Multi-node demand response method and system considering random volatility of renewable energy

Abstract

The invention discloses a multi-node demand response method and system considering the random volatility of renewable energy sources, and relates to the technical field of demand response, and the method comprises the steps: receiving the output data of the renewable energy sources, and carrying out the aggregation calculation of the output data of the renewable energy sources through a dynamic time bending algorithm and a K-means clustering algorithm, and obtaining an output curve; inputting the renewable energy output data into a pre-established multi-node demand response model, and outputting to obtain a CDL curve and a contribution factor corresponding to an individual load node; and performing similarity calculation according to the CDL curve corresponding to the individual load node and the contribution factor and output curve, and performing user incentive subsidy according to the similarity so as to realize user response effect evaluation.

Description

Technical Field

[0001] The present invention relates to the technical field of demand response, and in particular to a multi-node demand response method and system that considers random volatility of renewable energy. Background Art

[0002] With the continuous advancement of renewable energy technology, especially in the field of photovoltaics and solar energy, the proportion of renewable energy in the power system continues to rise. However, due to the interference of uncertain factors such as seasons and weather, the output of renewable energy shows a high degree of volatility and randomness, which makes it difficult to achieve effective control and regulation. At present, relying on traditional thermal power generation units can no longer meet the needs of real-time regulation of the power system.

[0003] In order to solve this problem, researchers have proposed a variety of strategies to improve the absorption capacity of renewable energy. Among them, safety-constrained unit commitment (SCUC) is used as an effective means to achieve large-scale grid connection of renewable energy. Some studies have improved the economy and flexibility of the system by optimizing the monthly and short-term scheduling of units. Other studies have proposed a medium- and long-term rapid unit commitment method based on the load state transfer curve to adapt to the time characteristics of renewable energy. In addition, there are also studies that improve the scheduling efficiency of wind power by establishing wind power equivalent units and medium- and short-term nested unit commitment models.

[0004] Demand response (DR), as an emerging strategy, provides new possibilities for the consumption of renewable energy. DR effectively reduces the peak load of the power grid by shifting loads during off-peak hours, and provides peak regulation and frequency regulation services for the power grid by integrating flexible load resources. Some studies have guided users to optimize load curves through time-of-day electricity prices and interruptible loads, thereby reducing the peak-shaving pressure of wind power on the system. Other studies have minimized the curtailment of wind power by combining DR and energy storage technology. In addition, the application of electricity-gas demand response to distribution networks improves the scheduling flexibility and controllability of the system, thereby more effectively absorbing renewable energy.

[0005] However, existing DR research is mostly focused on small-scale application scenarios. As the proportion of renewable energy continues to increase, the power system faces more regulation needs and needs to carry out larger-scale demand response. Demand response for large-scale users requires a large amount of communication and control resources. At the same time, coordination with a large number of load resources also makes it difficult to achieve global optimization at the large-scale power grid level.

[0006] To solve this problem, some studies have proposed the concept of an ideal load curve (Customer Direct Load, CDL), which can effectively smooth the fluctuations generated by renewable energy. By setting CDL as the load curve adjustment target for all users participating in demand response, users can make independent response decisions after weighing the cost of changing electricity behavior and the incentives obtained. Based on this strategy, the demand response center only needs to broadcast CDL to the entire power grid one day in advance without collecting and calculating a large amount of user load data.

[0007] Nevertheless, existing CDL studies apply the same CDL to each node of the system, ignoring the contribution ratio of different load nodes to renewable energy consumption, which makes it difficult to achieve accurate consumption of renewable energy in large-scale power grids. In large-scale power grids, different load nodes consume different amounts of energy supplied by renewable energy units. Due to the limitations of the power system network connection method and the constraints of power flow, there are differences in the ability of load nodes to consume renewable energy in the time dimension and space dimension. Therefore, it is necessary to formulate CDLs that match different loads according to their characteristics. Consider the dynamic contribution ratio of generators to loads. Summary of the invention

[0008] In order to address the deficiencies mentioned in the above background technology, the purpose of the present invention is to provide a multi-node demand response method and system that takes into account the random volatility of renewable energy, which can optimize the demand response strategy, improve the power grid's ability to absorb renewable energy and the economy and reliability of the system.

[0009] In a first aspect, the purpose of the present invention can be achieved by the following technical solution: a multi-node demand response method considering the random volatility of renewable energy, the method comprising the following steps:

[0010] Receiving renewable energy output data, using a dynamic time warping algorithm and a K-means clustering algorithm to aggregate and calculate the renewable energy output data, and obtaining an output curve;

[0011] Input the renewable energy output data into the pre-established multi-node demand response model, and output the corresponding CDL curve and contribution factor of the individual load node;

[0012] The similarity between the CDL curve and contribution factor of the individual load nodes and the output curve is calculated, and user incentive subsidies are provided based on the similarity, thereby realizing the evaluation of user response effects.

[0013] In combination with the first aspect, in some implementations of the first aspect, the method further includes: a process of calculating the dynamic time warping algorithm:

[0014] Dynamic time bending algorithm: According to the renewable energy output data, the renewable energy generation is a time series. Let two time series X = {x1, x2, x3, ... x T} and Y={y1,y2,y3,...y T}, their cumulative distance is expressed as:

[0015]

[0016] The typical output curve aggregated from renewable energy output data is obtained based on the dynamic time bending algorithm.

