New energy peak regulation scheduling optimization method based on source-load coordination

Abstract

The invention discloses a new energy peak regulation scheduling optimization method based on source-load coordination. The method comprises the following steps: constructing a wind-light and load output prediction model based on wind power, photoelectricity and load historical data; the prediction model is used for predicting predicted values of the input wind power, photoelectric and load demands in the next period; comparing the wind-light-electricity and load prediction data, and calculating a light-wind-electricity output value and a load value when the similarity is maximum and the constraint of the source load volatility is satisfied; establishing a direct-current wind-light-fire combined dispatching optimization mathematical model by taking the lowest peak regulation degree of the receiving-end power grid and the lowest dispatching peak regulation operation cost as objective functions; and an improved cattle grass-eating optimization algorithm is adopted to carry out optimization solution on the direct current wind-light-fire combined scheduling optimization mathematical model, and optimization is carried out on the whole situation, so that the peak regulation degree of the power grid is lowest, and the peak regulation operation cost is lowest at the same time. On the basis of reducing the source load disturbance rate, the peak regulation cost is reduced, the wind and light abandoning amount is reduced, and the efficiency of the thermal power generating unit is optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of peak load regulation of renewable energy power systems, and in particular to a new energy peak load regulation optimization method based on source-load coordination. Background Art

[0002] Vigorously developing renewable energy such as wind power and photovoltaics is a major measure to address energy security, environmental pollution, and achieve the goals of "carbon peak" and "carbon neutrality". my country is vast and rich in resources, with abundant wind and solar energy resources. Considering the basic requirements of the site, the exploitable capacity of wind power and photovoltaic power generation technology has reached 10.9 billion kilowatts and 45.6 billion kilowatts respectively, with a potential annual power generation of nearly 100 trillion kilowatt-hours, which can meet the growth of my country's energy and electricity demand and replace fossil energy.

[0003] Because of the uneven distribution of our energy, the western and northern regions have sufficient wind and solar resources, and the local power grid has insufficient absorption capacity. The eastern and southern regions have developed economies and large electricity demand. Therefore, it is urgent to achieve cross-regional transmission of a high proportion of renewable energy. Due to the large distance span from the west and north to the east and south and high capacity requirements, the use of ultra-high voltage direct current power transmission has obvious advantages. Under normal circumstances, in areas with abundant wind and solar resources, there is sufficient sunlight and small wind volume during the day, and sufficient wind volume and weak sunlight at night. Due to the strong volatility and randomness of wind power and photovoltaic power, the supporting capacity of thermal power reserve is determined to ensure the stability and reliability of the power system. The power consumption characteristics on the load side are random and sudden. Distributed energy has the characteristics of reasonable energy efficiency, flexible operation, and low economic cost. It can provide sufficient peak power and meet the peak load demand when the smart distribution network is in the peak load state, thereby alleviating the power congestion of the transmission line and playing an effective role in regulating the peak load of the power system. Therefore, the randomness on both the source and the load side puts forward new requirements for the peak-shaving capacity of the power system dispatching. Summary of the invention

[0004] 1. Technical problems to be solved:

[0005] In response to the above technical problems, the present invention provides a new energy peak-shaving scheduling optimization method based on source-load coordination. The new energy peak-shaving optimization scheduling method improves the peak-shaving capacity of the receiving power grid to a certain extent, absorbs wind power and photovoltaic power to the greatest extent, and ensures the safe operation of DC.

[0006] 2. Technical solution:

[0007] A new energy peak load dispatch optimization method based on source-load coordination, characterized by comprising:

[0008] Step 1: Obtain historical data of wind power, photovoltaic power and load in the area to be predicted, and build a wind, photovoltaic and load output prediction model based on the historical data; the prediction model is used to predict the corresponding prediction values ​​of the input wind power output, photovoltaic output and load demand in the next period;

[0009] Step 2: Compare the wind and solar power forecast data with the load forecast data, and obtain the wind and solar power output and load values ​​that have the greatest similarity between the two based on the improved DTW, while satisfying the source-load fluctuation rate constraint;

[0010] Step 3: Based on the similarity and constraint results of step 2, a mathematical model for DC wind-solar-thermal joint dispatch optimization is established with the lowest peak-shaving degree of the power grid and the lowest dispatch peak-shaving operation cost as the objective function;

[0011] Step 4: Take the wind power and photovoltaic forecast values ​​and load demand forecast values ​​of the substation to be dispatched as input, and use the improved cow-grazing optimization algorithm to optimize and solve the DC wind-solar-fire joint dispatch optimization mathematical model, and optimize the global situation to minimize the peak-shaving degree of the power grid and the lowest peak-shaving operation cost.

