function is as follows: (27); In formula (27), represents the overall economic benefit; , are the VPP heat and electricity sales revenue at time t; t is the total cost of flexible electrical load and thermal load dispatching compensation; wherein: (28); (29); (30); In the above formula, represents the thermal power directly supplied to the thermal load by the thermal storage system during the CSP period of the photo-thermal power plant; t represents the heat supply power of the electric boiler during the period ; t and and represent the heat release power and heat charging power of the thermal storage system during the period t ; represents the power generation of the photo-thermal power plant during the period t ; represents the compensation price per unit of thermal load adjustment; represents the thermal load adjustment amount of user i during the period t ; represents the compensation cost of the transferable electrical load during the period t ; represents the compensation cost of the reducible electrical load during the period t ; represents the total number of users participating in flexible thermal load adjustment; represents the total number of time periods divided in the dispatching period.

[0026] 3.3: Set the VPP electrical power balance constraint: (31); In formula (31), represents the total output plan of the virtual power plant declared to the power grid during the period t ; represents the power generation of the photo-thermal power plant CSP during the period t ;the generated power of the energy storage device; denotes the time interval length t denotes the adjustment amount of the flexible electrical load

[0027] 3.4: Energy storage device constraints: (32); (33); (34); (35); (36); wherein: , denote the upper and lower limits of the energy storage device capacity, respectively; denotes the energy storage device's state of charge t at the beginning of the time interval denotes the energy storage device's remaining energy at the end of the time interval t -1; denotes the time interval length denotes the energy storage device's charging power at the beginning of the time interval t denotes the energy storage device's discharging power at the beginning of the time interval denotes the energy storage device's state of charge t denotes the energy storage device's state of charge , denote the charging and discharging efficiencies, respectively; , denote the charging and discharging state variables, respectively; , denote the minimum charging and discharging power limits of the energy storage device, respectively; , denote the maximum charging and discharging power limits, respectively.

[0028] 3.5: Photovoltaic and wind farm power output constraints: (37); (38); wherein: denotes the time interval length t denotes the actual output of the wind farm denotes the maximum output of the wind turbine denotes the actual output of the photovoltaic power plant t denotes the maximum output of the photovoltaic array denotes the maximum output of the photovoltaic array

[0029] The hybrid dragonfly differential evolution algorithm based on chaos theory (CT-DADE) will be used to solve the model, including the following steps: To address the shortcomings of the traditional Dragonfly Algorithm (DA) algorithm, such as slow convergence speed and susceptibility to local optima, this paper proposes to enhance the algorithm's early-stage global optimization capability by adding an individual particle memory and introducing chaos theory (CT). Then, based on the cost-based convergence hybrid algorithm switching criterion, the current optimal solution of the DA algorithm is assigned to the Differential Evolutionary Algorithm (DE) to enhance the local optimization capability. Step 1: Iteratively generate chaotic sequences using the Tent mapping formula. , Let represent a set of pseudo-random number sequences with specific mathematical properties generated through chaotic mapping, and let represent an initial population with a uniform distribution. The expression is as follows: (39); (40); in: Represents the random sequence after iterative updates; This represents the random sequence before the iterative update; Indicates the first i The initial position of each individual dragonfly; and These represent the lower and upper limits of the search space, respectively; The first in the Tent chaotic sequence i Values.

[0030] Step 2: Optimize the dragonfly update step size using chaotic sequences: (41); In equation (41), This indicates the number of iterations for each dragonfly. t Position at +1 This indicates the number of iterations for each dragonfly. t The position at that time; Represents the first generation generated by the Tent chaotic map. i A chaotic number; express Levy The random step size generated by the flight function follows Levy distributed.

[0031] Step 3: Improve the mutation operation of the DE algorithm using chaotic sequences: The improved exploration capabilities of the DA algorithm are combined with the development capabilities of the DE algorithm to achieve a proper balance between diversification and intensification, thus reaching global optimum. The particle optimum of the DA algorithm (…) pBest ) and global optimal ( gBest The solution was used in the mutation process of the DE algorithm, and the Tent chaotic mapping was introduced to optimize the mutation parameters and enhance the diversity of the population. (42); (43); (44); (45); in: This represents the generated candidate solutions, which are used for subsequent crossover and selection operations; Indicates the current individual i In the k The position of the dimension; Represents a Tent chaotic sequence of a The power of the power is used as an adaptive weight; Represents a Tent chaotic sequence of b Power; Represents an individual i The historical best position; , These are the particle values ​​at the highest and lowest fitness levels, respectively. , These are the variation parameters; This represents the number of iterations for the current algorithm. This represents the mean of the variation factor; This indicates the number of iterations that have been completed so far. Indicates the maximum number of iterations; Represents a random number between 0 and 1.