[0017] In combination with the first aspect, in some implementations of the first aspect, the method further includes: a calculation process of the K-means clustering algorithm:

[0018] Use the K-means clustering algorithm to cluster the output scenarios. First, randomly select k renewable energy output curves as the initial reference line, set a state variable state, and set the initial value to false. As long as state is true, the cycle continues.

[0019] For each renewable energy output curve line i, perform the following operations: calculate the distance between curve i and each reference line j and save this distance in the memo; find the minimum value of all distances and assign curve i to the cluster C corresponding to the minimum distance j ;

[0020] If the cluster has changed, the state is set to true. For each cluster, the reference line is updated to the average value of all curves in the cluster. When the loop ends and there is no change, the updated reference line is returned as a typical scenario of renewable energy output. By calculating the distance between each curve and the reference line, and continuously iteratively updating the clusters and reference lines, a set of typical output curves representing the characteristics of the entire data set is finally obtained as the output curve.

[0021] In combination with the first aspect, in certain implementations of the first aspect, the method also includes: the pre-established multi-node demand response model includes upper and lower layer models, wherein the upper layer model is a joint optimization model of SCUC and CDL, and the lower layer model is a generator contribution calculation model of the load node.

[0022] In combination with the first aspect, in some implementations of the first aspect, the method further includes: the joint optimization model of the SCUC and the CDL minimizes the total cost consisting of the power generation cost of the generator set and the wind and solar power abandonment cost, and the objective function is as follows:

[0023]

[0024] Among them, P G,i (t) is the active output of the adjustable generator numbered i in the system during the time period; N G is the number of adjustable generators; P R (t) is the output of all renewable energy sources in the system, such as photovoltaic and wind power, at that time; P R,max (t) is the maximum output of the system at that time, which can be obtained through day-ahead prediction; a i 、b i 、c i is the cost coefficient of generator i; C R The cost of wind curtailment.

[0025] In combination with the first aspect, in certain implementations of the first aspect, the method also includes: the constraints of the joint optimization model of the SCUC and CDL include: power balance constraints, total power consumption constraints, flow constraints, rotating reserve constraints, power generation limitation constraints, minimum unit start-up time constraints and minimum downtime constraints.

[0026] In combination with the first aspect, in some implementations of the first aspect, the method further includes: a calculation process of a contribution factor corresponding to the separate load node is as follows:

[0027] P ijk =C ij *P jk

[0028]

[0029]

[0030] Where: C ij is the contribution of generator i to the power consumption of load node in common domain j, F jk is the power on the link between common domains j and k, F ijk is the power on the link between common domains j and k due to generator i, I k is the sum of the injected powers in the common domain k, C ik is the contribution of generator i to the power consumed by the load node in the common domain k.

[0031] In combination with the first aspect, in some implementations of the first aspect, the method further includes: the process of the user incentive subsidy is as follows:

[0032] Normalize the CDL curve:

[0033]

[0034] Where: l(t) is the actual user load curve, α is the total daily power consumption;

[0035] The similarity E is defined to measure the similarity between the shape of the actual user load curve and the shape of the CDL curve, and is calculated by the following formula:

[0036]

[0037] Where: l * (t) is the value of the load curve to be measured after normalization; d is the Euclidean distance between the two series; E is the measured value after the Euclidean distance is converted to the (0,1] interval;

[0038] User incentive subsidy plan:

[0039] F(i)=(5×P mid +7×P l ow+10×P high )×γ(i)

[0040] Where: F(i) is the amount of incentive subsidy received by user i; γ(i) is the price coefficient calculated based on E of user i; P low , P mid and P high They are the adjusted power of users during the demand response flat period, demand response low period and demand response peak period respectively.

[0041] In a second aspect, in order to achieve the above-mentioned object, the present invention discloses a multi-node demand response system considering the random volatility of renewable energy, comprising:

[0042] An aggregation calculation module is used to receive renewable energy output data, and use a dynamic time warping algorithm and a K-means clustering algorithm to perform aggregation calculation on the renewable energy output data to obtain an output curve;

[0043] The curve output module is used to input the renewable energy output data into the pre-established multi-node demand response model, and output the corresponding CDL curve and contribution factor of the individual load node;

[0044] The response evaluation module is used to calculate the similarity between the CDL curve and contribution factor corresponding to the individual load node and the output curve, and to provide user incentive subsidies based on the similarity, thereby realizing the user response effect evaluation.

[0045] In conjunction with the second aspect, in some implementations of the second aspect, the system further includes: a process of dynamic time warping algorithm calculation in the aggregation calculation module:

[0046] Dynamic time bending algorithm: According to the renewable energy output data, the renewable energy generation is a time series. Let two time series X = {x1, x2, x3, ... x T} and Y={y1,y2,y3,...y T}, their cumulative distance is expressed as:

[0047]

[0048] The typical output curve of renewable energy output data is obtained by dynamic time bending algorithm

[0049] Among them, the calculation process of the K-means clustering algorithm in the aggregation calculation module is:

[0050] Use the K-means clustering algorithm to cluster the output scenarios. First, randomly select k renewable energy output curves as the initial reference line, set a state variable state, and set the initial value to false. As long as state is true, the cycle continues.

