Low-carbon park capacity configuration and optimization method of photovoltaic and electric heavy truck

Abstract

The invention discloses a low-carbon park capacity configuration and optimization method for photovoltaic and electric heavy trucks. A Monte Carlo sampling method is adopted to simulate the daily driving mileage and the charging starting time of the pure electric heavy truck, and a pure electric heavy truck charging quantity model is established; the method comprises the following steps of: carrying out scientific division on peak, flat and valley periods by adopting a self-organizing map (SOM) neural network clustering method, and optimizing time-of-use electricity price and the charging amount of the pure electric heavy truck by taking a fuzzy demand response mechanism based on a Logistic function as a basis and adopting a non-dominated sorting genetic algorithm (non-dominated sorting genetic algorithm-II); an upper layer system capacity configuration planning model with the minimum park equipment investment cost as the target and a lower layer low-carbon optimization scheduling model with the minimum system operation cost as the target are established, and a green power supply mode of photovoltaic power station-charging pile-pure electric heavy truck is established. According to the strategy provided by the invention, carbon emission and economic cost are reduced, and collaborative development of green low-carbon power supply and economic benefits is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of green energy of wind power and photovoltaic, in particular to a low-carbon park capacity configuration and optimization method of photovoltaic and electric heavy truck. BACKGROUND

[0002] Energy shortage and ecological environment problems have become increasingly prominent, which promotes China's energy reform. Xinjiang region of China has long sunshine time and sufficient light resources, and a large amount of photovoltaic resources is not reasonably utilized every year. Meanwhile, photovoltaic is a kind of clean energy. Therefore, how to better utilize photovoltaic resources, it is necessary to construct a photovoltaic power station in the park, supply the electricity generated by photovoltaic to pure electric heavy truck and basic load in the park, and create a green power supply mode of "photovoltaic power station-charging pile-pure electric heavy truck", so as to solve the imbalance between the electricity purchase cost and the benefit of high energy consumption enterprises, which is a problem to be solved at present.

[0003] At present, scholars at home and abroad have carried out a large number of researches on low carbon in the field of transportation. However, there are relatively few researches on low carbon in the field of high energy consumption enterprises and heavy trucks, so the low carbon transformation of pure electric heavy truck transportation and high energy consumption enterprises is also an important goal. The combination of electric power and transportation field can realize low carbon emission reduction, so the utilization of low carbon clean technology and the development of pure electric heavy truck energy saving and emission reduction are extremely important.

[0004] At present, there are a few researches on the sequential charging modeling of electric vehicles, and there are few researches on the two-way interaction between renewable energy and electric vehicles. For the source-load two-way interaction, the existing technology separately studies photovoltaic power generation and electric vehicles, or studies the combination of photovoltaic power generation and electric vehicles, but there are few researches on the combination of photovoltaic power generation and pure electric heavy truck. SUMMARY

[0005] In order to solve the problems existing in the prior art, the purpose of the present application is to provide a low-carbon park capacity configuration and optimization method of photovoltaic and electric heavy truck. Firstly, the charging capacity of pure electric heavy truck is simulated based on the Monte Carlo method. Then, the peak, flat and valley periods are scientifically divided by using the neural network clustering method, and the time-of-use electricity price and pure electric heavy truck charging capacity are optimized based on the fuzzy demand response mechanism of the Logistic function. Finally, photovoltaic power generation is used as the main power supply mode on the source side, and an energy storage device is introduced. The upper layer takes the minimum investment cost as the target, and the capacity of the constructed device is constrained. The lower layer takes the minimum scheduling cost as the target, an economic optimal model of the lower layer is established, and the KKT (karush-kuhn-tucker) condition is transformed into a single-level programming problem for solving. Through two-level optimization scheduling, the best capacity of the system can be configured, so as to realize a low-carbon system and reduce the system cost.

[0006] In order to achieve the above purpose, the technical scheme of the present application is:

[0007] The low-carbon park capacity configuration and optimization method of photovoltaic and electric heavy truck first simulates the charging capacity of pure electric heavy truck based on the Monte Carlo method; then the peak, flat and valley periods are scientifically divided by using the neural network clustering method, the time-of-use electricity price and the charging capacity of pure electric heavy truck are optimized based on the fuzzy demand response mechanism of the Logistic function and by using the non-dominated sorting genetic algorithm; finally, the source side adopts photovoltaic power generation as the main power supply mode, and the energy storage device is introduced, the upper layer takes the minimum investment cost as the target, and the capacity of the constructed device is constrained, the lower layer takes the minimum scheduling cost as the target, the lower layer economic optimization model is established, and the KKT condition is transformed into a single-level programming problem for solving; through two-level optimization scheduling, the best capacity of the system is configured, so as to realize the low-carbon system and reduce the system cost.