[0032] This invention discloses an optimized scheduling method for a combined heat and power (CHP) virtual power plant based on concentrated solar power (CSP) technology, with the following technical advantages: 1) Step 1 of the present invention integrates multiple distributed energy resources by constructing a multi-energy virtual power plant model that includes wind power generation, photovoltaic power generation, solar thermal power plant, energy storage device and electric boiler, and makes full use of the spatiotemporal complementarity of each energy source to realize the joint production of multiple energy sources.

[0033] 2) Step 2 of the present invention uses Latin hypercube sampling to process the uncertainty of wind and solar power output, which can efficiently generate a globally covered wind and solar power output scene, effectively reduce the number of samplings, improve computational efficiency, and accurately reflect the randomness and fluctuation of wind and solar power output, providing reliable input for optimized scheduling and enhancing the stability and reliability of virtual power plant operation.

[0034] 3) The step 3 of the application adopts a double-layer optimization scheduling model and a hybrid dragonfly differential evolution algorithm (CT-DADE) based on chaos theory, the upper layer optimization smoothes the wind-solar output deviation, and the lower layer optimization adjusts the electric heating load curve, and gives consideration to economy and supply-demand balance. The CT-DADE algorithm combines chaos theory and differential evolution, improves the global and local optimization ability, effectively reduces the operation cost, and improves the economic benefit and power supply reliability of the virtual power plant.

[0035] 4) The application aggregates the photothermal power generation and the wind-solar distributed power generation unit into a virtual power plant with electric heating combined supply characteristics based on the photothermal power generation technology; meanwhile, the flexible electric load and the thermal load are modeled based on the load response type and the user comfort degree respectively; the model is solved by using the hybrid dragonfly differential evolution algorithm based on chaos theory to realize day-ahead scheduling, smooth the load curve, and improve the power supply reliability of the virtual power plant.

[0036] 5) The application considers the optimization scheduling of the electric heating combined supply type virtual power plant, can realize the functions of optimizing the load curve and relieving the high peak power supply pressure, and the comprehensive cost of the primary energy of the system can also be effectively reduced. BRIEF DESCRIPTION OF DRAWINGS

[0037] The application will be further described below in combination with the drawings and embodiments: Figure 1 It is a structure diagram of the electric heating combined supply type virtual power plant.

[0038] Figure 2 It is a flowchart of the DADE algorithm based on chaos search.

[0039] Figure 3 It is a curve diagram of the wind-solar processing predicted value and actual value.

[0040] Figure 4 It is a diagram of the time-of-use electricity price data of the power grid.

[0041] Figure 5 It is a comparison diagram of the electric load before and after optimization.

[0042] Figure 6 It is a comparison diagram of the thermal load before and after optimization. DETAILED DESCRIPTION

[0043] The electric heating combined supply type virtual power plant optimization scheduling method based on the photothermal power generation technology comprises the following steps: Step 1: constructing an electric heating combined supply type virtual power plant system framework; Step 2: modeling the scheduling resources based on the photothermal power generation adjustment characteristics; Step 3: generating a wind-solar output scene based on the Latin hypercube method; Step 4: setting a double-layer optimization scheduling objective function and constraint conditions based on the flexible load and the time-of-use electricity price; Step five: the hybrid dragonfly differential evolution algorithm based on chaos search will be used to solve the model.

[0044] In step one, the electric-thermal combined virtual power plant system framework is constructed, including the following: The electric-thermal combined virtual power plant system framework is constructed. The virtual power plant integrates various distributed units to form a multi-dimensional heterogeneous electric energy centralized management entity. In the entity, the virtual power plant controls and regulates the units efficiently according to system operation conditions and economic indicators. Wind power generation (WPP), photovoltaic power generation (PV), concentrated solar power (CSP), energy storage devices, and electric boilers are integrated into a virtual power plant (VPP). The temporal and spatial complementarity of different distributed energy sources is utilized to realize multi-energy joint production. According to the characteristics of electric and thermal loads, the controllable load is used as a demand-side resource to participate in market dispatching, improve system peak shaving and energy utilization efficiency, and the system framework is as shown in Figure 1 .