[0051] For each renewable energy output curve line i, perform the following operations: calculate the distance between curve i and each reference line j and save this distance in the memo; find the minimum value of all distances and assign curve i to the cluster C corresponding to the minimum distance j ;

[0052] If the cluster has changed, the state is set to true. For each cluster, the reference line is updated to the average value of all curves in the cluster. When the loop ends and there is no change, the updated reference line is returned as the typical scenario of renewable energy output. By calculating the distance between each curve and the reference line, and continuously iteratively updating the clusters and reference lines, a set of typical output curves representing the characteristics of the entire data set is finally obtained as the output curve;

[0053] The multi-node demand response model pre-established in the curve output module includes upper and lower layer models, wherein the upper layer model is the joint optimization model of SCUC and CDL, and the lower layer model is the generator contribution calculation model of the load node;

[0054] The joint optimization model of SCUC and CDL in the curve output module minimizes the total cost consisting of the power generation cost of the generator set and the wind and solar power abandonment cost. The objective function is as follows:

[0055]

[0056] Among them, P G,i (t) is the active output of the adjustable generator numbered i in the system during the time period; N G is the number of adjustable generators; P R (t) is the output of all renewable energy sources in the system, such as photovoltaic and wind power, at that time; P R,max(t) is the maximum output of the system at that time, which can be obtained through day-ahead prediction; a i 、b i 、c i is the cost coefficient of generator i; C R The cost of wind curtailment;

[0057] The constraints of the joint optimization model of SCUC and CDL in the curve output module include: power balance constraint, total power consumption constraint, power flow constraint, spinning reserve constraint, power generation limit constraint, minimum unit start-up time constraint and minimum downtime constraint;

[0058] The calculation process of the corresponding contribution factor of a single load node in the curve output module is as follows:

[0059] P ijk =C ij *P jk

[0060]

[0061] Where: C ij is the contribution of generator i to the power consumption of load node in common domain j, F jk is the power on the link between common domains j and k, F ijk is the power on the link between common domains j and k due to generator i, I k is the sum of the injected powers in the common domain k, C ik is the contribution of generator i to the power consumed by the load node in the common domain k;

[0062] The process of user incentive subsidies within the response evaluation module is as follows:

[0063] Normalize the CDL curve:

[0064]

[0065] Where: l(t) is the actual user load curve, α is the total daily power consumption;

[0066] The similarity E is defined to measure the similarity between the shape of the actual user load curve and the shape of the CDL curve, and is calculated by the following formula:

[0067]

[0068] Where: l * (t) is the value of the load curve to be measured after normalization; d is the Euclidean distance between the two series; E is the measured value after the Euclidean distance is converted to the (0,1] interval;

[0069] User incentive subsidy plan:

[0070] F(i)=(5×P mid +7×P low +10×P high )×γ(i)

[0071] Where: F(i) is the amount of incentive subsidy received by user i; γ(i) is the price coefficient calculated based on E of user i; P low , P mid and P high They are the adjusted power of users during the demand response flat period, demand response low period and demand response peak period respectively.

[0072] Beneficial effects of the present invention:

[0073] The present invention: (1) reduces the communication and control resource requirements during the implementation of large-scale demand response, and guides users by broadcasting CDL without collecting and calculating a large amount of user load data; (2) The demand response mechanism guides users to adjust their electricity consumption behavior through incentives, reducing dependence on traditional generators, thereby reducing the operating cost of the power system; (3) Through the multi-node demand response method, renewable energy can be absorbed and utilized more effectively, reducing the phenomenon of wind and solar power abandonment caused by the volatility of renewable energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0075] Figure 1 It is a schematic flow chart of the method of the present invention;

[0076] Figure 2 This is a schematic diagram of the IEEE standard six-node test system used for testing the present invention;

[0077] Figure 3 A schematic diagram of a typical scenario new energy output curve calculated by the present invention;

[0078] Figure 4 A schematic diagram of the CDL of each load node calculated by the present invention;

[0079] Figure 5 A schematic diagram of the load curve before and after the node 3 implements the demand response calculated by the present invention;

[0080] Figure 6 It is a schematic diagram of the output curve of renewable energy that can be absorbed when there is no demand response calculated by the present invention;

[0081] Figure 7 A schematic diagram of the output curve of renewable energy that can be absorbed when there is demand response calculated by the present invention;

[0082] Figure 8 It is a schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0083] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0084] Embodiment 1:

[0085] The following is an introduction to the relevant terms involved in the embodiments of the present application:

[0086] Demand response: Demand response (DR) is the abbreviation of power demand response. It means that when the wholesale power market price rises or the system reliability is threatened, the power user receives the direct compensation notice of induced load reduction or the signal of power price increase from the power supplier, and changes its inherent habitual power consumption mode to reduce or shift the power load in a certain period of time to respond to power supply, thereby ensuring the stability of the power grid and suppressing the short-term behavior of rising electricity prices. It is one of the solutions for demand side management (DSM).