[0008] Further, the above method consists of the following steps: step 1: overall architecture

[0009] It includes the power generation side, the energy storage side and the power consumption side; the power generation side is composed of photovoltaic power generation unit and power distribution network; the energy storage side is arranged with energy storage device; the power consumption side is composed of basic power load, i.e. public power consumption including security system, elevator and lighting, household power consumption including washing machine, television and cooking, and electric heavy truck charging;

[0010] The photovoltaic power generation is connected with the upper power grid to ensure the normal charging of the pure electric heavy truck. During the charging process of the pure electric heavy truck, during the day, the electricity generated by the photovoltaic power generation is preferentially supplied to the charging of the heavy truck and the lighting of the carbon fiber chemical plant, and the remaining electricity is stored in the energy storage device. After the energy storage is full, it is sold to the upper power grid. At night, the energy storage device preferentially generates electricity to supply the pure electric heavy truck and the lighting of the carbon fiber chemical plant, and then the upper power grid supplies power as an auxiliary;

[0011] Step 2: low-carbon modeling of carbon fiber chemical plant park energy system

[0012] The model includes photovoltaic power station output, pure electric heavy truck charging capacity and time-of-use electricity price optimization model, and the detailed construction process of the model is as follows:

[0013] Step 2.1 Establishment of photovoltaic power station output model

[0014] The output of the photovoltaic power station can be represented by the capacity of the photovoltaic power station and the unit output coefficient:

[0015] (1)

[0016] In the formula: is the photovoltaic output at time t; is the capacity of the photovoltaic power station; is the normalized photovoltaic predicted output; and The carbon fiber factory basic power at time t, the pure electric heavy truck charging power at time t, respectively;

[0017] Step 2.2 Pure electric heavy truck charging model

[0018] The Monte Carlo method is used to simulate the charging capacity at each time when a large number of pure electric heavy trucks are charged in disorder;

[0019] The following assumptions are made for the model: the probability density function of the driving distance is:

[0020] (2)

[0021] In the formula: and The variance and mean of the daily driving distance are 0.34 and 5.8, respectively; is the daily driving distance of the i-th pure electric heavy truck;

[0022] The probability density function of the charging start time is:

[0023] (3)

[0024] In the formula: and The variance and mean of the charging start time are 3.4 and 17.6, respectively; is the return time of the i-th pure electric heavy truck. The charging time calculation formula of the i-th pure electric heavy truck is:

[0025] (4)

[0026] In the formula: is the state of charge of the i-th pure electric heavy truck when it returns; is the charging power of the i-th pure electric heavy truck; E is the rated capacity of the pure electric heavy truck battery; is the charging efficiency; is related to the daily driving distance and the driving distance of the pure electric heavy truck:

[0027] (5)

[0028] In the formula: is the driving distance of the pure electric heavy truck;

[0029] The Monte Carlo method is used to simulate the charging situation of 100 pure electric heavy trucks, and the charging load characteristics of each pure electric heavy truck in each time period are obtained. Then, the load characteristics of each pure electric heavy truck are superimposed to obtain the total charging capacity data of the pure electric heavy truck in each time period;

[0030] (6)

[0031] In the formula: is the load demand of the pure electric heavy truck at time t before the load demand response; is the charging power of the ith pure electric heavy truck; is used to determine whether the pure electric heavy truck is charging, 1 means charging, and 0 means no charging.

[0032] Step 2.3 Time-of-use electricity price optimization of pure electric heavy truck charging capacity

[0033] SOM self-organizing mapping network is adopted to map high-dimensional data to low-dimensional space with simple structure and correlation for display, thereby realizing data visualization, clustering and classification function. The basic structure of SOM network is a neural network model for data visualization and clustering. The figure contains the following two main parts: the input layer represents the input of high-dimensional original data, usually composed of multiple neurons, each neuron corresponds to an input feature; the output layer is composed of a low-dimensional neuron grid, used to map and display the clustering results of input data; through the competitive learning mechanism, the output layer neurons can adaptively reflect the topological structure of the input data;

[0034] 3. Photovoltaic-pure electric heavy truck low-carbon park double optimization configuration model

[0035] Double-layer planning is a two-layer model, which is an upper model and a lower model. Each layer model has its own objective function and constraint condition, and the upper and lower models depend on and restrict each other. The upper model configures the capacity of the system equipment, the lower model optimizes the scheduling, and the upper and lower models are coordinated and optimized to obtain the optimal configuration capacity and optimal scheduling result.

[0036] 3.1 Upper planning model

[0037] The upper objective function is to minimize the investment cost, and the objective function includes equipment investment and maintenance cost, equipment residual value income, and equipment depreciation cost. The specific objective function is described as:

[0038] (12)

[0039] In the formula: is the annual investment, maintenance cost of equipment, etc.; is the equipment depreciation cost; is the equipment residual value income.

[0040] 3.2 Lower optimization scheduling model

[0041] The lower target function is the lowest cost, and the target function includes the system daily operation and maintenance cost, the system daily electricity purchase cost, the system daily carbon emission penalty cost, and the system daily electricity sale revenue.

[0042] (18)

[0043] In the formula: is the system daily operation cost; is the system daily electricity purchase cost; is the system daily carbon emission penalty cost; is the system daily electricity sale revenue.

[0044] 1) System daily operation cost

[0045] (19)

[0046] In the formula: T is the daily scheduling time, which is 24 hours; , , are the operation cost coefficients of photovoltaic, energy storage device, and charging pile respectively.