[0045] In step two, the scheduling resources are modeled based on the regulation characteristics of photothermal power generation, including the following specific contents: The kernel density estimation method is used to model the probability density function of wind power output prediction error. The kernel density estimation method in non-parametric estimation is used to model the photovoltaic output prediction error. The SOC represents the state characteristics of the electricity and heat storage devices. The flexible electric-thermal load model includes a compensating cost model of reducible / transferable electric load and a thermal load model based on comfort level. The CSP model includes thermal power balance constraints, concentrated heat collection and conversion constraints, heat storage link constraints, and power generation link constraints.

[0046] 2.1: Wind power generation model modeling: The kernel density estimation method in non-parametric estimation is used for estimation, The wind power output ultra-short-term prediction error historical data, , ,…, is the sample space of wind power ultra-short-term prediction error data, is the sample size, and let the random variable The probability density function of the random variable is .

[0047] (1); In the formula, is the bandwidth; represents the Gaussian kernel function.

[0048] (2); 2.2: Photovoltaic power generation model modeling. The normal distribution represents the prediction error of photovoltaic power generation output, and the specific probability density function can be represented as: (3); wherein, is the photovoltaic output prediction value at time t, t is the standard deviation of the photovoltaic output prediction error at time t; t is the percentage of the photovoltaic output prediction error to the prediction value at time t. t

[0049] Actual photovoltaic output is: (4); 2.3: Energy storage device model modeling. The overall operating characteristics of the thermal storage device and the electric storage device are similar, and the related constraints differ slightly, and the X is used to represent the electric storage device and the thermal storage device, and the state characteristics are represented by SOC, and the mathematical expression is: (5); 2.4: Flexible electric heating load model modeling: 1) Flexible electric load model: The compensation cost of the reducible electric load is as follows: (6); wherein represents the compensation cost of the reducible electric load; is the fixed cost of load reduction; is the compensation coefficient of unit capacity load reduction; is the electric load reduction amount in the time period t. t The transferable electric load output constraint and compensation cost are as follows:

[0050] (7); (8); In the formula, , respectively, is the transferable load power before and after scheduling in the time period t; t is the total compensation cost of the transferable load scheduling; is the unit transfer compensation coefficient; is the transferable load amount in the time period t. t 2) Thermal flexible load model based on comfort constraint:

[0051] ​​​​​The mathematical expression of the flexible heat load model is as follows: (9); In the formula, is t the heat load at the moment; and are t the indoor and outdoor temperatures at the moment; is the heat capacity per unit heating area.

[0052] The indoor temperature is an important indicator affecting the flexible heat load. The consumer's heat comfort range is measured by the change of the indoor temperature, which ultimately limits the participation amount of the flexible heat load. The consumer's heat comfort model is as follows: (10); In the formula, and are the upper and lower adjustment ranges of the indoor temperature; is the middle value of the set temperature range; is the change value of the indoor temperature; is the maximum set variable of the indoor temperature.

[0053] Based on the Weber-Fechner law, a heat load compensation price model is constructed, and based on this, a calculation method of heat load adjustment cost is set: (11); (12); In the formula, is the heat load flexible demand response compensation coefficient; is the heat load sales price; represents the compensation cost of heat load flexible demand response; represents the compensation price of unit power heat load; is the heat load adjustment amount of the VPP at the moment t .

[0054] 2.5: Perform modeling of the photo-thermal power station model: 1) Thermal power balance constraint: Neglecting heat loss, the mathematical expression of the thermal power balance relationship inside the CSP power station is: (13); In the formula, is the heat power delivered by the heat collection system to the heat transfer fluid at the moment t ; is the heat power released by the heat storage system to the heat transfer fluid at the moment t ; is the heat power of the heat transfer fluid at the moment tthe heat power absorbed by the thermal storage system from the heat transfer fluid at time t; for t the heat power supplied by the heat transfer fluid to the power generation system at time t.

[0055] 2) Concentrating solar power conversion constraints: (14); where is the optical-to-thermal efficiency; is the total area of the solar mirror field; is t the predicted value of the solar radiation intensity at time t; is t the heat not converted by the concentrating solar power system at time t.