[0087] Dynamic Time Warping (DTW) is an algorithm used to measure the similarity between two variable-length sequences. It is particularly suitable for processing time series data, even if they have nonlinear changes in the time scale. The DTW algorithm can find local similarities between time series, even if these sequences are scaled differently on the time axis. This algorithm is often used in speech recognition, DNA analysis, gesture recognition, time series classification in machine learning and other problems.

[0088] K-means clustering algorithm: The k-means clustering algorithm is an iterative clustering analysis algorithm. Its steps are to pre-divide the data into K groups, then randomly select K objects as the initial cluster centers, and then calculate the distance between each object and each seed cluster center, and assign each object to the cluster center closest to it. The cluster centers and the objects assigned to them represent a cluster. Each time a sample is assigned, the cluster center of the cluster will be recalculated based on the existing objects in the cluster. This process will be repeated until a certain termination condition is met. The termination condition can be that no (or a minimum number) of objects are reassigned to different clusters, no (or a minimum number) of cluster centers change again, and the sum of squared errors is locally minimized.

[0089] like Figure 1 As shown, a multi-node demand response method considering random volatility of renewable energy includes the following steps:

[0090] S101: Receiving renewable energy output data, and performing aggregation calculation on the renewable energy output data using a dynamic time warping algorithm and a K-means clustering algorithm to obtain an output curve;

[0091] Using the Dynamic Time Warping (DTW) algorithm and the K-means clustering algorithm, a large amount of renewable energy output data is aggregated into several typical scenarios for analysis. The main implementation steps are as follows:

[0092] Since the renewable energy generation is a time series, this paper uses the dynamic time warping (DTW) algorithm to calculate the cumulative distance between the two series as the similarity of the two curves. Assume that the two time series X = {x1, x2, x3, ... x T} and Y={y1,y2,y3,...y T}, their cumulative distance can be expressed as:

[0093]

[0094] According to this method, typical output curves can be aggregated from the historical data of renewable energy. Then, the K-means clustering algorithm is used to aggregate the output scenarios. First, k renewable energy output curves are randomly selected, which will be used as the initial reference lines. A state variable state is set, and the initial value is set to false. As long as state is true, the loop continues.

[0095] For each renewable energy output curve (denoted as curve i), the following operations are performed: ① Calculate the distance between curve i and each reference line (denoted as reference line j), and save this distance in the memo; ② Find the minimum value of all distances and assign curve i to the cluster C corresponding to the minimum distance j .

[0096] If the cluster has changed (that is, at least one curve has been reassigned to a different cluster), set state to true. For each cluster, update the reference line to the average of all curves in the cluster. When the loop ends and there are no more changes, return the updated reference line as a typical scenario for renewable energy output. The purpose of this algorithm is to simplify a large amount of renewable energy output data (such as wind power and solar power output data) into several representative typical scenarios through cluster analysis. Doing so can reduce the amount of data processed while retaining the main features of the data, facilitating subsequent analysis and model building. By calculating the distance between each curve and the reference line, and continuously iteratively updating the clusters and reference lines, a set of typical output curves that can represent the characteristics of the entire data set can eventually be obtained.

[0097] S102: inputting renewable energy output data into a pre-established multi-node demand response model, and outputting a CDL curve and contribution factor corresponding to a single load node;

[0098] The pre-established multi-node demand response model includes upper and lower layer models, where the upper layer model is a joint optimization model of SCUC and CDL, and the lower layer model is a generator contribution calculation model of the load node;

[0099] The goal of the SCUC and CDL joint optimization model is to minimize the total cost consisting of the power generation cost of the generator set and the wind and solar power abandonment cost while absorbing as much renewable energy output as possible. The objective function is expressed as:

[0100]

[0101] Among them, P G,i (t) is the active output of the adjustable generator numbered i in the system during the time period; N G is the number of adjustable generators; P R (t) is the output of all renewable energy sources in the system, such as photovoltaic and wind power, at that time; P R,max (t) is the maximum output of the system at that time, which can be obtained through day-ahead prediction; a i , b i 、c i is the cost coefficient of generator i; C R The cost of wind curtailment.

[0102] The constraints of the collaborative optimization model of SCUC and CDL are as follows:

[0103] Power balance constraints:

[0104]

[0105] The above equation is the system power balance constraint, which is different from the power balance of traditional demand response schemes. This equation shows the balance between adjustable resources and non-adjustable resources. In other words, the power balance of the entire system is maintained by adjusting the adjustable load and generator output. Among them, P R,j (t) is the effective output of the generator of the new energy source numbered j in the system during this time period, C i,d It is the ratio of the output of the conventional generator numbered i in the system transmitted to the load node d. is the total power consumption of the flexible load participating in demand response at the load node d during the time period. is the CDL load value at load node d. is the amount of power consumed by the fixed load at load node d during the time period.

[0106] Total power consumption constraints:

[0107]

[0108] The above formula is the total power consumption constraint of users. The demand response based on CDL emphasizes changing the shape of the original load curve rather than increasing or decreasing power consumption to ensure the efficiency of user power consumption. Therefore, this constraint needs to be considered. is the per-unit value of the user's CDL in period t.