[0047] 2) System daily electricity purchase cost

[0048] (20)

[0049] In the formula: is the time-of-use electricity price; is the system electricity purchase power;

[0050] 3) System daily electricity sale revenue

[0051] (21)

[0052] In the formula: is the electricity sale price to the grid; is the system electricity sale power;

[0053] 4) System daily carbon emission penalty cost

[0054] (22)

[0055] In the formula: is the carbon trading price, 75 yuan / ton; is the carbon emission of CO2 per unit of power, 0.9 tons / MW·h; (The present application does not consider carbon emission quota)

[0056] 3.2.1 Power balance constraint

[0057] System power balance constraint

[0058] (23)

[0059] wherein: is the photovoltaic output at time t; is the electricity purchase power from the upper grid at time t; is the electricity sale power to the upper grid at time t; is the energy storage discharge power at time t; is the carbon fiber chemical plant base load power at time t; is the green heavy truck charging power at time t; is the energy storage charging power at time t;

[0060] 3.3 Model solving

[0061] The model solving process of the application is divided into capacity configuration and optimization scheduling two parts, and the specific process is as shown in Figure 4

[0062] It can be seen from Figure 4 that the pure electric heavy truck charging quantity simulation data, the base load prediction data and the photovoltaic prediction data are normalized as planning data support. The planning solving model is a double-layer, wherein the upper layer is a planning model, and the minimum total cost of daily / year configuration is taken as the objective function; the lower layer is a low-carbon optimization operation model, and the minimum daily operation and maintenance cost is taken as the objective function. The upper layer model generates a set of optimal equipment capacity, and passes it to the lower layer, and the lower layer optimizes scheduling according to the variables passed by the upper layer, and the optimization scheduling depends on the optimal solution of the lower layer model. The optimal capacity and optimal scheduling result of the system are solved through iteration, the Lagrange function of the lower layer model is established, and then the lower layer model is converted into the constraint condition of the upper layer model by using the KKT condition. The two are realized through iterative interaction to realize collaborative optimization, and finally the optimal capacity configuration and low-carbon scheduling scheme of the system are obtained.

[0063] Further, the step 2.3 is specifically:

[0064] Step 2.3.1 SOM training process

[0065] In the training process, the SOM neural network optimizes the parameters of the network by continuously adjusting the weight values, so that the network can adaptively learn the distribution law of the input data. In the competition layer, each neuron will competitively process the input data, and similar data will be in adjacent positions in the competition layer, and different data will be in different positions in the competition layer. In this way, the SOM neural network can automatically discover the internal relationship between the input data and map it to the topological structure of the competition layer;

[0066] Step 2.3.2 Load transfer rate model based on Logistic function

[0067] ​To improve the accuracy of the time-of-use pricing factor and make it suitable for the actual load curve, the function model is as follows;

[0068] (7)

[0069] In the formula: Indicates the electricity price difference; Indicates the load transfer rate; , , These represent the known quantities in the Logistic function; Indicates a variable parameter;

[0070] Different parameters are used to calculate the "dead zone", "response zone", and "saturation zone", and the specific formulas are as follows;

[0071] (8)

[0072] (9)

[0073] In the formula: Indicates the actual load transfer rate; and These represent the load transfer rates for pessimistic and optimistic response forecasts, respectively. and This indicates the regional boundary points where electricity price differences are divided; The membership degree of the optimistic response is represented; then the actual load transfer rate from peak to flat and from flat to valley is calculated in turn; the values ​​after the user demand response are shown in Formula 10 and Formula 11, respectively.

[0074] (10)

[0075] (11)

[0076] In the formula: , , These represent peak, flat, and trough periods, respectively.

[0077] , ,and This represents the average load value for each time period before the implementation of peak-valley electricity pricing; express The amount of load shifting caused by real-time demand response; , These represent the load values ​​at time t before and after the implementation of peak-valley electricity pricing, respectively.

[0078] 4. The method according to claim 3, characterized in that: step 3.1, the upper-level planning model, specifically comprises:

[0079] 3.1.1 Upper-level objective function

[0080] 1) Annual value of equipment investment and maintenance costs

[0081] (13)

[0082] In the formula: For photovoltaic power stations, For energy storage devices, For charging stations; , This refers to the equipment's service life and depreciation rate after use. Unit investment cost of equipment; Equipment capacity (unit: MW·h); This is the annual maintenance cost coefficient for the equipment, which is 20% of the investment cost per unit capacity.

[0083] 2) Residual value income of equipment

[0084] (14)

[0085] In the formula: This is the residual value coefficient of the equipment;

[0086] 3) Equipment depreciation costs

[0087] (15)

[0088] 3.1.2 Upper-level constraints

[0089] 1) Equipment capacity constraints

[0090] (16)

[0091] In the formula: This is the normalized photovoltaic power output coefficient;

[0092] 2) Energy storage operation constraints

[0093] (17)

[0094] In the formula: This represents the initial electrical charge in the energy storage device. This represents the initial electrical charge in the energy storage device. Let be the amount of electricity stored in the energy storage device at time t; , This represents the charging and discharging operation status of the energy storage device, a 0-1 variable. , is the charging and discharging power of the energy storage at time t; , is the maximum charging and discharging power of the energy storage; , is the charging and discharging proportion coefficient of the energy storage.

[0095] Compared with the prior art, the present application has the following beneficial effects:

[0096] In the low-carbon park capacity configuration and optimization method of photovoltaic and electric heavy trucks:

[0097] 1) The carbon fiber chemical plant uses a new type of photovoltaic power station to supply power for the basic load of the park and pure electric heavy trucks. Although it increases the investment cost of the initial system configuration, it greatly reduces the total cost of the system. Compared with mode 1, the investment and construction cost of mode 3 and mode 2 can be recovered within one year, and the later operation cost is reduced by 75% and 71%.

