A Method for Evaluating the Regulation Potential of a Virtual Power Plant Driven by Dual Model Data

Abstract

The present invention discloses a method for evaluating the regulation potential of a virtual power plant driven by model data in a dual - drive manner. First, an energy consumption index is proposed for diverse power generation resources and the dynamic operating characteristics of equivalent power generation resources, and the improved Euclidean distance minimum clustering method is used for multi - heterogeneous load operation data to extract typical operating characteristics of the same type. Then, a physical operation model of virtual power plant resources based on multi - dimensional data - driven is established, and joint pre - dispatch considering traditional generating units and virtual power plants is carried out. Finally, a calculation method for the multi - scenario prediction response potential of virtual power plant resources based on data and model is established based on the pre - dispatch results. The present invention adopts a method combining physics and data, which not only reduces the difficulty of exponentially solving massive data but also improves the interpretability of the operation model, effectively realizes the accurate evaluation of the schedulable potential of virtual power plant resources under multiple operating scenarios, and improves the availability of virtual power plant resources.

Description

Technical Field

[0001] The present invention relates to a method for evaluating the regulation potential of a virtual power plant driven by model data, specifically to a technology for quantitatively evaluating the demand response potential of multi-load resources by means of a physical dynamic model and a data mining method, which can accurately evaluate the response ability of virtual power plant resources. Background Art

[0002] New energy often has strong intermittent and random characteristics. For example, wind power generation and solar power generation are affected by climate factors and have the characteristics of intermittency and volatility, making it difficult to accurately predict and control. This is one of the main bottlenecks restricting the efficient utilization of distributed renewable energy power sources. High-penetration renewable energy power sources can cause imbalance between supply and demand in the real-time operation of the system, which will have an adverse impact on the formulation of power generation plans, real-time scheduling, and reserve arrangements. If the grid operation mode cannot be reasonably arranged and effective integration cannot be achieved, unnecessary curtailment of wind and light will occur, affecting the economy of resource aggregators, and even seriously affecting the safe and stable operation of the power system.

[0003] To solve this problem, the concept of virtual power plant emerged as the times require. As a new model of integrated energy service business, the virtual power plant can aggregate flexible resources such as distributed generation, energy storage, electric vehicles, and controllable loads, effectively stimulate various flexible resources to participate in the power market, and can help aggregators effectively integrate various forms and characteristics of power sources and electricity loads in the region without restructuring the power grid, improve the consumption level of local flexible resources, and increase the economic benefits of regional virtual power plant operators. Through intelligent operation, it can reduce the energy consumption cost, promote the consumption of new energy, provide services for the operation of the power market, provide a new profit model for resource aggregators, help achieve "carbon peak and carbon neutrality", and has broad development prospects in China. Based on this, the present invention proposes a method for evaluating the potential of a virtual power plant driven by physical data to effectively improve the accuracy of virtual power plant resource scheduling. Summary of the Invention

[0004] To solve the problems and requirements in the background art, the present invention proposes a method for evaluating the regulation potential of a virtual power plant driven by model data. First, an energy consumption index for multi-source power generation resources and the dynamic operation characteristics of equivalent power generation resources is proposed, and the multi-source heterogeneous load operation data is clustered with the improved Euclidean distance minimum to extract the same type of typical operation characteristics. Then, a physical operation model of virtual power plant resources based on multi-dimensional data driving is established, and the joint pre-scheduling considering traditional generating units and virtual power plants is carried out. Finally, a calculation method for the multi-scenario estimated response potential of virtual power plant resources based on data and models is established based on the pre-scheduling results.