[0012] Furthermore, step one specifically includes:

[0013] S11: pre-processing the wind power output data, photovoltaic output data, and area load demand data in the acquired historical data; the pre-processing includes arranging the output data of each time period in the scheduling cycle according to the order of time;

[0014] S12: taking the data of the corresponding time period of each scheduling period as the actual value of the corresponding time period of the scheduling period, taking the average value of all data of the corresponding time period of the scheduling period before the scheduling period as the estimation of the actual value of the scheduling period, and calculating the error value of the corresponding output type data in all historical scheduling periods;

[0015] S13: Add the actual value estimate of the corresponding time period of the scheduling cycle and the corresponding error value to obtain the predicted value of the corresponding time period.

[0016] Furthermore, step 2 specifically includes:

[0017] S21: A scheduling cycle is divided into T time periods on average, and the total output forecast value of renewable energy in the nth time period is calculated. And the load power demand forecast value in the nth period As two time series of DTW, the total output forecast value is the sum of the wind power output and photovoltaic output in the corresponding time period; as shown in the following formula, the total output forecast value of new energy

[0018]

[0019] In the above formula, It represents the predicted value of wind power output in the nth period; Represents the predicted value of photovoltaic output in the nth period; the predicted value of total renewable energy output and the predicted value of load power demand within a scheduling cycle are arranged into two time series as follows:

[0020]

[0021] In the above formula, F N,F Represents the forecast time series data of the total output of renewable energy in a dispatching cycle; F Represents the forecast time series data of load power demand in a scheduling cycle;

[0022] S22: Construct the distance matrix D between the forecast time series data of the total output of new energy and the forecast time series data of the load power demand. The element values ​​in the matrix are defined as follows:

[0023]

[0024] In the above formula, the matrix D is a T×T matrix, D(x,y) is the element in the xth row and yth column of the matrix D; and

[0025]

[0026] S23: Construct the cumulative cost matrix C of the distance matrix D using the dynamic balance method. The element values ​​in the matrix C are as follows:

[0027]

[0028] In the above formula, in the matrix C, define C(0,0)=0; C(x,0)=C(0,y)=+∞;

[0029] Substitute x and y into formula (4) to recursively obtain the value of each element in the matrix C. Use formula (5) to select the optimal solution. Based on the optimal solution, find an optimal curved path P from point D(1,1) to D(T,T) in the D matrix:

[0030]

[0031] In the above formula, represents the path of the distance matrix D; P represents the optimal curved path, D(p s ) is the element of the distance matrix D corresponding to the optimal curved path P;

[0032] The element corresponding to the cost matrix C when obtaining the optimal curved path P is the minimum value of the forecast error between the total output of new energy and the load power demand when the optimal curved path is used;

[0033] S24: Calculate the time series F of the total output forecast value of new energy N,F and the time series of load power demand forecast values The covariance and variance of :

[0034]

[0035] In the above formula, V is the predicted value covariance between the new energy output and load demand in the nth period of a dispatch cycle; Var,N is the variance of the predicted value of the total output of renewable energy in this dispatch period, V Var,l is the variance of the load demand forecast value of the scheduling period; is the average output forecast value of the wind farm in the dispatching period; is the average output forecast value of the photovoltaic power station in this scheduling period; is the average predicted value of load demand within the dispatching period;

[0036] S25: Substitute the above covariance of the power between the new energy and the load and the variance of the power between the new energy and the load into the following correlation coefficient equation: The correlation coefficient between the predicted new energy output and load demand in the nth period is obtained:

[0037]

[0038] S26: The absolute value of the difference between the predicted values ​​of the total output of renewable energy in the nth period and the n+1th period minus the predicted value of the load in the nth period and the n+1th period is added together, and then divided by the average predicted output value of renewable energy in a scheduling cycle to obtain the source-load fluctuation rate δ:

[0039]

[0040] S27: Construct constraints based on the obtained correlation coefficient and source-load volatility, as shown in the following formula:

[0041]

[0042] In the above formula, R N-l,min Represents the minimum value of the correlation coefficient between the predicted value of new energy output and the predicted value of load; R N-l Represents the correlation coefficient between the predicted value of new energy output and the predicted value of load, with a value range of [-1,1]; δ N-1 Indicates the source-load fluctuation rate between renewable energy output and load;

[0043] judge Does the optimal solution of satisfy the constraints of the correlation coefficient R and the source-load volatility δ of the above formula? If so, continue with the following steps; if not, take x-1, and based on formula (5), take The second optimal curved route P' is obtained, and steps S25 and S26 are repeated, and so on until the constraints of the correlation coefficient and the source load fluctuation rate of formula (8) are satisfied.