[0109] Power flow constraints:

[0110]

[0111] Among them, P inj is the node injection power matrix; K P , K R , K D They are respectively the conventional generator connection matrix, the new energy generator connection matrix and the load connection matrix; P G , P R and P D are the conventional generator power matrix, the new energy generator power matrix and the load matrix respectively; B is the line conduction matrix; K L is the line connection matrix; X b Stores the conductance of each line; θ is the power angle matrix.

[0112] Spin reserve constraints:

[0113]

[0114] in, is the start and stop status of the conventional generator numbered i in the system during period t; is the maximum output of the conventional generator numbered i in the system; D t is the total power demand of the system in period t; R t represents the spinning reserve at time t.

[0115] Generation restrictions:

[0116]

[0117] in, is the minimum output of the conventional generator numbered i in the system; is the actual output of the conventional generator numbered i in the system during period t.

[0118] Minimum unit start time constraint:

[0119]

[0120] Minimum downtime constraints:

[0121]

[0122] Among them, M UTi and M DTi They are the minimum start-up time and minimum shutdown time of the unit respectively; and is the duration that unit i remains in the on state during period t.

[0123] It is defined that all load nodes powered by a generator are the domain of the generator. Since the load nodes in different generator domains have a large overlap, in order to calculate the contribution coefficient of the generator to the load, the nodes must be further divided. Therefore, a group of continuous buses powered by exactly the same generator is defined as a common domain, and the number of generators supplying the common domain is defined as the level of the common domain.

[0124] Based on the definition of common domain, when the power flow direction of all lines in the system is known, all load nodes of the system will be divided into different common domains that do not overlap each other. There will be one or more connected branches in the topology of adjacent common domains, and the power flow direction on these branches must be the same. We define these branches as links. When the power flow direction in all branches of the network is given, the network can be divided into a unique set of common points and links, and the state of the network can be described as a state diagram, where the common domain is represented by nodes and the link is represented by branches.

[0125] The algorithm for bottom-up recursive calculation of generator contribution to load nodes can be constructed with the following equation:

[0126] P ijk =C ij *P jk

[0127]

[0128]

[0129] Where: C ij is the contribution of generator i to the power consumption of load node in common domain j, F jk is the power on the link between common domains j and k, F ijk is the power on the link between common domains j and k due to generator i. k is the sum of the injected powers in the common domain k, C ik is the contribution of generator i to the power consumed by the load node in the common domain k.

[0130] The established multi-node demand response model is divided into two layers. The upper layer is a multi-node CDL calculation model, which takes into account the differences in the new energy consumption capacity of different load nodes and calculates the CDL curve corresponding to each individual load node. The upper model first calculates the contribution rate of each generator to the load of each node, and then passes the flow and load of each node to the lower model. The lower layer is the generator contribution calculation model of the load node, which calculates and updates the contribution factor of each generator to each load node by tracing the generator output from the load node to its source. The entire model runs in an iterative manner between the upper and lower layers until convergence.

[0131] S103: Calculate the similarity between the CDL curve and contribution factor corresponding to the individual load nodes and the output curve, and provide user incentive subsidies based on the similarity, thereby achieving user response effect evaluation.

[0132] In order to ensure that the CDL curve released by the DR Center can effectively guide customers to change their electricity consumption behavior and promote the consumption of renewable energy, it is necessary to establish a corresponding customer performance evaluation and incentive subsidy mechanism. First, since the CDL curve is the ideal load curve shape of each user node, the user's load curve must be normalized to facilitate a unified standard evaluation of the user's electricity consumption behavior. The specific formula for normalization is as follows:

[0133]

[0134] Where: l(t) is the actual user load curve, and α is the total daily electricity consumption.

[0135] (2) Define the similarity E to measure the similarity between the shape of the actual user load curve and the shape of the CDL curve. This indicator reflects whether the actual user's electricity consumption behavior meets the expectations of the DR center and is calculated using the following formula:

[0136]

[0137] Among them: * (t) is the value of the load curve to be measured after normalization; d is the Euclidean distance between the two series; E is the measured value after the Euclidean distance is converted to the (0,1] interval. E takes a value between 0 and 1. The closer the value of E is to 1, the more actively the user adjusts his electricity consumption behavior to participate in demand response.

[0138] (3) Finally, on this basis, formulate a user incentive subsidy plan.

[0139] F(i)=(5×P mid +7×P low +10×P high )×γ(i)

[0140] Where: F(i) is the amount of incentive subsidy received by user i; γ(i) is the price coefficient calculated based on E of user i; P low , P mid and P high They are the adjusted power of users during the demand response flat period, demand response low period and demand response peak period respectively.

[0141] Embodiment 2: In the second aspect, as Figure 8 As shown, in order to achieve the above-mentioned purpose, the present invention discloses a multi-node demand response system considering the random volatility of renewable energy, including:

[0142] Aggregation calculation module 11, used for receiving renewable energy output data, performing aggregation calculation on the renewable energy output data using dynamic time warping algorithm and K-means clustering algorithm, and obtaining an output curve;

[0143] The curve output module 12 is used to input the renewable energy output data into a pre-established multi-node demand response model, and output the CDL curve and contribution factor corresponding to the individual load node;

[0144] The response evaluation module 13 is used to calculate the similarity between the CDL curve and contribution factor corresponding to the individual load node and the output curve, and to provide user incentive subsidies based on the similarity, thereby realizing user response effect evaluation.