[0005] The technical solution adopted by the present invention to solve the above problems is as follows: A method for evaluating the regulation potential of a virtual power plant driven by dual model data, characterized by including the following steps:

[0006] Step 1: Calculate the energy consumption indicators for the dynamic operation characteristics of diverse power generation resources and equivalent power generation resources. For power generation resources and equivalent power generation resources, establish evaluation indicators covering their adjustment speed, adjustment boundary, and adjustment duration;

[0007] Step 2: Optimize the clustering of multi - heterogeneous virtual power plant data with the minimum of the improved Euclidean distance. Use the improved principal component analysis method to reduce the dimension of the virtual power plant resource data. All virtual power plant data are aggregated into k typical types and represented by the center point values;

[0008] Step 3: Establish a physical operation model of the virtual power plant's equivalent power generation resources driven by multi - dimensional data, and obtain k physical general models of the virtual power plant to represent the operating power characteristics of the virtual power plant resources;

[0009] Step 4: Establish a physical general model of the virtual power plant and obtain the physical general model of the virtual power plant to represent the operating power characteristics of the virtual power plant resources;

[0010] Step 5: Establish a joint pre - dispatch considering traditional generator sets and virtual power plants, and achieve the lowest total dispatch cost when comprehensively optimizing the dispatch of generators and loads in the virtual power plant system;

[0011] Step 6: Establish a calculation method for the multi - scenario rolling prediction response potential of virtual power plant resources, and continuously and real - time roll and correct based on the results of each real - time collection and detection.

[0012] The specific content of Step 1 is as follows:

[0013] Calculate the energy consumption indicators for the dynamic operation characteristics of diverse power generation resources and equivalent power generation resources.

[0014] For power generation resources and equivalent power generation resources, establish evaluation indicators covering their adjustment speed, adjustment boundary, and adjustment duration; for equivalent power generation resources, establish load resource operation characteristic indicators covering their peak power, peak duration, peak time, as well as valley power, valley duration, and valley time.

[0015] The specific content of Step 2 is as follows:

[0016] Optimize the clustering of multi - heterogeneous virtual power plant data with the minimum of the improved Euclidean distance.

[0017] Firstly, the improved principal component analysis method is used to reduce the dimension of the virtual power plant resource data. Subsequently, the virtual power plant data is divided into k groups, and the average value of each group is calculated as the initial center of the data set. Then, the distance from the virtual power plant data to the center point is calculated to classify the data. Finally, the new center point is calculated based on the center point of the classified data, and the above process is repeated until convergence.

[0018] The specific steps of step 3 are as follows:

[0019] Establish a physical operation model of the virtual power plant's equivalent power generation resources based on multi-dimensional data-driven.

[0020] Firstly, establish a general operation model for the virtual power plant resources of heat storage type. Secondly, establish a general operation model for the virtual power plant resources of electricity storage type. Then, establish a general physical model for other types of virtual power plant resources. Finally, use the fitting method to substitute the aggregated center point value of the virtual power plant data, and at this time, the characteristic coefficient values of each type of virtual power plant resource of this type can be obtained.

[0021] The specific steps of step 4 are as follows:

[0022] Establish a general physical model of the virtual power plant.

[0023] Using the fitting method, substitute the aggregated center point value of the virtual power plant data into the model to obtain the characteristic coefficient values of the virtual power plant resources of this type. Finally, the operating power characteristics representing the virtual power plant resources can be obtained from the general physical model of the virtual power plant.

[0024] The specific steps of step 5 are as follows:

[0025] Consider the combined pre-scheduling of traditional generating units and virtual power plants.

[0026] With the goal of minimizing the total scheduling cost, comprehensively optimize the scheduling of the power generation resources and equivalent power generation resources of the virtual power plant system to meet the system balance constraints, equipment operating power constraints, and user comfort requirements.

[0027] The specific steps of step 6 are as follows:

[0028] Establish a calculation method for the multi-scenario rolling prediction response potential of virtual power plant resources.

[0029] Firstly, establish an operation increment model for virtual power plant resources, including virtual power plant resources of heat storage type, virtual power plant resources of electricity storage type, and other types of virtual power plant resources. Then, based on the operation increment model, roll and correct the pre-scheduling results to continuously improve the evaluation accuracy.

[0030] Compared with the prior art, the present invention has the following advantages and effects:

[0031] The present invention adopts a method combining physics and data, which not only reduces the difficulty of exponentially solving massive data but also improves the interpretability of the operating model, effectively realizes the accurate evaluation of the dispatchable potential of virtual power plant resources under multiple operating scenarios, and improves the availability of virtual power plant resources.