[0056] 3) Thermal storage constraints: The mathematical expression of the thermal storage capacity constraint is: (15); is t the thermal energy stored in the thermal storage device at time t; , and

[0057] Thermal storage device and heat release power constraints: (16); (17); (18); where: is t the heat power supplied by the thermal storage system to the heat users at time t; , are t the charging and discharging power of the thermal storage device at time t, and charging and discharging cannot be performed simultaneously; , represent the maximum charging and discharging power of the thermal storage device; , are the charging and discharging efficiencies of the thermal storage device.

[0058] 4) Concentrating solar power generation constraints: The mathematical expression of the power generation of the concentrating solar power is: (19); where represents t the power generation of the solar thermal power generation unit at time t; is the start-up heat power of the thermal-to-electric conversion unit; For t the starting state variable of the generator set at time t; For the thermal-electric conversion efficiency of the CSP power generation unit.

[0059] The output constraint of the photothermal generator set is: (20); Wherein: , are the minimum and maximum values of the output of the photothermal generator set, respectively; , are the lower and upper rotational reserves of the unit at time t. t

[0060] In the third step, the wind and light output scene is generated based on the Latin hypercube method. First, the cumulative probability scale [0, 1] is divided into equal intervals, then random sampling is performed on each interval, and the sampling value is obtained through inverse transformation, including the following steps: Step S3.1: Let x 1, x 2…, x T be T independent random variables, and the cumulative probability distribution function is: (21); Step S3.2: Let M represent the sampling size, and divide the cumulative probability distribution function curve into M equal intervals with a width of , and each interval has a width of , n =1,2,…, M .

[0061] Step S3.3: Select the midpoint of each interval as the sampling value of , and calculate the inverse function of the cumulative distribution function to obtain the sampling value of x t , that is x The m th sampling value of t is: (22); All sampling values T x M form an initial sampling matrix X .

[0062] In the fourth step, the double-layer optimization scheduling objective function and constraint conditions are set based on the flexible load and time-of-use electricity price, including the following steps: ​4.1: Upper-level optimized scheduling model: To mitigate the impact of wind and solar power volatility on the optimal scheduling strategy, the Virtual Power Plant (VPP) corrects for output deviations using energy storage batteries and electric boilers based on the actual wind and solar power output the following day. Taking into account the operation and maintenance costs of distributed generation units and the penalty costs incurred due to wind and solar power output deviations, a higher-level optimal scheduling model is established with the objectives of maximizing VPP economic benefits and minimizing wind and solar power output deviations. The specific objective function is as follows: (twenty three); in (twenty four); (25); (26); in , respectively under time-of-use electricity pricing t The electricity purchase and sale price from the grid by the VPP at any given time; for t The penalty costs arising from deviations in output at all times; for t Electricity sales revenue in the VPP upper-layer model at any given time; for t The comprehensive operation and maintenance cost of wind and solar power generation at all times.

[0063] 4.2: Lower-level optimized scheduling model: To mitigate fluctuations in thermal and power loads, the lower-level optimization scheduling model adjusts the electrical load based on time-of-use pricing and the thermal load based on indoor temperature and user comfort indicators, thus potentially achieving peak shaving and valley filling. Furthermore, based on the adjusted net load curve and the output of the upper-level distributed generation units, the objective is to maximize the overall economic benefit of the lower-level VPP. The specific objective function is as follows: (27); in (28); (29); (30); in , They are respectively t Revenue from heat and electricity sales by VPP at all times; The total cost of flexible electrical load and thermal load dispatch compensation.

[0064] 4.3: Setting VPP power balance constraints: (31); 4.4: Set the energy storage device constraints: (32); (33); (34); (35); (36); where: , represent the upper and lower limits of the energy storage device capacity, respectively; represents t the energy storage device's energy at the time period; , are the charging and discharging efficiencies, respectively; , represent the state variables of charging and discharging, respectively; , represent the minimum charging and discharging power limits of the energy storage device, respectively; , represent the maximum charging and discharging power limits, respectively.

[0065] 4.5: Set the photovoltaic and wind power plant output constraints.