[0145] In conjunction with the second aspect, in some implementations of the second aspect, the system further includes: a process of dynamic time warping algorithm calculation in the aggregation calculation module 11:

[0146] Dynamic time bending algorithm: According to the renewable energy output data, the renewable energy generation is a time series. Let two time series X = {x1, x2, x3, ... x T} and Y={y1,y2,y3,...y T}, their cumulative distance is expressed as:

[0147]

[0148] The typical output curve of renewable energy output data is obtained by dynamic time bending algorithm

[0149] Among them, the calculation process of the K-means clustering algorithm in the aggregation calculation module 11 is:

[0150] Use the K-means clustering algorithm to cluster the output scenarios. First, randomly select k renewable energy output curves as the initial reference line, set a state variable state, and set the initial value to false. As long as state is true, the cycle continues.

[0151] For each renewable energy output curve line i, perform the following operations: calculate the distance between curve i and each reference line j and save this distance in the memo; find the minimum value of all distances and assign curve i to the cluster C corresponding to the minimum distance j ;

[0152] If the cluster has changed, the state is set to true. For each cluster, the reference line is updated to the average value of all curves in the cluster. When the loop ends and there is no change, the updated reference line is returned as the typical scenario of renewable energy output. By calculating the distance between each curve and the reference line, and continuously iteratively updating the clusters and reference lines, a set of typical output curves representing the characteristics of the entire data set is finally obtained as the output curve;

[0153] The multi-node demand response model pre-established in the curve output module 12 includes an upper and lower layer model, wherein the upper layer model is a joint optimization model of SCUC and CDL, and the lower layer model is a generator contribution calculation model of the load node;

[0154] The joint optimization model of SCUC and CDL in the curve output module 12 minimizes the total cost consisting of the power generation cost of the generator set and the wind and solar power abandonment cost. The objective function is as follows:

[0155]

[0156] Among them, P G,i(t) is the active output of the adjustable generator numbered i in the system during the time period; N G is the number of adjustable generators; P R (t) is the output of all renewable energy sources in the system, such as photovoltaic and wind power, at that time; P R,max (t) is the maximum output of the system at that time, which can be obtained through day-ahead prediction; a i 、b i 、c i is the cost coefficient of generator i; C R The cost of wind curtailment;

[0157] The constraints of the joint optimization model of SCUC and CDL in the curve output module 12 include: power balance constraint, total power consumption constraint, power flow constraint, spinning reserve constraint, power generation limit constraint, minimum unit start-up time constraint and minimum shutdown time constraint;

[0158] The calculation process of the corresponding contribution factor of the individual load node in the curve output module 12 is as follows:

[0159] P ijk =C ij *P jk

[0160]

[0161] Where: C ij is the contribution of generator i to the power consumption of load node in common domain j, F jk is the power on the link between common domains j and k, F ijk is the power on the link between common domains j and k due to generator i, I k is the sum of the injected powers in the common domain k, C ik is the contribution of generator i to the power consumed by the load node in the common domain k;

[0162] The process of user incentive subsidy in response evaluation module 13 is as follows:

[0163] Normalize the CDL curve:

[0164]

[0165] Where: l(t) is the actual user load curve, α is the total daily power consumption;

[0166] The similarity E is defined to measure the similarity between the shape of the actual user load curve and the shape of the CDL curve, and is calculated by the following formula:

[0167]

[0168] Where: l * (t) is the value of the load curve to be measured after normalization; d is the Euclidean distance between the two series; E is the measured value after the Euclidean distance is converted to the (0,1] interval;

[0169] User incentive subsidy plan:

[0170] F(i)=(5×P mid +7×P low +10×P high )×γ(i)

[0171] Where: F(i) is the amount of incentive subsidy received by user i; γ(i) is the price coefficient calculated based on E of user i; P low , P mid and P high They are the adjusted power of users during the demand response flat period, demand response low period and demand response peak period respectively.

[0172] In order to verify the multi-node system demand response method proposed in the present invention, an IEEE standard six-node test system is used for testing. Figure 2 As shown. Figure 2 In the figure, G1, G2 and G3 are traditional adjustable generators, and M1 and M2 are renewable energy non-adjustable generators. In addition, the proportion of user loads that can participate in demand response in the total demand side load is defined as the demand response penetration rate. The demand response penetration rate is set to 50%.

[0173] The output of wind power and photovoltaic power generation is selected from the data released by PJM in 2018. The clustering algorithm can effectively obtain the distribution of renewable energy output, cluster it into 8 typical output scenarios, and calculate their probabilities. The results are as follows: Figure 3 The occurrence probability of each typical scenario is shown in Table 1. The user CDL of each node is calculated based on the controllable unit parameters, renewable energy generator unit parameters, flexible load and fixed load. The calculation results are shown in Figure 4 shown.