[0098] 2) Considering the time-of-use electricity price and the newly built photovoltaic power station, the carbon emission of mode 3 is reduced by 1.8% compared with mode 2, and the carbon emission of mode 3 is reduced by 163% compared with mode 1. Reducing costs while achieving low carbon can effectively achieve low carbon and economic operation. BRIEF DESCRIPTION OF DRAWINGS

[0099] Figure 1 Light-Storage-Electric Heavy Truck Low-Carbon Park Energy System;

[0100] Figure 2 SOM network structure;

[0101] Figure 3 User actual response mechanism model based on Logistic function;

[0102] Figure 4 Planning solution flowchart;

[0103] Figure 5 Probability density and probability distribution of daily driving mileage;

[0104] Figure 6 Probability density and probability distribution of charging start time;

[0105] Figure 7 Daily driving mileage and required charging amount scatter point distribution;

[0106] Figure 8 Charging duration and charging start time scatter point distribution;

[0107] Figure 9 Pure electric heavy truck charging amount diagram;

[0108] Figure 10Time-of-use price optimization pure electric heavy truck charging results

[0109] Figure 11 Mode 1 scheduling results

[0110] Figure 12 Mode 2 scheduling results

[0111] Figure 13 Mode 3 scheduling results. DETAILED DESCRIPTION

[0112] The technical solutions of the present application will be further described in detail below in combination with the drawings and specific embodiments:

[0113] As Figures 1-13 shown,

[0114] The low-carbon park capacity configuration and optimization method of photovoltaic and electric heavy truck, first, based on the Monte Carlo method, the charging capacity of pure electric heavy truck is simulated. Then the peak, flat and valley period is scientifically divided by using the neural network clustering method, and the fuzzy demand response mechanism based on the Logistic function is used as the basis and the non-dominated sorting genetic algorithm is used to optimize the time-of-use electricity price and the charging capacity of pure electric heavy truck. Finally, the source side uses photovoltaic power generation as the main power supply mode, and introduces energy storage devices, the upper layer takes the minimum investment cost as the target, and the capacity of the constructed device is constrained, the lower layer takes the minimum scheduling cost as the target, establishes the lower economic optimal model, and uses KKT (karush-kuhn-tucker) condition to transform into single-level programming problem for solving. Through two-level optimization scheduling, the best capacity of the system can be configured, so as to realize the low-carbon system and reduce the system cost.

[0115] 1 Overall architecture

[0116] Using photovoltaic to replace self-provided power plant can effectively reduce the carbon emissions in the process of power generation, in order to ensure the reliability of power supply for pure electric heavy truck and basic load, energy storage devices need to be added and connected with the upper power grid to reduce the economic risk caused by power supply shortage. The overall architecture of the low-carbon park capacity configuration and optimization operation strategy of photovoltaic and pure electric heavy truck is shown as Figure 1 .

[0117] As Figure 1 can be seen, the low-carbon park energy system includes power generation side, energy storage side and power consumption side. The power generation side is composed of photovoltaic power generation unit and distribution network; the energy storage side is arranged with energy storage devices; the power consumption side is composed of basic electric power load, i.e. public power consumption including security system, elevator and lighting, household power consumption including washing machine, television and cooking, and electric heavy truck charging.

[0118] The photovoltaic power generation is connected with the upper power grid, to ensure the normal charging of the pure electric heavy truck, during the charging process of the pure electric heavy truck, during the day, the electricity generated by the photovoltaic power generation is preferentially supplied to the charging of the heavy truck and the lighting of the carbon fiber chemical plant, the remaining electricity is stored in the energy storage device, after the energy storage is full, it is sold to the upper power grid; at night, the energy storage device preferentially generates electricity to supply the pure electric heavy truck and the lighting of the carbon fiber chemical plant, and then the upper power grid supplies power.

[0119] 2 Carbon fiber chemical plant park energy system low carbon modeling

[0120] The model provided by the application mainly includes a photovoltaic power station output, a pure electric heavy truck charging capacity and a time-of-use electricity price optimization model, and the detailed construction process of the model is as follows:

[0121] 2.1 Photovoltaic power station output model establishment

[0122] The output of the photovoltaic power station can be represented by the capacity of the photovoltaic power station and the unit output coefficient:

[0123] (1)

[0124] In the formula, is the photovoltaic output at t time; is the capacity of the photovoltaic power station; is the photovoltaic predicted output normalization; and are respectively the basic electricity power of the carbon fiber chemical plant at t time and the charging power of the pure electric heavy truck at t time;

[0125] Step 2.2 pure electric heavy truck charging model

[0126] The Monte Carlo method is used to simulate the charging capacity at each time when large-scale pure electric heavy trucks are charged in disorder;

[0127] The following assumptions are made for the model: the probability density function of the driving mileage is:

[0128] (2)

[0129] In the formula, and are respectively the variance and the mean value of the daily driving mileage, and the values are 0.34 and 5.8; is the daily driving mileage of the i-th pure electric heavy truck;

[0130] The probability density function of the charging start time is:

[0131] (3)

[0132] In the formula, and The variance and mean of the charging start time, respectively, are 3.4 and 17.6; is the return time of the ith pure electric heavy truck; the charging time calculation formula of the ith pure electric heavy truck is:

[0133] (4)

[0134] In the formula: is the state of charge of the ith pure electric heavy truck when it returns; is the charging power of the ith pure electric heavy truck; E is the rated capacity of the battery of the pure electric heavy truck; is the charging efficiency; is related to the daily driving distance and the cruising range of the pure electric heavy truck:

[0135] (5)

[0136] In the formula: is the driving distance of the pure electric heavy truck;