[0044] Furthermore, step three specifically includes:

[0045] S31: Obtain the transaction volume of the DC channel in the dispatching cycle and the total output forecast value of the wind and solar power in the cycle, and determine the supporting thermal power installed capacity according to the power balance relationship of the following formula:

[0046]

[0047] In the above formula, are respectively the thermal power output and the estimated total output of the DC channel in the dispatch period T; are the output forecast values ​​of wind power and photovoltaic power in the nth period of the scheduling cycle respectively;

[0048] S32: Taking the lowest peak load regulation degree of the power grid and the lowest peak load regulation operation cost as the objective function, a DC wind-solar-thermal joint dispatch mathematical model is established, as shown in the following equations (10) and (11):

[0049]

[0050] minF=f f +f w +f pv +f l +f p (11)

[0051] In the above formula, A is the peak load regulation degree of the receiving power grid; T is the total time period of a dispatching cycle; x1, x2 and x3 are the weight coefficients of the three influencing factors preset in function A; α is the preset proportional coefficient, which is usually 1000; F is the system peak load regulation operation cost; f f 、f w 、f pv are the output costs of thermal power, wind power and photovoltaic power respectively; f l Operating costs caused by load changes; p It is the penalty value for the amount of new energy not consumed;

[0052] S33: Peak load regulation operation cost of thermal power units; operation cost includes thermal power cost f f , power generation cost g , start-stop cost s and the response cost f r , as follows:

[0053]

[0054] In the above formula, N G is the total amount of thermal power units; P Gi,t , P Gi,t-1 are the output values ​​of the i-th generator set participating in the peak load regulation of the power grid in period t and period t-1 respectively; P Gi,T,min , P Gi,T , P Gi,T,max are the minimum output, output value, and maximum output of the i-th thermal power unit respectively; is the start / stop status of the i-th generator set, 0 means it is stopped, and 1 means it is put into operation; a i 、b i 、c i is the consumption characteristic coefficient of the i-th thermal power unit; is the startup cost of the i-th generator set; is the downtime cost of the i-th generator set; is the cost coefficient for increasing the output of the i-th thermal power unit; is the cost coefficient for reducing the output of the i-th thermal power unit;

[0055] S34: Wind power operation cost, which is the sum of the power generation costs of each wind farm in a dispatching cycle:

[0056]

[0057] Among them, P w is the cost per unit of wind power output; Nw represents the total number of wind farms; Pw represents the cost per unit of wind power output; represents the actual output value of the i-th wind farm;

[0058] S35: Photovoltaic operating costs pv , is the total power generation cost of each photovoltaic power station in a scheduling cycle:

[0059]

[0060] Among them, P pv is the cost per unit of photovoltaic output; N pv is the total number of photovoltaic power stations; is the actual output value of the i-th photovoltaic power station;

[0061] S36: According to different requirements in different regions, penalty costs will be imposed on unabsorbed wind and solar power. p :

[0062]

[0063] Among them, μ1 and μ2 are the penalty coefficients for wind and solar power abandonment, and the values ​​are preset according to different requirements of different regions;

[0064] S37: Calculate the operating cost when the load changes, which is the operating cost of the equivalent adjustable load f l :

[0065]

[0066] In the above formula, P l is the unit operating cost of equivalent adjustable load; Indicates the actual demand of load;

[0067] S38: Add the thermal power output and the renewable energy output to get the DC channel transmission power

[0068]

[0069] Furthermore, step three also includes a constraint on the DC channel ramp rate, as shown below:

[0070]

[0071] in, They are the lower and upper limits of the DC channel ramp rate respectively.

[0072] Furthermore, step four specifically includes:

[0073] S41: Acquire historical data of the substation to be dispatched, and obtain output forecast data of the substation based on step 1; the output forecast data includes wind power output, photovoltaic output and load demand forecast time series data;

[0074] S42: Preset thermal power unit parameters, the number of iterations K of the cow grazing algorithm, the number of iterations J of the improved differential algorithm, and the total number of populations L;

[0075] S43: Simulate the behavior of cattle eating grass, randomly select more delicious grass from the current multiple grasslands with probability for iterative update, and obtain the global optimal solution that meets the iteration conditions, that is, the optimal solution of the DC wind-solar-fire joint scheduling optimization mathematical model with the lowest peak-shaving degree of the power grid and the lowest peak-shaving operation cost.

[0076] 3. Beneficial effects:

[0077] (1) This method provides a new energy peak-shaving dispatch optimization method based on source-load coordination. It builds a wind and solar power and load output prediction model based on historical data to predict wind power output, photovoltaic output and load demand. It is easy to calculate and can be updated according to the latest data, making its prediction more accurate.

[0078] (2) In the method provided by this method, a new energy peak-shaving scheduling optimization method based on source-load coordination is provided. The wind power and load values ​​corresponding to the time when the similarity between the wind power forecast data and the load forecast data is the largest are obtained based on the improved DTW, so that the similarity between the source and the load is maximized, so that the output of new energy is best matched with the load. At the same time, the requirements of volatility must be met to avoid excessive source-load fluctuations. On the premise that the correlation coefficient and the source-load volatility meet the requirements, it provides the premise for constructing a DC wind-solar-fire joint scheduling optimization mathematical model.