[0174] In order to intuitively express the effect of this method on renewable energy consumption, the amount of renewable energy abandoned under different curve similarities was calculated, and the ratio of the abandoned amount to the maximum output of the renewable energy unit was defined as the abandonment rate. Figure 6 is the output curve of renewable energy that can be absorbed when not participating in demand response, Figure 7 It is the output curve of renewable energy that can be absorbed when participating in demand response.

[0175] Assuming that the power generation cost of renewable energy units is ignored, the cost of the system will include two parts, the power generation cost of the controllable units and the adjustment cost paid to users, that is, the incentive subsidy obtained by users. The total load of users in a day remains unchanged, and the electricity price remains unchanged. The cost analysis of running this method is as follows, as shown in Table 2.

[0176] Table 1 shows the probability of occurrence of each typical scenario calculated by the present invention.

[0177]

[0178] Table 2 is the cost analysis of different similarities calculated by the present invention.

[0179]

[0180] Based on the same inventive concept, the present invention also provides a computer device, which includes: one or more processors, and a memory for storing one or more computer programs; the program includes program instructions, and the processor is used to execute the program instructions stored in the memory. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is used to implement one or more instructions, specifically used to load and execute one or more instructions in a computer storage medium to implement the above method.

[0181] It needs to be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium, on which a computer program is stored, and the computer program is executed by a processor to execute the above method. The storage medium can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electrical, magnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0182] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0183] The above shows and describes the basic principles, main features and advantages of the present disclosure. Those skilled in the art should understand that the present disclosure is not limited by the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present disclosure. Without departing from the spirit and scope of the present disclosure, the present disclosure may have various changes and improvements, and these changes and improvements fall within the scope of the present disclosure to be protected.

Claims

1. A multi-node demand response method considering the random volatility of renewable energy, characterized in that: The method comprises the following steps: Receiving renewable energy output data, using a dynamic time warping algorithm and a K-means clustering algorithm to aggregate and calculate the renewable energy output data, and obtaining an output curve; Input the renewable energy output data into the pre-established multi-node demand response model, and output the corresponding CDL curve and contribution factor of the individual load node; The similarity between the CDL curve and contribution factor of the individual load nodes and the output curve is calculated, and user incentive subsidies are provided based on the similarity, thereby realizing the evaluation of user response effects.

2. The multi-node demand response method considering the random volatility of renewable energy according to claim 1 is characterized in that: The dynamic time warping algorithm calculates the process: Dynamic time bending algorithm: According to the renewable energy output data, the renewable energy generation is a time series. Let two time series X = {x1, x2, x3, ... x T } and Y={y1,y2,y3,...y T }, their cumulative distance is expressed as: The typical output curve aggregated from renewable energy output data is obtained based on the dynamic time bending algorithm.

3. The multi-node demand response method considering the random volatility of renewable energy according to claim 1 is characterized in that: The calculation process of the K-means clustering algorithm: Use the K-means clustering algorithm to cluster the output scenarios. First, randomly select k renewable energy output curves as the initial reference line, set a state variable state, and set the initial value to false. As long as state is true, the cycle continues. For each renewable energy output curve i, perform the following operations: calculate the distance between curve i and each reference line j and save this distance in the memo; find the minimum value of all distances and assign curve i to the cluster C corresponding to the minimum distance j ; If the cluster has changed, the state is set to true. For each cluster, the reference line is updated to the average value of all curves in the cluster. When the loop ends and there is no change, the updated reference line is returned as a typical scenario of renewable energy output. By calculating the distance between each curve and the reference line, and continuously iteratively updating the clusters and reference lines, a set of typical output curves representing the characteristics of the entire data set is finally obtained as the output curve.

4. The multi-node demand response method considering the random volatility of renewable energy according to claim 1 is characterized in that: The pre-established multi-node demand response model includes upper and lower layer models, wherein the upper layer model is a joint optimization model of SCUC and CDL, and the lower layer model is a generator contribution calculation model of the load node.

5. The multi-node demand response method considering the random volatility of renewable energy according to claim 4 is characterized in that: The joint optimization model of SCUC and CDL minimizes the total cost consisting of the power generation cost of the generator set and the wind and solar power abandonment cost. The objective function is as follows: Among them, P G,i (t) is the active output of the adjustable generator numbered i in the system during the time period; N G is the number of adjustable generators; P R (t) is the output of all renewable energy sources in the system, such as photovoltaic and wind power, at that time; P R,max (t) is the maximum output of the system at that time, obtained by day-ahead prediction; a i 、b i 、c i is the cost coefficient of generator i; C R The cost of wind curtailment.

6. The multi-node demand response method considering the random volatility of renewable energy according to claim 5 is characterized in that: The constraints of the joint optimization model of SCUC and CDL include: power balance constraint, total power consumption constraint, power flow constraint, spinning reserve constraint, power generation limit constraint, minimum unit start-up time constraint and minimum downtime constraint.

7. The multi-node demand response method considering the random volatility of renewable energy according to claim 1 is characterized in that: The calculation process of the corresponding contribution factor of the individual load node is as follows: P ijk =C ij *P jk Where: C ij is the contribution of generator i to the power consumption of load node in common domain j, F jk is the power on the link between common domains j and k, F ijk is the power on the link between common domains j and k due to generator i, I k is the sum of the injected powers in the common domain k, C ik is the contribution of generator i to the power consumed by the load node in the common domain k.

