Artificial intelligence-based virtual power plant participation in multi-objective market co-optimization method

Abstract

The application discloses a kind of based on artificial intelligence's virtual power plant participates in multi-objective market collaborative optimization method, and the time period and weather condition of virtual power plant participating in multi-objective market are classified;For each historical period and weather condition, according to the characteristics of 5 markets, the conclusion of virtual power plant participating in multi-objective market is determined;Artificial intelligence model corresponding to each historical period and virtual power plant capacity range is learned using support vector machine algorithm;When online application, input the week, season, weather, time period and virtual power plant capacity range of current period, directly obtain the market that virtual power plant can participate in through corresponding support vector machine model, and calculate the income.The application applies artificial intelligence method, and by the statistical analysis of historical period virtual power plant participates in each market situation, online can directly guide virtual power plant to participate in each market to obtain maximum income.

Description

Technical Field

[0001] This invention relates to a collaborative optimization method, and more particularly to a collaborative optimization method for virtual power plants participating in multi-objective markets based on artificial intelligence, belonging to the field of intelligent virtual power plant technology. Background Technology

[0002] The construction of virtual power plants aligns with the national energy strategy and represents a feasible solution for enterprises to address the uncertainties surrounding the integration and introduction of new energy sources, as well as to explore alternative economic growth drivers. However, existing research focuses on the aggregation of multiple energy sources and the optimization of individual markets, neglecting multi-market synergistic optimization. Currently, there is no evidence of virtual power plants participating in multi-target market synergy optimization, which hinders their potential to simultaneously participate in multiple markets and generate additional revenue. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method for collaborative optimization of virtual power plants participating in multi-objective markets based on artificial intelligence, so as to enable virtual power plants to participate in multiple objective markets at the same time and earn more revenue.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0005] A method for collaborative optimization in a multi-objective market using a virtual power plant based on artificial intelligence includes the following steps: S1. Classify the time periods and weather conditions for virtual power plants to participate in multi-target markets; S2. For each historical period and weather condition, the conclusion of virtual power plant participation in multi-objective markets is determined based on the characteristics of the five markets: electrical energy, primary frequency regulation, peak shaving, secondary frequency regulation, and voltage regulation. S3. After obtaining the conclusions of each historical period segment and each virtual power plant capacity level and the corresponding virtual power plants participating in the multi-objective market to earn profits, the support vector machine algorithm is used to learn the artificial intelligence model corresponding to each historical period and virtual power plant capacity level. S4. When using the online application, input the current day of the week, season, weather, time of day, and virtual power plant capacity level. The corresponding support vector machine model will directly obtain the market in which the virtual power plant can participate and calculate the revenue.

[0006] Furthermore, in step S1, for different historical times within the same week, season, weather, and time period, the multi-target market in which the virtual power plant participates is learned.

[0007] Furthermore, in step S2, the profit earned by the virtual power plant participating in the multi-target market is jointly determined by the virtual power plant's capacity and the profitability sensitivity of participating in the market for electrical energy per unit capacity, primary frequency regulation, peak shaving, secondary frequency regulation, and voltage regulation.

[0008] Further, step S2 specifically includes: At a given time, based on the order of the electricity energy, primary frequency regulation, peak shaving, and secondary frequency regulation markets, we determine which market we need to participate in. The active power increase or decrease in the first-ranked market is used as the basis. If there are no subsequent markets to participate in, the participation amount is the active power increase or decrease value of the current market. If there are subsequent markets to participate in, we need to determine whether the direction of the active power increase or decrease in the market to participate in is consistent with the direction of the active power increase or decrease in the first-ranked market. If they are inconsistent, we abandon the market to participate in. If they are consistent, the participation amount is recorded as the maximum of the active power change required by the market to participate in and the active power change required by the first-ranked market. For the voltage regulation market, if participation in the voltage regulation market is required during a certain period, then the active power participation quota for that period in the energy / primary frequency regulation / peak shaving / secondary frequency regulation market is denoted as P. Based on the virtual power plant capacity S, the reactive power output / absorption Q during that period shall not exceed [a certain value]. Based on this, the reactive power output / absorption amount Q is obtained.