[0137] The Monte Carlo method is used to simulate the charging of 100 pure electric heavy trucks, and the charging load characteristics of each pure electric heavy truck in each time period are obtained. Then, the load characteristics of each pure electric heavy truck are superimposed to obtain the total charging amount data of pure electric heavy trucks in each time period;

[0138] (6)

[0139] In the formula: is the load demand of the pure electric heavy truck at time t before the load demand response; is the charging power of the ith pure electric heavy truck; is used to determine whether the pure electric heavy truck is charging, 1 means charging, and 0 means no charging;

[0140] Step 2.3 Time-of-use electricity price optimization of pure electric heavy truck charging amount

[0141] SOM self-organizing mapping network is used to map high-dimensional data to low-dimensional space with simple structure and correlation for display, thereby realizing data visualization, clustering, and classification functions. The basic structure of SOM network is a neural network model for data visualization and clustering. It consists of two main parts: the input layer, which represents the input of high-dimensional original data, usually composed of multiple neurons, each corresponding to an input feature; the output layer, composed of a low-dimensional neuron grid, used to map and display the clustering results of input data; through the competitive learning mechanism, the output layer neurons can adaptively reflect the topological structure of input data;

[0142] 3. Photovoltaic-pure electric heavy truck low-carbon park double optimization configuration model

[0143] The double-layer planning has two models, i.e., an upper model and a lower model, and each model has its own objective function and constraint condition, and the upper and lower models depend on and restrict each other, the upper model performs capacity configuration of system equipment, the lower model performs optimization scheduling, the upper and lower models are coordinated and optimized to obtain optimal configuration capacity and optimal scheduling result;

[0144] 3.1 Upper planning model

[0145] The upper objective function is to minimize investment cost, and the objective function includes equipment investment and maintenance cost, equipment residual value income and equipment depreciation cost; the specific objective function is described as:

[0146] (12)

[0147] In the formula: is the annual investment, maintenance cost of equipment and the like; is the equipment depreciation cost; is the equipment residual value income;

[0148] 3.2 Lower optimization scheduling model

[0149] The lower objective function is to minimize cost, and the objective function includes system daily operation and maintenance cost, system daily electricity purchase cost, system daily carbon emission penalty cost and system daily electricity sale income;

[0150] (18)

[0151] In the formula: is the system daily operation cost; is the system daily electricity purchase cost; is the system daily carbon emission penalty cost; is the system daily electricity sale income;

[0152] 1) System daily operation cost

[0153] (19)

[0154] In the formula: T is daily scheduling time, which is 24 hours; 、 、 are respectively operation cost coefficients of photovoltaic, energy storage device and charging pile;

[0155] 2) System daily electricity purchase cost

[0156] (20)

[0157] In the formula: For time-of-use electricity price; For system to buy electricity power;

[0158] 3) System daily electricity sales revenue

[0159] (21)

[0160] In the formula: P is the electricity price sold to the power grid; P is the system electricity sales power;

[0161] 4) System daily carbon emission penalty cost

[0162] (22)

[0163] In the formula: C is the carbon trading price, 75 yuan / ton; CO2 is the carbon emission per unit of power, 0.9 tons / MW·h; (The present application does not consider carbon emission quota)

[0164] 3.2.1 Power balance constraint

[0165] System power balance constraint

[0166] (23)

[0167] In the formula: P is the t-time photovoltaic output; P is the t-time electricity purchase power from the upper-level power grid; P is the t-time electricity sales power to the upper-level power grid; P is the t-time energy storage discharge power; P is the t-time carbon fiber chemical plant basic load power; P is the t-time green heavy truck charging power; P is the t-time energy storage charging power;

[0168] 3.3 Model solution

[0169] The model solution process of the present application is divided into two parts of capacity configuration and optimization scheduling, and the specific process is as shown in Figure 4

[0170] Figure 4 ​​It can be seen that the charging quantity simulation data of the pure electric heavy truck, the basic load prediction data and the photovoltaic prediction data are normalized as planning data support. The planning solving model is a double-layer model, wherein the upper layer is a planning model, and the minimum total cost of daily / year configuration is taken as an objective function; the lower layer is a low-carbon optimal operation model, and the minimum daily operation maintenance cost is taken as an objective function. The upper layer model generates a set of optimal equipment capacity and transmits it to the lower layer, and the lower layer optimizes scheduling according to the variables transmitted by the upper layer, while the optimal scheduling depends on the optimal solution of the lower layer model. The optimal capacity and optimal scheduling result of the system are solved by iteration, the Lagrange function of the lower layer model is established, and then the lower layer model is converted into the constraint condition of the upper layer model by using the KKT condition. The two are iteratively interacted to realize collaborative optimization, and finally the optimal capacity configuration and low-carbon scheduling scheme of the system are obtained.

[0171] 4 Example analysis

[0172] 4.1 Example setting

[0173] The power consumption load and photovoltaic power generation output data of a carbon fiber chemical plant park in Xinjiang are selected for analysis, the operation life of the photovoltaic power station, the energy storage device and the charging pile device in the park is 15 years, 20 years and 12 years respectively. The discount rate is 0.067, the equipment participation value coefficient is 0.067, the equipment fixed maintenance cost coefficient is 0.02, the carbon emission coefficient of power purchase is 0.9, the carbon trading price is 75 yuan / ton, and the unit capacity configuration cost of photovoltaic, energy storage and charging pile is 2000 yuan / kW, 2600 yuan / kW and 1700 yuan / kW respectively.