[0066] (37); (38); In step five, the hybrid mayfly differential evolution algorithm based on chaos theory (CT-DADE) will be used to solve the model, including the following steps: To solve the problems of slow convergence speed and easy to fall into local optimum of traditional mayfly algorithm (DA), by increasing the individual particle memory, and introducing chaos theory (CT) to improve the global optimization ability of the algorithm in the early stage, then according to the hybrid algorithm switching criterion based on cost convergence, the current optimal solution of DA algorithm is assigned to differential evolution algorithm (DE), to enhance the local optimization ability.

[0067] S5.1: Use Tent mapping formula to generate chaotic sequence and generate an initial population with uniform distribution, the expression is: (39); (40); where and represent the lower and upper limits of the search space, respectively; is the i th value in the Tent chaotic sequence.

[0068] S5.2: Dragonfly update step size optimization using chaotic sequence: (41); S5.3: DE algorithm mutation operation improved by chaotic sequence: The improved DA exploration ability combined with the development ability of DE achieves a proper balance between diversification and intensification, reaching the global optimum. The DA particle optimal ( pBest ) and global optimal ( gBest ) solutions are used in the mutation process of DE, and the Tent chaotic mapping is introduced to optimize the mutation parameters, enhancing the diversity of the population: (42); (43); (44); (45); where , are the particle values at the highest and lowest fitness, respectively; , are the mutation parameters; is the current iteration number of the algorithm.

[0069] S5.4: Solving the double-layer optimization scheduling model constructed on the MATLAB platform using CPLEX software.

[0070] To verify the correctness of the proposed method, the following case analysis is performed: The example model is constructed in MATLAB, and the upper model is composed of 1 500 kW wind farm, 1 200 kW photovoltaic power station, 1 100 kW electricity storage device, and 1 150 kW electric boiler; the lower model is composed of 1 photoelectric heat station (both power generation and heating power are 200 kW), 1 150 kW electric boiler, and 1 100 kW heat storage device.

[0071] The wind and light predicted output and actual output curves are shown in Figure 3 The differences between the wind and light predicted output and actual output can be seen from Figure 3 , which intuitively reflects the intermittency and volatility of wind and light output. By comparing the curves, the accuracy of the prediction model can be evaluated, providing a basis for optimizing the scheduling strategy and helping the virtual power plant better arrange the charging and discharging strategy of the energy storage device and the adjustment of the flexible load, thereby improving the utilization rate of renewable energy and the stability of the system.

[0072] The time-of-use electricity price data are shown in Figure 4 The differences between the wind and light predicted output and actual output can be seen from Figure 4It can be seen that the change rule of time-of-use electricity price, and the price difference is obvious in different time periods. Time-of-use electricity price provides economic incentives for VPP optimal dispatch, guiding it to store energy or reduce load during low price period, and discharge or increase power generation during peak price period, thereby reducing operating costs and improving economic benefits.

[0073] Table 1 Relationship between CSP heat storage capacity, VPP heat rejection rate and operating income

[0074] The heat storage duration of the photothermal power station represents the number of hours that the photothermal power station runs at rated power under lightless conditions, and can indirectly represent the capacity of the heat storage system. As can be seen from Table 1, as the heat storage capacity of the photothermal power station increases, the heat rejection rate of the thermal system decreases from 23.31% to 3.01%, the heat rejection penalty cost gradually decreases, and the operating income of the VPP continuously increases. Therefore, the selection of a photothermal power station with appropriate heat storage capacity is of great significance to improving the overall operating reliability and economic benefits of the VPP.

[0075] As shown in the optimization results shown in Figure 5 , Figure 6 , after considering the flexible electric and thermal loads, the standard deviation of the electric load curve decreases from 69.49 kW to 48.9 kW, and the overall smoothness increases by 29.6%; the standard deviation of the thermal load curve decreases from 26.81 kW to 22.58 kW, and the overall smoothness increases by 18.5%. The optimization results verify the feasibility of changing user energy demand based on time-of-use electricity price strategy, and the user comfort value index also provides a good reference for the transfer of thermal load, which is conducive to the realization of unit heat and power decoupling, and lays a foundation for improving the operating stability and operating income of the VPP system.