8. The multi-node demand response method considering the random volatility of renewable energy according to claim 1, characterized in that: The process of user incentive subsidy is as follows: Normalize the CDL curve: Where: l(t) is the actual user load curve, α is the total daily power consumption; The similarity E is defined to measure the similarity between the shape of the actual user load curve and the shape of the CDL curve, and is calculated by the following formula: Where: l * (t) is the value of the load curve to be measured after normalization; d is the Euclidean distance between the two series; E is the measured value after the Euclidean distance is converted to the (0,1] interval; User incentive subsidy plan: F(i)=(5×P mid +7×P low +10×P high )×γ(i) Where: F(i) is the amount of incentive subsidy received by user i; γ(i) is the price coefficient calculated based on E of user i; P low , P mid and P high They are the adjusted power of users during the demand response flat period, demand response low period and demand response peak period respectively.

9. A multi-node demand response system considering random volatility of renewable energy, characterized in that: include: An aggregation calculation module is used to receive renewable energy output data, and use a dynamic time warping algorithm and a K-means clustering algorithm to perform aggregation calculation on the renewable energy output data to obtain an output curve; The curve output module is used to input the renewable energy output data into the pre-established multi-node demand response model, and output the corresponding CDL curve and contribution factor of the individual load node; The response evaluation module is used to calculate the similarity between the CDL curve and contribution factor corresponding to the individual load node and the output curve, and to provide user incentive subsidies based on the similarity, thereby realizing the user response effect evaluation.

10. The multi-node demand response system considering the random volatility of renewable energy according to claim 9, characterized in that: The process of dynamic time warping algorithm calculation in the aggregation calculation module: Dynamic time bending algorithm: According to the renewable energy output data, the renewable energy generation is a time series. Let two time series X = {x1, x2, x3, ... x T } and Y={y1,y2,y3,...y T }, their cumulative distance is expressed as: The typical output curve of renewable energy output data is obtained by dynamic time bending algorithm Among them, the calculation process of the K-means clustering algorithm in the aggregation calculation module is: Use the K-means clustering algorithm to cluster the output scenarios. First, randomly select k renewable energy output curves as the initial reference line, set a state variable state, and set the initial value to false. As long as state is true, the cycle continues. For each renewable energy output curve i, perform the following operations: calculate the distance between curve i and each reference line j and save this distance in the memo; find the minimum value of all distances and assign curve i to the cluster C corresponding to the minimum distance j ; If the cluster has changed, the state is set to true. For each cluster, the reference line is updated to the average value of all curves in the cluster. When the loop ends and there is no change, the updated reference line is returned as the typical scenario of renewable energy output. By calculating the distance between each curve and the reference line, and continuously iteratively updating the cluster and the reference line, a set of typical output curves representing the characteristics of the entire data set is finally obtained as the output curve; The multi-node demand response model pre-established in the curve output module includes upper and lower layer models, wherein the upper layer model is the joint optimization model of SCUC and CDL, and the lower layer model is the generator contribution calculation model of the load node; The joint optimization model of SCUC and CDL in the curve output module minimizes the total cost consisting of the power generation cost of the generator set and the wind and solar power abandonment cost. The objective function is as follows: Among them, P G,i (t) is the active output of the adjustable generator numbered i in the system during the time period; N G is the number of adjustable generators; P R (t) is the output of all renewable energy sources in the system, such as photovoltaic and wind power, at that time; P R,max (t) is the maximum output of the system at that time, which can be obtained through day-ahead prediction; a i 、b i 、c i is the cost coefficient of generator i; C R The cost of wind curtailment; The constraints of the joint optimization model of SCUC and CDL in the curve output module include: power balance constraint, total power consumption constraint, power flow constraint, spinning reserve constraint, power generation limit constraint, minimum unit start-up time constraint and minimum downtime constraint; The calculation process of the corresponding contribution factor of a single load node in the curve output module is as follows: P ijk =C ij *P jk Where: C ij is the contribution of generator i to the power consumption of load node in common domain j, F jk is the power on the link between common domains j and k, F ijk is the power on the link between common domains j and k due to generator i, I k is the sum of the injected powers in the common domain k, C ik is the contribution of generator i to the power consumed by the load node in the common domain k; The process of user incentive subsidies within the response evaluation module is as follows: Normalize the CDL curve: Where: l(t) is the actual user load curve, α is the total daily power consumption; The similarity E is defined to measure the similarity between the shape of the actual user load curve and the shape of the CDL curve, and is calculated by the following formula: Where: l * (t) is the value of the load curve to be measured after normalization; d is the Euclidean distance between the two series; E is the measured value after the Euclidean distance is converted to the (0,1] interval; User incentive subsidy plan: F(i)=(5×P mid +7×P low +10×P high )×γ(i) Where: F(i) is the amount of incentive subsidy received by user i; γ(i) is the price coefficient calculated based on E of user i; P low , P mid and P high They are the adjusted power of users during the demand response flat period, demand response low period and demand response peak period respectively.

Images