[0174] The present application sets three kinds of operation modes. Operation mode 1: without considering photovoltaic and time-of-use electricity price, only using power purchase to supply power for the basic load and pure electric heavy truck operation in the park. Operation mode 2: without considering time-of-use electricity price, using photovoltaic power generation and power purchase to supply power for the basic load and pure electric heavy truck operation in the park. Operation mode 3: considering time-of-use electricity price, using photovoltaic power generation and power purchase to supply power for the basic load and pure electric heavy truck operation in the park.

[0175] 4.2 Example result analysis

[0176] 4.2.1 Monte Carlo method simulation charging quantity result

[0177] The probability density and probability distribution of the driving mileage of the pure electric heavy truck and the charging start time in the Monte Carlo method are as shown in Figure 5 , Figure 6 .

[0178] The daily driving mileage of the pure electric heavy truck mainly concentrates on 300-400km, and the charging start time mainly concentrates on 15:00-20:00. In the Monte Carlo method, the driving mileage, charging quantity, charging time required, and charging start time of 100 vehicles are simulated according to the probability distribution, and the results are shown in Figure 7 , Figure 8 .

[0179] The charging quantity of the pure electric heavy truck in 24 hours is shown in Figure 9 . The peak period of the charging quantity of the pure electric heavy truck mainly concentrates on 18:00-23:00.

[0180] 4.2.2 Charging quantity results optimized by time-of-use electricity price

[0181] Firstly, the SOM neural network clustering method is used to divide the peak and valley periods, and then the obtained results are verified and adjusted by using the load change rate index to obtain a reasonable and accurate scheme; secondly, based on the fuzzy demand response mechanism of the Logistic function, finally, the non-dominated sorting genetic algorithm is used to optimize the time-of-use electricity price and the charging quantity of the pure electric heavy truck, and the load comparison before and after the optimization of the time-of-use electricity price is shown in Figure 10 . The time-of-use electricity price is shown in Table 1:

[0182] Table 1 Electricity price at different purchase times

[0183] Peak time 17:00-23:00 Valley time 1:00,14:00,16:003:00-11:00 Flat price 2:00,24:00,14:0012:00-13:00 Electricity price 0.796 yuan / kW·h 0.247 yuan / kW·h 0.618 yuan / kW·h

[0184] The electricity price in Table 1 is used as the electricity price guided by the power grid to purchase electricity in the park, and the charging quantity is high when the electricity price is low, and the charging quantity is low when the electricity price is high. The time-of-use electricity price optimizes the load, reduces the load peak-valley difference, and has the function of peak clipping and valley filling.

[0185] 4.2.3 System capacity configuration and optimized scheduling results

[0186] The photovoltaic power generation is used as the incremental distribution network of the carbon fiber chemical plant, which will increase the investment cost of the whole system in the early stage, but has a substitution effect on the purchase of electricity. Most of the purchased electricity comes from thermal power generation, which will produce carbon emissions and affect the environment. The photovoltaic power generation will replace the thermal power generation, reduce the carbon emissions of the system, and help achieve the goal of carbon peak and carbon neutral.

[0187] By comparing the three modes, using photovoltaic power generation to supply power for the basic load in the park and the operation of the pure electric heavy truck, and using the time-of-use electricity price to optimize the charging quantity of the pure electric heavy truck, the peak-valley difference of the charging quantity of the pure electric heavy truck is reduced, the carbon emissions are reduced, and the scheduling cost is reduced. The configuration results and annual scheduling results in the present application are shown in Table 2. Table 2 Configuration results and annual operation cost of different modes

[0188] Operating mode Photovoltaic power station capacity (MW) Charging pile capacity (MW) Energy storage device capacity (1000 m³) Configuration cost (ten thousand yuan) Annual carbon emissions (ten thousand tons) Annual dispatching cost (ten thousand yuan) Total cost (ten thousand yuan) 1 0 2.55 0 87.23 4.37 5845.07 5932.30 2 16.54 2.55 81.47 3608.87 -2.71 1700.54 5309.41 3 16.73 2.36 85.39 3751.73 -2.76 1466.17 5217.90

[0189] As can be seen from Table 2, the total cost of mode 2 and mode 3 is reduced by 622.89 million yuan and 714.4 million yuan respectively relative to mode 1, the configuration cost of mode 2 and mode 3 is higher than that of mode 1 by 3521.64 million yuan and 3664.5 million yuan, but the annual scheduling cost is reduced by 71% and 75% relative to mode 1, and the configuration cost of mode 3 is higher than that of mode 2 by 142.86 million yuan, but the annual scheduling cost is reduced by 13.8%. Therefore, it can be concluded that the mode of using photovoltaic power generation and purchasing power proposed in the present application is to use photovoltaic power generation and purchasing power to supply power for the basic load of the park and the pure electric heavy truck, and relative to the traditional pure power purchase mode, although the configuration cost in the early stage is increased, in the long run, the scheduling cost and the total cost are significantly reduced relative to the traditional mode, and low carbon and economy are realized.

[0190] 1) The carbon fiber chemical plant uses a new type of photovoltaic power station to supply power for the basic load of the park and the pure electric heavy truck, although the investment cost of the system configuration in the early stage is increased, the total cost of the system is greatly reduced, the investment and construction cost of mode 3 and mode 2 is recovered within one year compared with mode 1, and the running cost in the later stage is reduced by 75% and 71%.