Claims

1. A method for optimal dispatch of a combined heat and power virtual power plant based on photothermal power generation technology, characterized in that The method comprises the following steps: Step 1: constructing a multi-energy virtual power plant model comprising wind power generation, photovoltaic power generation, a photo-thermal power station, energy storage devices and an electric boiler; Step 2: processing wind and light output uncertainty by using Latin hypercube sampling; Step 3: adopting a double-layer optimization strategy and a hybrid dragonfly differential evolution algorithm based on chaos theory for optimization scheduling: The upper-layer optimization scheduling model aims to minimize wind and light output deviation and maximize economic benefits, and corrects the output deviation by using energy storage and an electric boiler; The lower-layer optimization scheduling model optimizes the electric-thermal load curve in combination with time-of-use electricity prices and the adjustment characteristics of flexible loads.

2. The method of claim 1, wherein the method is characterized in that: In the step 1, the multi-energy virtual power plant model comprises a wind power generation model, and the wind power generation model comprises: Let be the historical data of wind power output ultra-short-term prediction error, , ,…, be the historical data of prediction error of the 1st to the th sample, respectively, which together constitute the sample space of wind power output ultra-short-term prediction error; be the sample size; let the probability density function of the random variable be , then the probability density function of the wind power output ultra-short-term prediction error is : （1）； In formula (1), is the bandwidth; represents a Gaussian kernel function; denotes the i-th observation in the sample space; i denotes the i-th observation in the sample space; （2）； In formula (2), a Gaussian kernel function represented by an attenuation exponential term of the Gaussian kernel function.

3. The method of claim 2, wherein the method further comprises: The multi-energy virtual power plant model comprises a photovoltaic power generation model, and the photovoltaic power generation model comprises: Adopting normal distribution The error of photovoltaic power generation output prediction is represented, and the specific probability density function is represented as: （3）； In formula (3), denotes a probability density function; denotes a photovoltaic power output prediction error; is t a photovoltaic power output prediction value at a time instant, is t an error of a photovoltaic power output prediction error at a time instant; is t a percentage of a photovoltaic power output prediction error to a prediction value at a time instant; Actual photovoltaic output is: （4）。 4. The method of claim 3, wherein the method further comprises: Adopting X The SOC represents the state characteristics of the electric storage device and the heat storage device, and the mathematical expression is: （5）； In formula (5), denotes a state of charge of the energy storage device at a time ; denotes an initial state of charge of the energy storage device at a time t ; denotes a charging power at a time t ; denotes a discharging power at a time t ; denotes a charging efficiency; denotes a discharging efficiency; denotes a time interval of power conversion.

5. The method of claim 4, wherein the method further comprises: The multi-energy virtual power plant model comprises an energy storage device model, and the energy storage device model comprises: 1) a flexible electric load model: The cuttable electric load compensation cost is as follows: （6）； In formula (6), represents the cuttable electric load compensation cost; is the load cut fixed cost; is the unit capacity load cut compensation coefficient; is t is the period electric load cut amount; represents the total number of time periods contained in the entire optimization scheduling period; represents a certain time; The transferable electric load output constraint and compensation cost are as follows: （7）； （8）； In the formula: , are respectively t transferable load power before and after period scheduling; is total compensation cost of transferable load scheduling; is unit transfer compensation coefficient; is t period transferable load amount; and respectively represent charge and discharge time constants of the energy device; 2) a thermal flexible load model based on comfort constraint: The flexible thermal load model is mathematically expressed as: （9）； In formula (9), is t the heating load at the time; and respectively represent t the indoor and outdoor temperatures at the time; is the heat capacity per unit heating area; represents the heating area; represents the time interval; represents t the indoor temperature at the time -1; represents the heat loss coefficient per unit area; The consumer thermal comfort model is as follows: （10）； In formula (10), and are an upper and lower adjustment range of the indoor temperature, respectively; is a middle value of the set temperature range; is a change value of the indoor temperature; is a maximum set variable of the indoor temperature; represents t the indoor temperature at the time t; represents t the indoor temperature at the time t-1; A thermal load compensation price model is constructed based on the Weber-Fechner law, and based on this, a calculation method of the thermal load adjustment cost is set: （11）； （12）； wherein: represents the unit power heat load compensation price; is the heat load flexible demand response compensation coefficient; is the heat load sales price; represents the constant offset of the compensation model; represents the compensation cost of heat load flexible demand response; is the heat load adjustment amount of the VPP at t is the time; represents the number of user categories; represents the number of individuals in each category of users.