[0079] (3) This method provides a new energy peak-shaving scheduling optimization method based on source-load coordination. A DC wind-solar-fire joint scheduling optimization mathematical model is constructed with the lowest cost and the lowest grid peak-shaving degree as the objective function. This can smooth the peak-shaving curve, effectively reduce the peak-shaving cost, reduce the amount of wind and solar power curtailment, and optimize the transmission efficiency of the entire grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 A flowchart of the method is shown in FIG.

[0081] Figure 2 It is a flowchart of a specific embodiment. DETAILED DESCRIPTION

[0082] The present invention will be described in detail below with reference to the accompanying drawings.

[0083] As attached Figure 1 As shown, a new energy peak load scheduling optimization method based on source-load coordination is characterized by comprising:

[0084] Step 1: Obtain historical data of wind power, photovoltaic power and load in the area to be predicted, and build a wind, photovoltaic and load output prediction model based on the historical data; the prediction model is used to predict the corresponding prediction values ​​of the input wind power output, photovoltaic output and load demand in the next period;

[0085] Step 2: Compare the wind and solar power forecast data with the load forecast data, and obtain the wind and solar power output and load values ​​that have the greatest similarity between the two based on the improved DTW, while satisfying the source-load fluctuation rate constraint;

[0086] Step 3: Based on the similarity and constraint results of step 2, a mathematical model for DC wind-solar-thermal joint dispatch optimization is established with the lowest peak-shaving degree of the power grid and the lowest dispatch peak-shaving operation cost as the objective function;

[0087] Step 4: Take the wind power and photovoltaic forecast values ​​and load demand forecast values ​​of the substation to be dispatched as input, and use the improved cow-grazing optimization algorithm to optimize and solve the DC wind-solar-fire joint dispatch optimization mathematical model, and optimize the global situation to minimize the peak-shaving degree of the power grid and the lowest peak-shaving operation cost.

[0088] Furthermore, step one specifically includes:

[0089] S11: pre-processing the wind power output data, photovoltaic output data, and area load demand data in the acquired historical data; the pre-processing includes arranging the output data of each period in the scheduling cycle according to the order of time;

[0090] S12: taking the data of the corresponding time period of each scheduling period as the actual value of the corresponding time period of the scheduling period, taking the average value of all data of the corresponding time period of the scheduling period before the scheduling period as the estimation of the actual value of the scheduling period, and calculating the error value of the corresponding output type data in all historical scheduling periods;

[0091] S13: Add the actual value estimate of the corresponding time period of the scheduling cycle and the corresponding error value to obtain the predicted value of the corresponding time period.

[0092] Furthermore, step 2 specifically includes:

[0093] S21: The dispatch cycle is evenly divided into T time periods, and the total output forecast value of renewable energy in the nth time period is And the load power demand forecast value in the nth period As two time series of DTW, the total output forecast value is the sum of the wind power output and photovoltaic output in the corresponding time period; as shown in the following formula, the total output forecast value of new energy

[0094]

[0095] In the above formula, It represents the predicted value of wind power output in the nth period; Represents the predicted value of photovoltaic output in the nth period; the predicted value of total renewable energy output and the predicted value of load power demand within a scheduling cycle are arranged into two time series as follows:

[0096]

[0097] In the above formula, F N,F Represents the forecast time series data of the total output of renewable energy in a dispatching cycle; F Represents the forecast time series data of load power demand in a scheduling cycle;

[0098] S22: Construct the distance matrix D between the forecast time series data of the total output of new energy and the forecast time series data of the load power demand. The element values ​​in the matrix are defined as follows:

[0099]

[0100] In the above formula, the matrix D is a T×T matrix, D(x,y) is the element in the xth row and yth column of the matrix D; and

[0101]

[0102] S23: Construct the cumulative cost matrix C of the distance matrix D using the dynamic balance method. The element values ​​in the matrix C are as follows:

[0103]

[0104] In the above formula, in the matrix C, define C(0,0)=0; C(x,0)=C(0,y)=+∞;

[0105] Substitute x and y into formula (4) to recursively obtain the value of each element in the matrix C. Use formula (5) to select the optimal solution. Based on the optimal solution, find an optimal curved path P from point D(1,1) to D(T,T) in the D matrix:

[0106]

[0107] In the above formula, represents the path of the distance matrix D; P represents the optimal curved path, D(p s ) is the element of the distance matrix D corresponding to the optimal curved path P;

[0108] The element corresponding to the cost matrix C when obtaining the optimal curved path P is the minimum value of the forecast error between the total output of new energy and the load power demand when the optimal curved path is used;

[0109] S24: Calculate the time series F of the total output forecast value of new energy N,F and the time series of load power demand forecast values The covariance and variance of :