[0191] 2) Considering the time-of-use electricity price and the newly built photovoltaic power station, the carbon emission of mode 3 is reduced by 1.8% compared with mode 2, and the carbon emission of mode 3 is reduced by 163% compared with mode 1, so that low carbon and economy are realized, and low carbon and economic operation are effectively realized.

[0192] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any change or replacement without creative labor should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be limited by the protection scope defined in the claims.

Claims

1. A method for capacity configuration and optimization of low-carbon park of photovoltaic and electric heavy-duty truck, characterized in that: Firstly, the charging amount of pure electric heavy truck is simulated based on Monte Carlo method; then the peak, flat and valley periods are scientifically divided by using neural network clustering method, and the time-of-use electricity price and pure electric heavy truck charging amount are optimized based on the fuzzy demand response mechanism of Logistic function and the non-dominated sorting genetic algorithm; finally, the source side adopts photovoltaic power generation as the main power supply mode, and introduces energy storage device, the upper layer takes the minimum investment cost as the target, and the capacity of the constructed device is constrained, the lower layer takes the minimum scheduling cost as the target, and the lower layer economic optimal model is established, and the KKT condition is transformed into single-level programming problem for solving; Through two-level optimization scheduling, the best capacity of the system is configured, so as to realize low-carbon system and reduce system cost.

2. The method of claim 1, wherein: The following steps constitute: step 1: overall architecture It includes power generation side, energy storage side and power consumption side; the power generation side is composed of photovoltaic power generation unit and power distribution network; the energy storage side is arranged with energy storage device; the power consumption side is composed of basic power load, i.e. public power consumption including security system, elevator and lighting, household power consumption including washing machine, television and cooking, and electric heavy truck charging; During the charging process of pure electric heavy truck, photovoltaic power generation is connected with the upper power grid to ensure the normal charging of pure electric heavy truck, during the day, the generated electricity of photovoltaic power generation is preferentially supplied to heavy truck charging and carbon fiber plant lighting, the remaining electricity is stored in energy storage device, after the energy storage is full, it is sold to the upper power grid; at night, the energy storage device preferentially generates electricity to supply pure electric heavy truck and carbon fiber plant lighting, and then the upper power grid assists power supply; Step 2: low-carbon modeling of carbon fiber plant park energy system The model includes photovoltaic power station output, pure electric heavy truck charging amount and time-of-use electricity price optimization model, and the detailed construction process of the model is as follows: Step 2.1: photovoltaic power station output model establishment The output of photovoltaic power station can be represented by photovoltaic power station capacity and unit output coefficient: （1） In the formula: is the photovoltaic output at time t; is the photovoltaic power station capacity; is the photovoltaic predicted output normalization; and are the carbon fiber chemical plant base electricity power at time t and the pure electric heavy truck charging power at time t, respectively. Step 2.2: pure electric heavy truck charging model Monte Carlo method is used to simulate the charging amount at each time when large-scale pure electric heavy truck charges in disorder; The following assumptions are made for the model: the probability density function of driving mileage is: （2） In the formula: and are the variance and mean of daily travel distance, respectively, with values of 0.34 and 5.8; is the daily travel distance of the ith pure electric heavy truck. The probability density function of charging start time is: （3） In the formula: and are the variance and mean of the charging start time, respectively, with values of 3.4 and 17.6, respectively; is the return time of the ith pure electric heavy truck; the charging time calculation formula of the ith pure electric heavy truck is: （4） In the formula: is the state of charge when the i-th pure electric heavy truck returns; is the charging power of the i-th pure electric heavy truck; E is the rated capacity of the battery of the pure electric heavy truck; is the charging efficiency; is related to the daily driving mileage and the cruising range of the pure electric heavy truck: （5） In the formula: is the driving range of the pure electric heavy truck; The charging situation of 100 pure electric heavy trucks is simulated by using Monte Carlo method, the charging load characteristics of each pure electric heavy truck in each time period are obtained, and then the load characteristics of each pure electric heavy truck are superimposed to obtain the total charging amount data of pure electric heavy truck in each time period; （6） In the formula: is the load demand of the pure electric heavy truck at time t before load demand response; is the charging power of the ith pure electric heavy truck; is used to determine whether the pure electric heavy truck is charging, 1 indicates that it is charging, and 0 indicates that it is not charging; Step 2.3: time-of-use electricity price optimization of pure electric heavy truck charging amount SOM self-organizing mapping network is used to map high-dimensional data to low-dimensional space with simple structure and correlation to realize data visualization, clustering and classification function. The basic structure of SOM network is a neural network model for data visualization and clustering; it contains the following two main parts: input layer, representing the input of high-dimensional original data, usually composed of multiple neurons, each neuron corresponds to an input feature; output layer, composed of low-dimensional neuron grid, used to map and display the clustering results of input data; through competitive learning mechanism, the output layer neurons can adaptively reflect the topological structure of input data; 3. A double-optimization configuration model for photovoltaic-pure electric heavy-duty trucks in a low-carbon park Double-layer planning is a two-layer model, including an upper layer model and a lower layer model. Each layer has its own objective function and constraint conditions, and the upper and lower layer models depend on and restrict each other. The upper layer model configures the capacity of system equipment, the lower layer model optimizes scheduling, and the upper and lower layers are coordinated and optimized to obtain the optimal configuration capacity and optimal scheduling result. 3.1 Upper layer planning model The upper layer objective function is to minimize the investment cost, including equipment investment and maintenance cost, equipment residual value income, and equipment depreciation cost. The specific objective function is described as follows: （12） wherein: is the annual investment in equipment, etc., maintenance costs; is the depreciation of the equipment; is the residual value of the equipment. 3.2 Lower layer optimization scheduling model The lower layer objective function is to minimize the cost, including system daily operation and maintenance cost, system daily electricity purchase cost, system daily carbon emission penalty cost, and system daily electricity sale income. （18） In the formula: is the system day operating cost; is the system day electricity purchase cost; is the system day carbon emission penalty cost; is the system day electricity sale revenue; 1) System daily operation cost （19） In the formula, T is the daily scheduling time, which is 24 hours; , , are the operation cost coefficients of photovoltaic, energy storage device, and charging pile, respectively; 2) System daily electricity purchase cost （20） In the formula: is the time-of-use electricity price; is the system electricity purchase power; 3) System daily electricity sale income （21） In the formulae: is the electricity price for selling electricity to the grid; is the system sold electricity power; 4) System daily carbon emission penalty cost （22） In the formula: is the carbon trading price, 75 yuan / ton; is the carbon emission per unit of power, 0.9 tons / MW·h; (the present application does not consider carbon emission quota) 3.2.1 Power balance constraint System power balance constraint （23） In the formula: is the photovoltaic output at time t; is the power purchased from the upper-level power grid at time t; is the power sold to the upper-level power grid at time t; is the energy storage discharge power at time t; is the power for the carbon fiber chemical plant basic load at time t; is the green heavy truck charging power at time t; is the energy storage charging power at time t; 3.3 Model solution The model solution process is divided into two parts: capacity configuration and optimization scheduling. The charging capacity simulation data of pure electric heavy-duty trucks, the basic load prediction data, and the photovoltaic prediction data are normalized as planning data support. The planning solution model is a double-layer model, in which the upper layer is a planning model, and the objective function is to minimize the total daily / yearly configuration cost. The lower layer is a low-carbon optimization operation model, and the objective function is to minimize the daily operation and maintenance cost. The upper layer model generates a set of optimal equipment capacity, which is passed to the lower layer. The lower layer optimizes scheduling according to the variables passed by the upper layer, and the optimization scheduling depends on the optimal solution of the lower layer model. The optimal capacity and optimal scheduling result of the system are obtained by iterative solution. The Lagrange function of the lower layer model is established, and then the lower layer model is converted into the constraint condition of the upper layer model by using the KKT condition. The two are iteratively interacted to achieve collaborative optimization, and finally the optimal capacity configuration and low-carbon scheduling scheme of the system are obtained.