[0032] The improved principal component analysis method established by the present invention can effectively improve the speed of dimensionality reduction of massive data and enhance the accuracy of the dimensionality reduction result. The established improved Euclidean distance minimum clustering method can accelerate the convergence speed of the clustering algorithm.

[0033] The present invention simultaneously uses a physical model and a data method, which can introduce the implicit quantitative relationship between data into the physical model and also solve the problem that it is difficult to determine the parameters of the physical model. Based on this, the response potential evaluation result can achieve the real-time performance and accuracy of quantitative evaluation. Brief Description of the Drawings

[0034] Figure 1 is the flowchart of the virtual power plant response potential evaluation in the embodiment of the present invention;

[0035] Figure 2 is the virtual power plant response potential evaluation result diagram in the embodiment of the present invention;

[0036] Figure 3 is the evaluation index diagram in the embodiment of the present invention. Detailed Embodiment

[0037] The present invention will be further described in detail below with reference to the drawings and through embodiments. The following embodiments are explanations of the present invention, and the present invention is not limited to the following embodiments.

[0038] Embodiment

[0039] Refer to Figure 1 , a method for evaluating the regulation potential of a virtual power plant driven by both model and data, including the following steps:

[0040] Step 1: Calculate the energy consumption index for the dynamic operating characteristics of multiple power generation resources and equivalent power generation resources. A virtual power plant is a collection of multiple types of resources, including distributed power generation resources, traditional generator set power generation resources, and equivalent power generation resources centered on load.

[0041] For power generation resources and equivalent power generation resources, evaluation indexes covering their adjustment speed, adjustment boundary, and adjustment duration are established, as shown in Figure 3 .

[0042] Taking downward regulation as an example, after receiving the control signal, the power generation resources and equivalent power generation resources will gradually reduce their operating power, and finally stabilize near a certain stable power, with an allowable fluctuation range of P os α%. Among them, the adjustment speed is expressed as:

[0043]

[0044] Among them, P init and P end are the initial operating power and the final operating power after regulation respectively, and T ra is the power decline process. The regulation boundary is is the minimum power generation of the power generation resource and the equivalent power generation resource, is the maximum power generation of the power generation resource and the equivalent power generation resource; the regulation duration is the duration that the power generation resource and the equivalent power generation resource can last at a certain power generation level.

[0045] For the equivalent power generation resource, load resource operation characteristic indexes covering its peak power, peak duration, peak time, valley power, valley duration, and valley time are established. The peak power P l max and the valley power P l min are the maximum and minimum operating powers of the load resource during a day respectively, that is

[0046]

[0047] The peak duration and valley duration are the durations when the load resource operation power is greater than 0.7P l max and less than 1.3P l min respectively; the peak time and valley time are the time ranges when the peak power and valley power of the load resource appear. Taking 24 hours a day as an assignment cycle, each hour is divided into an interval and assigned a value, and this value is used to represent the peak time and valley time.

[0048] Step 2: Optimize the clustering of multi-source heterogeneous virtual power plant data with the minimum improved Euclidean distance. First, the virtual power plant resource data includes its operating data on a typical day and the energy consumption indexes of the multi-source dynamic operation characteristics proposed in Step 1. Among them, the operating data on a typical day has a measurement cycle of 15 minutes, and there are 96 measurement points in a day. Considering the large number of virtual power plant resources and the huge virtual power plant resource data, a data dimensionality reduction method is established. In order to improve the importance of the proposed energy consumption index system, a coefficient is multiplied by the proposed energy consumption index, that is

[0049]

[0050] Among them, X ij is the original virtual power plant resource data matrix, It is the corrected virtual power plant resource data matrix. Then, all the data is divided by its maximum value to standardize the virtual power plant resource data. The process is as follows:

[0051]

[0052] Among them, is the standardized virtual power plant resource data. Based on this, the principal component analysis method is used to reduce the dimension of the virtual power plant resource data to obtain the key information contained in the massive data.