[0110]

[0111]

[0112] In the above formula, V is the predicted value covariance between the new energy output and load demand in the nth period of a dispatch cycle; Var,N is the variance of the predicted value of the total output of renewable energy in this dispatch period, V Var,l is the variance of the load demand forecast value of the scheduling period; is the average output forecast value of the wind farm in the dispatching period; is the average output forecast value of the photovoltaic power station in this scheduling period; is the average predicted value of load demand within the dispatching period;

[0113] S25: Substitute the above covariance of the power between the new energy and the load and the variance of the power between the new energy and the load into the following correlation coefficient equation: The correlation coefficient between the predicted new energy output and load demand in the nth period is obtained:

[0114]

[0115] S26: The absolute value of the difference between the predicted values ​​of the total output of renewable energy in the nth period and the n+1th period minus the predicted value of the load in the nth period and the n+1th period is added together, and then divided by the average predicted output value of renewable energy in a scheduling cycle to obtain the source-load fluctuation rate δ:

[0116]

[0117] S27: Construct constraints based on the obtained correlation coefficient and source-load volatility, as shown in the following formula:

[0118] R N-l,min ≤R N-l ≤R N-l,max

[0119] δ N-l,min ≤δ N-l ≤δ N-l,max (8)

[0120] In the above formula, R N-l,min Represents the minimum value of the correlation coefficient between the predicted value of new energy output and the predicted value of load; R N-l Represents the correlation coefficient between the predicted value of new energy output and the predicted value of load, with a value range of [-1,1]; δ N-1 Indicates the source-load fluctuation rate between renewable energy output and load;

[0121] judge Does the optimal solution of satisfy the constraints of the correlation coefficient R and the source-load volatility δ of the above formula? If so, continue with the following steps; if not, take x-1, and based on formula (5), take The second optimal curved route P' is obtained, and steps S25 and S26 are repeated, and so on until the constraints of the correlation coefficient and the source load fluctuation rate of formula (8) are satisfied.

[0122] Furthermore, step three specifically includes:

[0123] S31: Obtain the transaction volume of the DC channel in the dispatching cycle and the total output forecast value of the wind and solar power in the cycle, and determine the supporting thermal power installed capacity according to the power balance relationship of the following formula:

[0124]

[0125] In the above formula, are the thermal power output and the estimated total output of the DC channel dispatch period T respectively; are the output forecast values ​​of wind power and photovoltaic power in the nth period of the scheduling cycle respectively;

[0126] S32: Taking the lowest peak load regulation degree of the power grid and the lowest peak load regulation operation cost as the objective function, a DC wind-solar-thermal joint dispatch mathematical model is established, as shown in the following equations (10) and (11):

[0127]

[0128] minF=f f +f w +f pv +f l +f p (11)

[0129] In the above formula, A is the peak load regulation degree of the receiving power grid; T is the total time period of a dispatching cycle; x1, x2 and x3 are the weight coefficients of the three influencing factors preset in function A; α is the preset proportional coefficient, which is usually 1000; F is the system peak load regulation operation cost; f f 、f w 、f pv are the output costs of thermal power, wind power and photovoltaic power respectively; f l Operating costs caused by load changes; p It is the penalty value for the amount of new energy not consumed;

[0130] S33: Peak load regulation operation cost of thermal power units; operation cost includes thermal power cost f f , power generation cost g , start-stop cost s and the response cost f r , as follows:

[0131]

[0132] In the above formula, N G is the total amount of thermal power units; P Gi,t , P Gi,t-1 are the output values ​​of the i-th generator set participating in the peak load regulation of the power grid in period t and period t-1 respectively; P Gi,T,min , P Gi,T , P Gi,T,max are the minimum output, output value, and maximum output of the i-th thermal power unit respectively; is the start / stop status of the i-th generator set, 0 means it is stopped, and 1 means it is put into operation; a i 、b i 、c iis the consumption characteristic coefficient of the i-th thermal power unit; is the startup cost of the i-th generator set; is the downtime cost of the i-th generator set; is the cost coefficient for increasing the output of the i-th thermal power unit; is the cost coefficient for reducing the output of the i-th thermal power unit;

[0133] S34: Wind power operation cost, which is the sum of the power generation costs of each wind farm in a dispatching cycle:

[0134]

[0135] Among them, P w is the cost per unit of wind power output; Nw represents the total number of wind farms; Pw represents the cost per unit of wind power output; represents the actual output value of the i-th wind farm;

[0136] S35: Photovoltaic operating costs pv , is the total power generation cost of each photovoltaic power station in a scheduling cycle:

[0137]

[0138] Among them, P pv is the cost per unit of photovoltaic output; N pv is the total number of photovoltaic power stations; is the actual output value of the i-th photovoltaic power station;