3. The method of claim 2 wherein: The step 2.3 is specifically: Step 2.3.1 SOM training process During the training process, the SOM neural network optimizes the network parameters by continuously adjusting the weight values, so that the network can adaptively learn the distribution rules of the input data. In the competition layer, each neuron will competitively process the input data, and similar data will be located adjacent to each other in the competition layer, and different data will be located in different positions in the competition layer. In this way, the SOM neural network can automatically discover the internal relationship between the input data and map it to the topology of the competition layer. Step 2.3.2 Load transfer rate model based on Logistic function In order to improve the accuracy of the time-of-use electricity price coefficient and make it suitable for the load curve in actual situation, the function model is as follows: （7） In the formulae: represents the price difference; represents the load transfer rate; , , respectively represent known quantities in the Logistic function; represents a variable parameter; In the "dead zone", "response zone", and "saturation zone", different parameters are used for calculation, and the specific formula is as follows: （8） （9） In the formula: represents the actual load transfer rate; and respectively represent the pessimistic response prediction and the optimistic response prediction of the load transfer rate; and represent the regional demarcation point of the electricity price difference division; represents the membership degree of the optimistic response; then the actual load transfer rates of the peak-to-flat and the flat-to-valley are sequentially obtained; the values after the demand response of the user are respectively shown in the formula 10 and the formula 11; （10） （11） In the formulae: , , respectively represent peak, flat, and valley periods. , , and denotes the average load value of each time period before the peak-valley electricity price implementation; denotes the load transfer amount caused by demand response at time t; , denote the load values at time t before and after the peak-valley electricity price implementation, respectively.

4. The method of claim 3, wherein: The step 3.1 upper layer planning model is specifically: 3.1.1 Upper layer objective function 1) Annual value of equipment investment and maintenance cost （13） In the formula: is a photovoltaic power station, is an energy storage device, is a charging pile; , is the operating life and depreciation rate of the equipment after use; is the unit investment cost of the equipment; is the equipment capacity (unit: MW·h); is the annual maintenance cost coefficient of the equipment, which is 20% of the unit capacity investment cost; 2) Equipment residual value income （14） In the formula: is the equipment residual value coefficient; 3) Equipment depreciation cost （15） 3.1.2 Upper layer constraint conditions 1) Equipment capacity constraint （16） In the formula: is the normalized photovoltaic power coefficient; 2) Energy storage operation constraint （17） In the formula: is the initial electric quantity in the energy storage device; is the initial electric quantity in the energy storage device; is the storage electric quantity of the energy storage device at time t; , is the charge and discharge operation state of the energy storage device, a 0-1 variable; , is the charge and discharge power of the energy storage at time t; , is the maximum charge and discharge power of the energy storage; , is the charge and discharge proportionality coefficient of the energy storage.

Images