[0009] Furthermore, the calculation process for the virtual power plant to earn revenue by participating in a multi-target market is as follows: Electricity market: Revenue = Peak-valley price difference × Declared capacity; Primary frequency regulation market: Taking low frequency as an example, the real-time increase in active power is as follows:

[0010] in, This represents the increase in active power. For frequency limits, k is the droop coefficient; Multiply by the actual frequency modulation time to obtain the actual operating power. Subtract 70% of the theoretical operating power, multiply by the peak-valley price difference, and you get the profit from participating in the frequency regulation market. Peak shaving market: Revenue = Demand capacity × Demand price; Secondary frequency regulation market: Revenue = Capacity compensation + Mileage compensation, Capacity compensation = Total regulation mileage in the month × 360 yuan / MW·month, Mileage compensation = Change in active power × 3 yuan / MW; Voltage regulation market: Revenue = Cumulative reactive power output / Absorption ∑Q × 15 yuan / Mvarh.

[0011] Furthermore, in step S3, the support vector machine algorithm adjusts the penalty coefficient c and the polynomial kernel-specific parameter gamma to match the historical time periods of each segment of the input information with the capacity levels of each virtual power plant and the conclusion of the output information that the virtual power plant participates in the multi-target market to earn revenue.

[0012] Furthermore, the penalty coefficient c is initially set to 1, and the polynomial kernel-specific parameter gamma is initially set to 0.1.

[0013] Compared with the prior art, the present invention has the following advantages and effects: The present invention applies artificial intelligence methods to conduct statistical analysis on the participation of virtual power plants in various markets during historical periods, and can directly guide virtual power plants to participate in various markets online to obtain maximum benefits. Attached Figure Description

[0014] Figure 1 This is a flowchart of a method for collaborative optimization of a virtual power plant in a multi-objective market based on artificial intelligence, according to the present invention. Detailed Implementation

[0015] To illustrate in detail the technical solutions adopted by the present invention to achieve the intended technical objectives, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Furthermore, the technical means or technical features in the embodiments of the present invention can be replaced without creative effort. The present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0016] like Figure 1 As shown, the present invention provides a method for collaborative optimization of virtual power plants participating in multi-objective markets based on artificial intelligence, comprising the following steps: S1. Classify the time periods and weather conditions for virtual power plants to participate in multi-target markets.

[0017] For offline learning of historical periods, it is necessary to conduct separate learning according to season, month, week, and holiday, and also to take into account the different weather (sunny / cloudy / rainy) and time periods (every hour or 15 minutes).

[0018] For different historical times (years may vary) within the same week, season, weather, and period, learn how the virtual power plant participates in multi-target markets to maximize profits.

[0019] The virtual power plant capacity can also be divided into tiers (such as 0~5MW, 5~10MW, 10~15MW, etc.). Using the above method, the participation of virtual power plants in various markets during the same period can be obtained under different tiers, and the revenue can be calculated.

[0020] S2. For each historical period and weather condition, the conclusion of virtual power plant participation in the multi-objective market is determined based on the characteristics of the five markets: electrical energy, primary frequency regulation, peak shaving, secondary frequency regulation, and voltage regulation (including but not limited to their requirements for active power output and their compensation (revenue)).

[0021] The profitability of virtual power plants participating in multi-target markets is jointly determined by the virtual power plant's capacity and the sensitivity of its profitability to market participation in primary frequency regulation, peak shaving, secondary frequency regulation, and voltage regulation, based on the power output per unit capacity.

[0022] At a given time period, based on the order of the electricity market, primary frequency regulation, peak shaving, and secondary frequency regulation market, we determine which market we need to participate in at the moment. The participation amount is based on the increase or decrease in active power generation in the first-ranked market (if we do not participate in the electricity market at the moment, but can participate in the primary frequency regulation market, the participation amount is based on the increase or decrease in active power generation when participating in the primary frequency regulation market). If there are no subsequent markets to participate in, the participation amount is the increase or decrease in active power generation in the current market. If there are subsequent markets to participate in, we need to determine whether the direction of the increase or decrease in active power generation in the market to participate in is consistent with the direction of the increase or decrease in active power generation in the first-ranked market. If they are inconsistent, we abandon the market to participate in. If they are consistent, the participation amount is recorded as the maximum of the change in active power generation required by the market to participate in and the change in active power generation required by the first-ranked market.

[0023] For the voltage regulation market, if participation in the voltage regulation market is required during a certain period, then the active power participation quota for that period in the energy / primary frequency regulation / peak shaving / secondary frequency regulation market is denoted as P. Based on the virtual power plant capacity S, the reactive power output / absorption Q during that period shall not exceed [a certain value]. Based on this, the reactive power output / absorption amount Q is obtained.