[0139] S36: According to different requirements in different regions, penalty costs will be imposed on unabsorbed wind and solar power. p :

[0140]

[0141] Among them, μ1 and μ2 are the penalty coefficients for wind and solar power abandonment, and the values ​​are preset according to different requirements of different regions;

[0142] S37: Calculate the operating cost when the load changes, which is the operating cost of the equivalent adjustable load f l :

[0143]

[0144] In the above formula, P l is the unit operating cost of equivalent adjustable load; Indicates the actual demand of load;

[0145] S38: Add the thermal power output and the renewable energy output to get the DC channel transmission power

[0146]

[0147] Furthermore, step three also includes a constraint on the DC channel ramp rate, as shown below:

[0148]

[0149] in, They are the lower and upper limits of the DC channel ramp rate respectively.

[0150] Furthermore, step four specifically includes:

[0151] S41: Acquire historical data of the substation to be dispatched, and obtain output forecast data of the substation based on step 1; the output forecast data includes wind power output, photovoltaic output and load demand forecast time series data;

[0152] S42: Preset thermal power unit parameters, the number of iterations K of the cow grazing algorithm, the number of iterations J of the improved differential algorithm, and the total number of populations L;

[0153] S43: Simulate the behavior of cattle eating grass, randomly select more delicious grass from the current multiple grasslands with probability for iterative update, and obtain the global optimal solution that meets the iteration conditions, that is, the optimal solution of the DC wind-solar-fire joint scheduling optimization mathematical model with the lowest peak-shaving degree of the power grid and the lowest peak-shaving operation cost. Specific embodiment:

[0155] The specific process of this embodiment is shown in the attached Figure 2 shown.

[0156] In the above process, the specific process of improving the optimal solution of the multi-objective problem by using the cow-eating-grass algorithm is as follows:

[0157] 1) Assume that the grassland where the cow lives is divided into n pieces, and the probability that the cow randomly chooses a piece of grassland to eat grass is ρ,

[0158] 2) Simulate the behavior of cattle grazing, that is, select some grassland from the current grassland and select more delicious grassland through random walking method.

[0159] 3) If the cow finds a better grassland, that is, ρ>rand, it will be iteratively updated according to the following formula.

[0160]

[0161] in, is the solution of the i+1th generation; is the solution of the i-th generation; L represents the step length, which obeys the two-dimensional Levy distribution; g * It is the global optimal solution.

[0162] The calculation process of L is as follows:

[0163]

[0164] Among them, μ and v obey the normal distribution and meet the following requirements:

[0165]

[0166] Wherein, β=1.5; Γ(β) is the gamma function.

[0167] 4) If the probability of the cow eating grass satisfies ρ<rand, it is updated according to the following formula:

[0168]

[0169] Where η is a random number in the interval [0,1], which satisfies uniform distribution. Belong to different types of grass.

[0170] 5) Calculate the evaluation index of the solutions obtained in step 3) and step 4). If the evaluation index of the new solution is better, then replace the current solution with the new solution. If the evaluation index of the new solution is worse than the current evaluation index, then retain the current solution.

[0171] 6) If the evaluation index of the new solution is better than the global optimal evaluation index, then update the global optimal evaluation index and the global optimal solution.

[0172] 7) Determine whether the algorithm meets the end condition. If the end condition is met, the algorithm ends and outputs the global optimal solution and the global optimal evaluation index. If the end condition is not met, continue the calculation from step 3).

[0173] 8) For the optimal choice of grassland for multiple cows, construct the following function:

[0174]

[0175] in, It means that the mth cow is grazing on the i-th grassland during the period t.

[0176] The ranking index of the i-th feasible solution in time period t is as follows:

[0177]

[0178] Among them, RI represents the selection criterion.

[0179] The random selection index of each cow is calculated by the bell-shaped membership function:

[0180]

[0181] in, is the random selection of the mth cow under the c category in period t, a c is a randomly selected center under category c; b c is the random selection width under category c; K is the function order.

[0182] 9) The fuzzy selection function is used to integrate the selections of multiple cows before and after, and the comprehensive objective function FSI is used to improve the differential evolution algorithm to obtain the current optimal selection:

[0183]

[0184] in, is the current choice of the mth cow; The next choice for the mth cow; is the next hypothetical choice for the mth cow; f is the objective function.

[0185] Although the present invention has been disclosed as above in terms of preferred embodiments, they are not intended to limit the present invention. Anyone skilled in the art can make various changes or modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection defined by the claims of this application.

