New energy power station abandoned electricity storage auxiliary decision method

Abstract

The application discloses a new energy power station abandoned electricity storage auxiliary decision-making method, and the auxiliary decision-making method comprises the following processes: a new energy power generation model and an energy storage release model are established, and Monte Carlo simulation is adopted for prediction; according to the prediction results of the new energy power generation model and the energy storage release model, a new energy power station abandoned electricity storage auxiliary decision-making is obtained; and the new energy power station abandoned electricity storage auxiliary decision-making is obtained according to the commonness and characteristics of various energy storages, combined with the indexes of the energy storage types, and the auxiliary decision-making suggestions of the power station storable power are obtained.

Description

Technical Field

[0001] This invention relates to the field of coordinated peak shaving of centralized new energy power plants and energy storage power plants, and in particular to a guidance method for coordinated capabilities based on data mining. Background Technology

[0002] In recent years, to achieve the "dual carbon" goals, the composition of the power energy sector has undergone significant adjustments. With the rapid changes in the power grid structure and power source composition, the State Grid Corporation of China faces new challenges. The proportion of renewable energy integration is increasing year by year. Renewable energy generation has its own volatility and randomness, making it difficult for the grid to adjust its active and reactive power. Currently, grid peak shaving is almost entirely handled by conventional thermal power units, but as carbon emission reduction targets continue to intensify, traditional thermal power units will inevitably be gradually reduced.

[0003] New energy power generation is typically greatly affected by geographical location and weather conditions. Taking wind power as the most common example: wind power generation is affected by the seasons, with significant variations in wind strength and direction, resulting in large fluctuations in power generation. This fluctuation cycle is uncontrollable. If a large amount of electricity generated is wasted when the grid load is low, it results in substantial waste; conversely, when the load is high, there may be no wind, preventing any power output. Therefore, to increase the penetration rate of new energy sources, their highly volatile nature must be addressed.

[0004] Energy storage power stations are a type of power station that has developed rapidly in recent years. The biggest difference between them and traditional power stations is that they do not actually generate electricity. Instead, they convert electrical energy into other forms of energy (such as chemical energy, potential energy, etc.), and then convert these other forms of energy back into electrical energy to supply power to the grid when needed. The ability of energy storage power stations to regulate peak and off-peak energy is their biggest feature.

[0005] The current State Grid dispatching system still relies primarily on thermal power generation for output, with peak and valley load regulation mainly handled by thermal power units capable of peak shaving. It typically dispatches power output from various substations based on current load conditions, employing traditional dispatching methods. This lacks a holistic approach to green energy consumption supported by accurate forecasting data, hindering the improvement of clean energy penetration. Ultimately, the goal of the entire power system is to consider how to increase the proportion of renewable energy consumption, improve the cleanliness of our power generation, and make our power grid safer and more reliable. Summary of the Invention

[0006] To address the problems in existing technologies, this invention aims to establish power load models and new energy power generation models by analyzing historical electricity consumption and new energy power generation data through data mining, thereby increasing the penetration rate of new energy sources and their ability to participate in peak shaving.

[0007] This application achieves the above effects through the following technical solution:

[0008] A decision-making support method for storing abandoned electricity from renewable energy power plants, the decision-making support method comprising the following processes:

[0009] Establish new energy power generation models and energy storage release models, and use Monte Carlo simulation for prediction;

[0010] Based on the prediction results of the new energy power generation model and the energy storage release model, auxiliary decision-making for the storage of abandoned new energy power plants is obtained.

[0011] The auxiliary decision-making method for storing abandoned electricity in new energy power plants is based on the commonalities and characteristics of various types of energy storage, combined with the indicators of energy storage type, to obtain auxiliary decision-making suggestions on the amount of electricity that the power plant can store.

[0012] Furthermore, the auxiliary decision-making suggestion for the storable electricity of the power station is as follows: the formula for storable electricity is:

[0013] E store =E v -E upload

[0014] When E store ≥E all-can-store hour:

[0015] When E store <E all-can-store Time: E store =E store

[0016] In the formula E store E is a new energy source that can be used for storage. all-can-release For the available capacity of all energy storage power stations, E all-can-store This refers to the storage capacity available for all energy storage power stations.