[0024] The calculation process for virtual power plants to earn revenue by participating in multi-target markets is as follows: Electricity market: Revenue = Peak-valley price difference (0.65 yuan / kWh) × Declared capacity; Primary frequency regulation market: Taking low frequency as an example, the real-time increase in active power is as follows:

[0025] in, This represents the increase in active power. For frequency limits, k is the droop coefficient; Multiply by the actual frequency modulation time to obtain the actual operating power. Subtract 70% of the theoretical operating power, multiply by the peak-valley price difference, and you get the profit from participating in the frequency regulation market. Peak shaving market: Revenue = Demand capacity × Demand price; Secondary frequency regulation market: Revenue = Capacity compensation + Mileage compensation, Capacity compensation = Total regulation mileage in the month × 360 yuan / MW·month, Mileage compensation = Change in active power × 3 yuan / MW; Voltage regulation market: Revenue = Cumulative reactive power output / Absorption ∑Q × 15 yuan / Mvarh.

[0026] The calculation of revenue will be further explained below through specific examples.

[0027] Peak demand on a typical day between 8 and 9 a.m.

[0028] Virtual power plant pricing information

[0029] Winning bid capacity of each virtual power plant

[0030] Peak-valley electricity price for a typical day between 8-9 am: 8-9 AM: Peak hour, 0.5583 yuan / kWh, requires discharging; Off-peak hours: 0.3583 yuan / kWh; AGC frequency regulation capacity at 8-9 am on a typical day (before the response phase, frequency regulation resources participating in market bidding need to submit bidding information such as capacity quotation, mileage quotation and frequency regulation capacity, with a time scale of 1 hour).

[0031] The frequency regulation capacity is 150MW from 8 to 9 o'clock, with a frequency regulation duration of 1 hour. The capacity price is 360 yuan / MW·month, and the mileage price is 3 yuan / MW.

[0032] Combining the above (peak shaving, electrical energy, AGC frequency regulation), the total revenue is: Peak shaving: (50) 215+70 220+30 310) 0.25+(50) 215+70 220+60 310) 0.25+(50) 215+70 220+80 310) 0.25+(50) 215+70 220+80 310) 0.25 = 35450 0.25+44750 0.25+50950 0.25+50950 0.25 = 45525 yuan; Electricity market: ≥150 (0.5583-0.3583) 1000 = 30000 yuan, peak-valley arbitrage is at least 150MW.

[0033] AGC FM market: Basic compensation: 150 60 / 60 360 = 54,000 yuan; Call compensation: 150 3 = 450 yuan; Voltage regulation: The current reactive power at the grid connection point is 1131.1 MVar, and the received reactive power dispatch instruction from the upper level is 1000 MVar, so the reactive power needs to be reduced by 131.1 Mvar; Assuming the total capacity of the virtual power plant is 200MVA or higher It can simultaneously handle changes in active and reactive power, with voltage regulation compensation: 15. 131.1 = 1966.5 yuan.

[0034] Total: 45525 + 30000 + 54000 + 450 + 1966.5 = 131941.5 yuan.

[0035] This demonstrates the advantages of overlapping compensation.

[0036] The revenue from a single frequency modulation, which can be added in real time, has not yet been factored in.

[0037] S3. After obtaining the historical time periods and virtual power plant capacity levels (inputs) and the corresponding conclusions (outputs) of virtual power plants participating in the multi-objective market to earn profits, the AI ​​model corresponding to each historical time period and virtual power plant capacity level is learned using the support vector machine algorithm.

[0038] The support vector machine algorithm, by adjusting the penalty coefficient c and the polynomial kernel-specific parameter gamma, ensures that the historical time periods of each segment of the input information and the capacity levels of each virtual power plant match the conclusions of the output information regarding virtual power plants participating in multi-objective markets to earn profits.

[0039] The penalty coefficient c is initially set to 1, and the polynomial kernel-specific parameter gamma is initially set to 0.1.

[0040] The support vector machine algorithm can also be replaced by algorithms such as decision trees, convolutional neural networks, and long short-term memory networks.

[0041] S4. When using the online application, input the current day of the week, season, weather, time of day, and virtual power plant capacity level. The corresponding support vector machine model will directly obtain the market in which the virtual power plant can participate and calculate the revenue.

[0042] This invention applies artificial intelligence methods to statistically analyze the participation of virtual power plants in various markets over historical periods, and can directly guide virtual power plants to participate in various markets online to maximize their profits.

[0043] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent substitutions, and improvements made to the above embodiments without departing from the scope of the present invention, based on the technical essence of the present invention and within the spirit and principles of the present invention, shall still fall within the protection scope of the present invention.