Claims

1. A new energy peak load dispatch optimization method based on source-load coordination, characterized by: include: Step 1: Obtain historical data of wind power, photovoltaic power and load in the grid area to be predicted, and build a wind, photovoltaic and load output prediction model based on the historical data; The prediction model is used to predict the corresponding prediction values ​​of the input wind power output, photovoltaic output and load demand in the next period; Step 2: Compare the wind and solar power forecast data with the load forecast data, and obtain the wind and solar power output and load values ​​that have the greatest similarity between the two based on the improved DTW, while satisfying the source-load fluctuation rate constraint; Step 3: Based on the similarity and constraint results of step 2, a mathematical model for DC wind-solar-thermal joint dispatch optimization is established with the lowest peak-shaving degree of the power grid and the lowest dispatch peak-shaving operation cost as the objective function; Step 4: Take the wind power and photovoltaic forecast values ​​and load demand forecast values ​​of the substation to be dispatched as input, and use the improved cow-grazing optimization algorithm to optimize and solve the DC wind-solar-fire joint dispatch optimization mathematical model, and optimize the global situation to minimize the peak-shaving degree of the power grid and the lowest peak-shaving operation cost.

2. According to claim 1, a new energy peak load dispatch optimization method based on source-load coordination is characterized in that: Step 1 specifically includes: S11: pre-processing the wind power output data, photovoltaic output data, and area load demand data in the acquired historical data; the pre-processing includes arranging the output data of each time period in the scheduling cycle according to the order of time; S12: taking the data of the corresponding time period of each scheduling period as the actual value of the corresponding time period of the scheduling period, taking the average value of all data of the corresponding time period of the scheduling period before the scheduling period as the estimation of the actual value of the scheduling period, and calculating the error value of the corresponding output type data in all historical scheduling periods; S13: Add the actual value estimate of the corresponding time period of the scheduling cycle and the corresponding error value to obtain the predicted value of the corresponding time period.

3. According to claim 1, a new energy peak load dispatch optimization method based on source-load coordination is characterized in that: Step 2 specifically includes: S21: A scheduling cycle is divided into T time periods on average, and the total output forecast value of renewable energy in the nth time period is calculated. And the load power demand forecast value in the nth period As two time series of DTW, the total output forecast value is the sum of the wind power output and photovoltaic output in the corresponding time period; as shown in the following formula, the total output forecast value of new energy In the above formula, It represents the predicted value of wind power output in the nth period; Represents the predicted value of photovoltaic output in the nth period; The total output forecast value of renewable energy and the load power demand forecast value within a dispatching cycle are arranged into two time series as follows: In the above formula, F N,F Represents the forecast time series data of the total output of renewable energy in a dispatching cycle; F Represents the forecast time series data of load power demand in a scheduling cycle; S22: Construct the distance matrix D between the forecast time series data of the total output of new energy and the forecast time series data of the load power demand. The element values ​​in the matrix are defined as follows: In the above formula, the matrix D is a T×T matrix, D(x, y) is the element in the xth row and yth column of the matrix D; and S23: Construct the cumulative cost matrix C of the distance matrix D using the dynamic balance method. The element values ​​in the matrix C are as follows: In the above formula, in the matrix C, define C(0,0)=0; C(x,0)=C(0,y)=+∞; Substitute x and y into formula (4) to recursively obtain the value of each element in the matrix C. Use formula (5) to select the optimal solution. Based on the optimal solution, find an optimal curved path P from point D(1,1) to D(T,T) in the D matrix: In the above formula, represents the path of the distance matrix D; P represents the optimal curved path, D(p s ) is the element in the distance matrix D corresponding to the optimal curved path P; The element corresponding to the cost matrix C when obtaining the optimal curved path P is the minimum value of the forecast error between the total output of new energy and the load power demand when the optimal curved path is used; S24: Calculate the time series F of the total output forecast value of new energy N,F and the time series of load power demand forecast value l F n The covariance and variance of : In the above formula, V is the predicted value covariance between the new energy output and load demand in the nth period of a dispatch cycle; Var,N is the variance of the predicted value of the total output of renewable energy in this dispatch period, V Var,l is the variance of the load demand forecast value of the scheduling period; is the average output forecast value of the wind farm in the dispatching period; is the average output forecast value of the photovoltaic power station in this scheduling period; is the average predicted value of load demand within the dispatching period; S25: Substitute the above covariance of power between renewable energy and load and the variance of power between renewable energy and load into the following correlation coefficient equation: The correlation coefficient between the predicted new energy output and load demand in the nth period is obtained: S26: The absolute value of the difference between the predicted values ​​of the total output of renewable energy in the nth period and the n+1th period minus the predicted value of the load in the nth period and the n+1th period is added together, and then divided by the average predicted output value of renewable energy in a scheduling cycle to obtain the source-load fluctuation rate δ: S27: Construct constraints based on the obtained correlation coefficient and source-load volatility, as shown in the following formula: R N-l,min ≤R N-l ≤R N-l,max d N-l,min ≤δ N-l ≤δ N-l,max (8) In the above formula, R N-l,min Represents the minimum value of the correlation coefficient between the predicted value of new energy output and the predicted value of load; R N-l Represents the correlation coefficient between the predicted value of new energy output and the predicted value of load, with a value range of [-1,1]; δ N-1 Indicates the source-load fluctuation rate between renewable energy output and load; judge Does the optimal solution satisfy the correlation coefficient R of the above formula? N-l and source-charge volatility δ N-1 If the constraint is satisfied, continue with the following steps; if not, take x-1 and based on formula (5), take The second optimal curved route P' is obtained, and steps S25 and S26 are repeated, and so on until the constraints of the correlation coefficient and the source load fluctuation rate of formula (8) are satisfied.