[0053] Secondly, select the initial center value: randomly divide the virtual power plant data into k groups, and calculate the average value of each group as the initial center of the data set, that is

[0054]

[0055] Among them, X ij is the original data of the virtual power plant resources, k q refers to the data set that is divided into the k q th group, and q is the amount of its data. Then classify the data based on the distance: the distance matrix of all data to these k data The calculation method is:

[0056]

[0057] Among them, α q is the average value adjustment coefficient to change the calculation accuracy and speed. Based on the distance matrix, all data is re-divided into k categories with the goal of minimizing the distance to the average value point. Then recalculate the center point value: calculate the average value point of the divided data Calculate the new center point value according to the following formula:

[0058]

[0059] Among them, is the center point adjustment coefficient, which can effectively prevent the deviation of the center point selection. Finally, repeat the above process according to the new center point, and the center point will finally converge on a stable convergence point. At this time, all virtual power plant data is aggregated into k typical types and represented by the center point value.

[0060] Step 3: Establish a physical operation model of the virtual power plant equivalent generating resources based on multi-dimensional data-driven. First, establish a general operation model for the heat storage type virtual power plant resources, that is

[0061]

[0062] Among them, C A and R A are the thermal resistance and heat capacity of the heat storage type virtual power plant resources respectively, TA is the internal temperature of the resource, Q AC is the heat load power of the virtual power plant resource of the heat storage type, Q OUT and Q IN are the external heat transfer power and the internal heat generation power respectively.

[0063] Secondly, establish a general operation model for the virtual power plant resources of the electricity storage type, that is

[0064]

[0065] Among them, P E is the operating power of the virtual power plant resource of the electricity storage type, P sta is the steady-state operating power, and β is the non-steady-state operation correction coefficient, which is related to the operating state of the virtual power plant resource of the electricity storage type.

[0066] Then, establish a general physical model for other types of virtual power plant resources, that is

[0067]

[0068] Among them, is the characteristic coefficient of the general physical model of other types of virtual power plant resources, and X is the virtual power plant resource data on the potential calculation day.

[0069] Step 4: Use the fitting method to substitute the aggregated virtual power plant data center point value. At this time, the characteristic coefficient values of each type of virtual power plant resource of this type can be obtained. Finally, k virtual power plant general physical models will be obtained to represent the operating power characteristics of the virtual power plant resources.

[0070] Step 5: Consider the joint pre-scheduling of traditional generating units and virtual power plants. Based on the virtual power plant resource physical operation model established in the previous step, when the virtual power plant system comprehensively optimizes and schedules power generation resources and equivalent power generation resources, the goal is to minimize the total scheduling cost, thereby reducing the total operating cost.

[0071]

[0072] Among them, F(P,t) is the total virtual power plant scheduling cost; F g (P g ,t) is the power generation cost of the power generation resource at time t, and F l1 , F l2 , F l3 are the equivalent power generation costs of the virtual power plant resources of the heat storage type, the virtual power plant resources of the electricity storage type, and other types of virtual power plant resources in the equivalent power generation resources respectively.

[0073] In addition, the joint pre-scheduling needs to meet some constraint conditions, including:

[0074] (1) System balance constraint

[0075]

[0076] Among them, P Gi (t) is the optimized scheduling power of unit i participating in the virtual power plant system scheduling at time t; when considering network losses during scheduling, P Loss (t) is the line loss of the system at time t; P L (t) is the load size used during the virtual power plant system scheduling; T1(t) is the corresponding constraint relaxation amount.

[0077] (2) Equipment operating power constraint

[0078]

[0079] The three formulas are respectively the upper and lower limits constraints and the ramp rate constraints of the equipment operating power. Among them, P j (t) is the operating power of the jth load equipment at time t; and are respectively the upper and lower limits of the operating power of the jth load equipment; and are respectively the upper and lower limits of the ramp rate.

[0080] (3) User comfort requirements

[0081] 1) Comfort constraint of temperature

[0082] T min (t) < T in (t) < T max (t)

[0083] In the formula, T in (t) is the real-time indoor temperature, T min and T max are respectively the lower and upper limits of the acceptable temperature, and the upper and lower limits of the temperature can be changed.