6. The method of claim 5, wherein the method further comprises: The multi-energy virtual power plant model comprises a photo-thermal power station model, and the photo-thermal power station model comprises: 1) a thermal power balance constraint: Neglecting thermal energy loss, the mathematical expression of the thermal power balance relationship inside the photo-thermal power station CSP is as follows: （13）； wherein: is the thermal power delivered by the heat collection system to the heat transfer fluid at time t; t is the thermal power released by the heat storage system to the heat transfer fluid at time t; is the thermal power absorbed by the heat storage system from the heat transfer fluid at time t; t is the thermal power released by the heat storage system to the heat transfer fluid at time t; is the thermal power absorbed by the heat storage system from the heat transfer fluid at time t; t is the thermal power released by the heat storage system to the heat transfer fluid at time t; is the thermal power absorbed by the heat storage system from the heat transfer fluid at time t; t is the thermal power delivered by the heat transfer fluid to the power generation system for power generation at time t; 2) a concentrated heat collection and conversion constraint: （14）； wherein: is the light-to-heat efficiency; is the total area of the solar mirror field; is t is the predicted value of the instantaneous solar radiation intensity; is t is the heat not converted by the thermal collection system at the instant. 3) a heat storage link constraint: The mathematical expression of the heat storage capacity constraint is as follows: （15）； wherein: is t the thermal energy stored in the thermal storage device at the time instant; , are the minimum and maximum values of the thermal energy stored in the thermal storage device, respectively. The storage heat device and heat release power constraint: （16）； （17）； （18）； wherein: is t the thermal power supplied by the thermal storage system to the thermal user at time t; , are t the charging and discharging power of the thermal storage device at time t, and the charging and discharging cannot be performed simultaneously; , represent the maximum charging and discharging power of the thermal storage device, respectively; , are the charging and discharging efficiencies of the thermal storage device, respectively; 4) a photo-thermal power station CSP power generation link constraint: The mathematical expression of the CSP power generation power is as follows: （19）； wherein: represents t the power generated by the power generator at the time instant t; is the start-up thermal power of the thermoelectric conversion unit; is t is the start-up state variable of the power generator at the time instant t; is the thermoelectric conversion efficiency of the CSP power generation unit; represents t is the thermal power transferred from the heat transfer fluid to the power generation system at the time instant t; The output constraint of the photo-thermal generator set is as follows: （20）； in: , These are the minimum and maximum output values ​​of the solar thermal power generator set, respectively. , The units are respectively in t The time for the down rotation and the up rotation is reserved.

7. The method of claim 1, wherein the method further comprises: The step 2 comprises the following steps: Step 2.1: Set x 1, x 2…, x T is T an independent random variable with cumulative probability distribution function: （21）； In formula (21), denotes a cumulative probability distribution function; denotes an empirical cumulative distribution function based on kernel density estimation; denotes t a wind and light output prediction error value at the time point Step 2.2: Let M represent the sampling scale, divide the ordinate of the cumulative probability distribution function curve into M equal intervals, each interval having a width of n = 1, 2, …, M M represents the number of interval divisions.​​​ Step 2.3: Choose the midpoint of each interval as the sample value, F the cumulative distribution function the inverse function to calculate x t the sample value, i.e. x t the first m sample value of the interval （22）； all the sampled values forming one T x M initial sampled matrix X , denotes the inverse function of the cumulative distribution function mapping uniform distributed probability values back to the value space of the original variable.