4. The method for optimizing the peak load regulation of renewable energy based on source-load coordination according to claim 1 is characterized in that: Step three specifically includes: S31: Obtain the transaction volume of the DC channel in the dispatching cycle and the total output forecast value of the wind and solar power in the cycle, and determine the supporting thermal power installed capacity according to the power balance relationship of the following formula: In the above formula, are the thermal power output and the estimated total output of the DC channel dispatch period T respectively; are the output forecast values ​​of wind power and photovoltaic power in the nth period of the scheduling cycle respectively; S32: Taking the lowest peak load regulation degree of the power grid and the lowest peak load regulation operation cost as the objective function, a DC wind-solar-thermal joint dispatch mathematical model is established, as shown in the following equations (10) and (11): minF=f f +f w +f pv +f l +f p (11) In the above formula, A is the peak load regulation degree of the receiving power grid; T is the total time period of a dispatching cycle; x1, x2 and x3 are the weight coefficients of the three influencing factors preset in function A; α is the preset proportional coefficient, which is usually 1000; F is the system peak load regulation operation cost; f f 、f w 、f pv are the output costs of thermal power, wind power and photovoltaic power respectively; f l Operating costs caused by load changes; p It is the penalty value for the amount of new energy not consumed; S33: Peak load regulation operation cost of thermal power units; operation cost includes thermal power cost f f , power generation cost g , start-stop cost s and the response cost f r , as follows: In the above formula, N G is the total amount of thermal power units; P Gi,t , P Gi,t-1 are the output values ​​of the i-th generator group participating in the peak load regulation of the power grid in period t and period t-1 respectively; P Gi,T,min , P Gi,T , P Gi,T,max are the minimum output, output value, and maximum output of the i-th thermal power unit respectively; is the start / stop status of the i-th generator set, 0 means it is stopped, and 1 means it is put into operation; a i 、b i 、c i is the consumption characteristic coefficient of the i-th thermal power unit; is the startup cost of the i-th generator set; is the downtime cost of the i-th generator set; is the cost coefficient for increasing the output of the i-th thermal power unit; is the cost coefficient for reducing the output of the i-th thermal power unit; S34: Wind power operation cost, which is the sum of the power generation costs of each wind farm in a dispatching cycle: Among them, P w is the cost per unit of wind power output; Nw represents the total number of wind farms; Pw represents the cost per unit of wind power output; represents the actual output value of the i-th wind farm; S35: Photovoltaic operating costs pv , is the total power generation cost of each photovoltaic power station in a scheduling cycle: Among them, P pv is the cost of unit photovoltaic output; N pv is the total number of photovoltaic power stations; is the actual output value of the i-th photovoltaic power station; S36: According to different requirements in different regions, penalty costs will be imposed on unabsorbed wind and solar power. p : Among them, μ1 and μ2 are the penalty coefficients for wind and solar power abandonment, and the values ​​are preset according to different requirements of different regions; S37: Calculate the operating cost when the load changes, which is the operating cost of the equivalent adjustable load f l : In the above formula, P l is the unit operating cost of equivalent adjustable load; Indicates the actual demand of load; S38: Add the thermal power output and the renewable energy output to get the DC channel transmission power 5. The method for optimizing the peak load scheduling of new energy based on source-load coordination according to claim 4 is characterized in that: Step 3 also includes the DC channel ramp rate constraint, as shown below: in, They are the lower and upper limits of the DC channel ramp rate respectively.

6. The method for optimizing the peak load scheduling of new energy based on source-load coordination according to claim 5 is characterized in that: Step 4 specifically includes: S41: Acquire historical data of the substation to be dispatched, and obtain output forecast data of the substation based on step 1; the output forecast data includes wind power output, photovoltaic output and load demand forecast time series data; S42: Preset thermal power unit parameters, the number of iterations K of the cow grazing algorithm, the number of iterations J of the improved differential algorithm, and the total number of populations L; S43: Simulate the behavior of cattle eating grass, randomly select more delicious grass from the current multiple grasslands with probability for iterative update, and obtain the global optimal solution that meets the iteration conditions, that is, the optimal solution of the DC wind-solar-fire joint scheduling optimization mathematical model with the lowest peak-shaving degree of the power grid and the lowest peak-shaving operation cost.

Images