[0084] 2) Equipment usage requirements

[0085] |t par -t lea | ≤ τ no

[0086] In the formula, t par 、t lea and τ no are respectively the start time of calling this equipment, the stop time of calling this equipment, and the upper limit of the unacceptable duration. Taking the air conditioner as an example, t par is the time when the air conditioner starts to participate in the power system scheduling, t leais the moment when the scheduling task is completed and the indoor temperature starts to recover, τ no is the user's maximum willingness to participate during a period, and its value is related to the characteristics of the air conditioner and the house. For energy storage loads, it is related to the state of charge SOC and is determined by the minimum acceptable SOC when leaving the schedule.

[0087] 3) Number of device startups and shutdowns

[0088] Θ ≤ Θ max

[0089] Θ is the total number of startups and shutdowns of the load during the scheduling process. To reduce the damage to the device caused by frequent calls and startups and shutdowns, its number of startups and shutdowns must meet certain limiting requirements.

[0090] Step 6: Establish a calculation method for the multi-scenario rolling prediction response potential of virtual power plant resources. Based on the pre-scheduling result calculated in Step 5, the responsive amount under the optimal economic conditions can be obtained. However, this amount has poor accuracy in real-time situations. Based on this amount, using the general physical model of virtual power plant resources established in Step 4, a real-time feedback correction method is established.

[0091] First, establish the operation increment models of virtual power plant resources, namely virtual power plant resources of heat storage type, virtual power plant resources of electricity storage type, and other types of virtual power plant resources, that is

[0092]

[0093] Then, roll and correct the pre-scheduling result, that is

[0094]

[0095] Among them, is the real-time regulation potential of the virtual power plant system, γ cop is the thermoelectric conversion coefficient of the virtual power plant resources of heat storage type.

[0096] Finally, that is, take the newly corrected value as the base value. During the evaluation process, based on the real-time acquisition and detection results each time, continuous real-time rolling correction will be carried out. The evaluation results are as Figure 2 shown.

[0097] The content not described in detail in this specification belongs to the prior art well-known to those skilled in the art.

[0098] Although the present invention has been disclosed above with embodiments, it is not intended to limit the protection scope of the present invention. Any changes and modifications made by those skilled in the art without departing from the concept and scope of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for evaluating the regulation potential of a virtual power plant driven by dual model data, characterized in that It includes the following steps: Step 1: Calculate the energy consumption index for the dynamic operation characteristics of diversified power generation resources and equivalent power generation resources; For power generation resources and equivalent power generation resources, establish evaluation indexes covering their regulation speed, regulation boundary, and regulation duration; for equivalent power generation resources, establish load resource operation characteristic indexes covering their peak power, peak duration, peak time, as well as valley power, valley duration, and valley time; Step 2: Optimize and cluster the data of the multi - heterogeneous virtual power plant with the minimum improved Euclidean distance; First, use the improved principal component analysis method to reduce the dimension of the virtual power plant resource data. Subsequently, divide the virtual power plant data into k groups, calculate the average value of each group as the initial data set center, then calculate the distance from the virtual power plant data to the center point to classify the data. Finally, calculate a new center point based on the center points of the classified data, and repeat the above process until convergence; Step 3: Establish a physical operation model of the virtual power plant's equivalent power generation resources based on multi - dimensional data - driven; First, establish a general operation model for the resources of the heat - storage type virtual power plant. Second, establish a general operation model for the resources of the electricity - storage type virtual power plant. Then, establish a general physical model for the resources of other types of virtual power plants; Step 4: Establish a general physical model of the virtual power plant; Using the fitting method, substitute the center point values of the aggregated virtual power plant data into the model to obtain the characteristic coefficient values of the virtual power plant resources. Finally, obtain the operation power characteristics of the virtual power plant represented by the general physical model of the virtual power plant; Step 5: Consider the combined pre - dispatch of traditional generator sets and virtual power plants; With the goal of minimizing the total dispatch cost, comprehensively optimize and dispatch the power generation resources and equivalent power generation resources of the virtual power plant system to meet the system balance constraints, equipment operation power constraints, and user comfort requirements; Step 6: Establish a calculation method for the multi - scenario rolling prediction response potential of virtual power plant resources; First, establish operation increment models for the resources of the heat - storage type virtual power plant, the electricity - storage type virtual power plant, and the resources of other types of virtual power plants. Then, based on the operation increment models, roll - correct the pre - dispatch results to continuously improve the evaluation accuracy.

Images