8. The method of claim 1, wherein the method further comprises: In the step 3, a double-layer optimization scheduling objective function and constraint condition are set based on the flexible load and time-of-use electricity price: 3.1: an upper-layer optimization scheduling model is established, and the specific objective function is as follows: （23）； In formula (23), represents the first k optimal fitness value at the first iteration; represents the first k optimal fitness value at the first iteration; represents a relative threshold value representing an improvement range. Wherein: （24）； （25）； （26）； wherein: , are the purchase and sale electricity price of the grid at time VPP under time-of-use price respectively; t is the penalty cost of the wind and solar power output deviation at time VPP; t is the sale electricity revenue of the VPP upper model at time VPP; t is the comprehensive operation and maintenance cost of wind and solar power generation at time VPP; t represents the declared total output plan of the virtual power plant in period t represents the actual output of the photovoltaic in period t represents the actual output of the wind power in period t represents the discharging power of the energy storage device; represents the charging power of the energy storage device; represents the electric energy consumed by the electric heat conversion; represents the unit operation and maintenance cost coefficient of the photovoltaic; represents the unit operation and maintenance cost coefficient of the wind power;​​​​​​​ 3.2: the specific objective function of the lower-layer optimization scheduling model is as follows: （27）； In formula (27), represents the overall economic benefits; , respectively t VPP heat and electricity sales revenue at time; is the total compensation cost of flexible electric and thermal load dispatching; Wherein: （28）； （29）； （30）； In the above formula, Indicates the CSP period of a solar thermal power plant t The heat power is directly supplied to the heat load through the heat storage system; Indicates the time period of the electric boiler t The heating capacity; and These represent the time periods of the thermal storage system. t The heat release power and the heat charge power; Indicates the solar thermal power plant during the time period t The power generation capacity; This indicates the compensation price per unit of heat load adjustment; Indicates user i During the period t Heat load adjustment amount; Indicates time period t Compensation costs for transferable electrical loads; Indicates time period t It can reduce the cost of compensating for electrical load; This indicates the total number of users participating in flexible heat load regulation; This indicates the total number of time periods divided within the scheduling cycle.

9. The method according to claim 8, characterized in that: 1) a VPP electric power balance constraint is set: （31）； In formula (31), representing the time period t total power output plan declared by the virtual power plant to the grid; representing the power generation of the CSP plant in the time period t ; representing the time period t regulation amount of the flexible electrical load; 2) an energy storage device constraint is set: （32）； （33）； （34）； （35）； （36）； wherein: , respectively represent the upper and lower limits of the energy storage device capacity; represents t the electric quantity of the energy storage device at the time interval; represents the time interval length; t represents the residual electric quantity of the energy storage device at the time interval represents the time interval length; represents the charging power of the energy storage device at the time interval t represents the time interval length; represents the discharging power of the energy storage device at the time interval t represents the time interval length; , respectively represent the charging and discharging efficiencies; , respectively represent the state variables of charging and discharging; , respectively represent the minimum charging and discharging power limits of the energy storage device; , respectively represent the maximum charging and discharging power limits. 3) a photovoltaic and wind power plant output constraint is set: （37）； （38）； wherein: denotes a time period t actual power of a wind farm; denotes the maximum power of a wind turbine generator denotes a time period t actual power of a photovoltaic plant; denotes the maximum power of a photovoltaic array.

10. The method of claim 9, wherein the method further comprises: The hybrid dragonfly differential evolution algorithm based on chaos theory is used to solve the model, comprising the following steps: Step 1: Iterative generation of chaotic sequence using Tent mapping formula , respectively represent a set of pseudo-random number sequences with specific mathematical properties generated by chaotic mapping, and generate an initial population with uniform distribution, expressed as: （39）； （40）； wherein: denotes the iteratively updated random sequence; denotes the iteratively updated random sequence; denotes the initial position of the i dragonfly individual; and denote the lower and upper bounds of the search space, respectively; is the i value in the Tent chaotic sequence; Step 2: a dragonfly update step is optimized by using a chaos sequence: （41）； In formula (41), denotes the position of the dragonfly individual at iteration number t +1 denotes the position of the dragonfly individual at iteration number t +1 denotes the jth chaotic number generated by the Tent chaotic map; i denotes the jth chaotic number generated by the Tent chaotic map; denotes Levy the random step generated by the flight function, which is subject to a Levy uniform distribution; Step 3: the mutation operation of the DE algorithm is improved by using a chaos sequence: The particle optimal and global optimal solutions of the DA algorithm are used in the mutation process of the DE algorithm, and a Tent chaos mapping is introduced to optimize the mutation parameter, so that the diversity of the population is enhanced: （42）； （43）； （44）； （45）； wherein: represents the generated candidate solution for subsequent crossover and selection operations; represents the current individual i in the first k dimension; represents the Tent chaotic sequence raised to the power of a as an adaptive weight; represents the Tent chaotic sequence raised to the power of b ; represents the historical best position of the individual i ; , are the particle values at the highest and lowest fitness, respectively; , are the mutation parameters, respectively; is the number of iterations of the current algorithm; represents the mean of the mutation factors; represents the current number of completed iterations; represents the maximum number of iterations; represents a random number between 0 and 1.

Images