Claims

1. A method for collaborative optimization of a virtual power plant in a multi-objective market based on artificial intelligence, characterized in that... Includes the following steps: S1. Classify the time periods and weather conditions for virtual power plants to participate in multi-target markets; S2. For each historical period and weather condition, the conclusion of virtual power plant participation in multi-objective markets is determined based on the characteristics of the five markets: electrical energy, primary frequency regulation, peak shaving, secondary frequency regulation, and voltage regulation. S3. After obtaining the conclusions of each historical period segment and each virtual power plant capacity level and the corresponding virtual power plants participating in the multi-objective market to earn profits, the support vector machine algorithm is used to learn the artificial intelligence model corresponding to each historical period and virtual power plant capacity level. S4. When using the online application, input the current day of the week, season, weather, time of day, and virtual power plant capacity level. The corresponding support vector machine model will directly obtain the market in which the virtual power plant can participate and calculate the revenue.

2. The method for collaborative optimization of virtual power plants participating in multi-objective markets based on artificial intelligence according to claim 1, characterized in that: In step S1, for different historical times of the same week, season, weather, and period, the multi-target market in which the virtual power plant participates is learned.

3. The method for collaborative optimization of virtual power plants participating in multi-objective markets based on artificial intelligence according to claim 1, characterized in that: In step S2, the profit earned by the virtual power plant in the multi-target market is jointly determined by the virtual power plant's capacity and the profitability sensitivity of the market participation in primary frequency regulation, peak shaving, secondary frequency regulation, and voltage regulation, based on the power energy per unit capacity.

4. The method for collaborative optimization of virtual power plants participating in multi-objective markets based on artificial intelligence according to claim 3, characterized in that: Step S2 specifically involves: At a given time, based on the order of the electricity energy, primary frequency regulation, peak shaving, and secondary frequency regulation markets, we determine which market we need to participate in. The active power increase or decrease in the first-ranked market is used as the basis. If there are no subsequent markets to participate in, the participation amount is the active power increase or decrease value of the current market. If there are subsequent markets to participate in, we need to determine whether the direction of the active power increase or decrease in the market to participate in is consistent with the direction of the active power increase or decrease in the first-ranked market. If they are inconsistent, we abandon the market to participate in. If they are consistent, the participation amount is recorded as the maximum of the active power change required by the market to participate in and the active power change required by the first-ranked market. For the voltage regulation market, if participation in the voltage regulation market is required during a certain period, then the active power participation quota for that period in the energy / primary frequency regulation / peak shaving / secondary frequency regulation market is denoted as P. Based on the virtual power plant capacity S, the reactive power output / absorption Q during that period shall not exceed [a certain value]. Based on this, the reactive power output / absorption amount Q is obtained.

5. The method for collaborative optimization of a virtual power plant participating in a multi-objective market based on artificial intelligence, as described in claim 4, is characterized in that: The calculation process for the virtual power plant to earn revenue by participating in a multi-target market is as follows: Electricity market: Revenue = Peak-valley price difference × Declared capacity; Primary frequency regulation market: Taking low frequency as an example, the real-time increase in active power is as follows: in, This represents the increase in active power. For frequency limits, k is the droop coefficient; Multiply by the actual frequency modulation time to obtain the actual operating power. Subtract 70% of the theoretical operating power, multiply by the peak-valley price difference, and you get the profit from participating in the frequency regulation market. Peak shaving market: Revenue = Demand capacity × Demand price; Secondary frequency regulation market: Revenue = Capacity compensation + Mileage compensation, Capacity compensation = Total regulation mileage in the month × 360 yuan / MW·month, Mileage compensation = Change in active power × 3 yuan / MW; Voltage regulation market: Revenue = Cumulative reactive power output / Absorption ∑Q × 15 yuan / Mvarh.

6. The method for collaborative optimization of virtual power plants participating in multi-objective markets based on artificial intelligence according to claim 1, characterized in that: In step S3, the support vector machine algorithm is adjusted by adjusting the penalty coefficient c and the polynomial kernel-specific parameter gamma to match the historical time periods of each segment of the input information with the capacity level of each virtual power plant and the conclusion of the output information that the virtual power plant participates in the multi-target market to earn profits.

7. The method for collaborative optimization of a virtual power plant participating in a multi-objective market based on artificial intelligence, as described in claim 6, is characterized in that: The penalty coefficient c is initially set to 1, and the polynomial kernel-specific parameter gamma is initially set to 0.1.

Images