[0017] Furthermore, the specific process for establishing the new energy power generation model and the energy storage release model is as follows:

[0018] Acquire historical data on power generation and consumption of new energy power plants and energy storage power plants. The historical data includes new energy power generation data, information on power generation equipment and real-time monitoring data of new energy power plants, and information on energy storage equipment and real-time monitoring data of energy storage power plants.

[0019] The historical data is segmented and analyzed to obtain the first constraint on new energy power generation. Then, based on the first constraint and the acquired new energy power generation data, the new energy power generation model is obtained using time series method and discrete data sampling method.

[0020] Cluster analysis is performed based on the characteristics of energy storage types to determine the amount of electricity that an energy storage station can receive and store, as well as the amount of power it can release to the grid at a given time, and an energy storage release model is established.

[0021] Furthermore, the new energy power generation model is established as follows:

[0022] The historical data is divided into time slices of a set duration according to the time series, and additional attributes are added to the segmented data to obtain the processed historical data.

[0023] The processed historical data is extracted according to different attributes, and its distribution is analyzed to obtain the first constraint affecting new energy power generation. Then, based on the first constraint and the obtained new energy output data, the new energy power generation model is obtained.

[0024]

[0025] Where X is the output data sequence retrieved from the database according to the additional attribute conditions, fi is the output weight value, β is the redundancy coefficient, and l, s, t, and w are additional attributes.

[0026] Furthermore, the energy storage release model is established as follows:

[0027] According to the type of energy storage, the power generation process is divided into pumped hydro storage, battery energy storage, hydrogen energy storage, and compressed air energy storage;

[0028] Cluster analysis is performed based on the characteristics of energy storage types to determine the amount of electricity that an energy storage station can receive and store, as well as the amount of power it can release to the grid at a given time, and an energy storage release model is established.

[0029] Furthermore, the energy storage release model is as follows:

[0030] Based on the output results of the new energy power generation model and the energy storage release model, the commonalities and characteristics of various types of energy storage are analyzed, and the amount of electricity that can be stored is calculated according to the indicators of the energy storage type. and the amount of electricity that can be released

[0031] Where Erate is the rated capacity of the storage unit, P is the current percentage of power in the storage unit, and Ez is the minimum reserved capacity of the storage unit;

[0032] By summarizing all storable energy, auxiliary decision-making suggestions for the power plant's storable energy are obtained:

[0033] and

[0034] The upper limit of electricity available for grid connection and the amount of electricity available for storage are calculated based on real-time current storage data and new energy forecast data.

[0035] In a preferred embodiment of this application, the additional attributes are l: latitude and longitude coordinates, s: season, t: temperature, and w: time period.

[0036] As a preferred embodiment of this application, the prediction using Monte Carlo simulation specifically involves: using Monte Carlo simulation to analyze the impact of changes in the first constraint on new energy power generation, and plotting the trend curve of the change.

[0037] Beneficial effects

[0038] The advantage of the collaborative peak shaving and valley filling strategy of new energy power plants and energy storage power plants based on data mining is that it can maximize the penetration rate of green new energy and reduce carbon emissions while ensuring the safe operation of the power grid. The new energy power plant abandoned power storage auxiliary decision-making method provided in this application integrates new energy power generation model and energy storage release model to understand the grid-connected new energy situation at that time. Combined with the current energy storage percentage, it can store the power that cannot be connected to the grid in the energy storage power plant as much as possible. Only when the energy storage power plant is full will the power be truly abandoned. In this way, the power generated by new energy can be used to the maximum extent and the benefits can be maximized. Attached Figure Description

[0039] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application.

[0040] Figure 1 This is a flowchart illustrating the model construction process of the present invention. Detailed Implementation

[0041] The invention will be further described below with reference to the accompanying drawings:

[0042] A decision-making support method for storing abandoned electricity from renewable energy power plants, the decision-making support method comprising the following processes:

[0043] Establish new energy power generation models and energy storage release models, and use Monte Carlo simulation for prediction;

[0044] Based on the prediction results of the new energy power generation model and the energy storage release model, auxiliary decision-making for the storage of abandoned new energy power plants is obtained.

[0045] The auxiliary decision-making method for storing abandoned electricity in new energy power plants is based on the commonalities and characteristics of various types of energy storage, combined with the indicators of energy storage type, to obtain auxiliary decision-making suggestions on the amount of electricity that the power plant can store.

[0046] Furthermore, the auxiliary decision-making suggestion for the storable electricity of the power station is as follows: the formula for storable electricity is:

[0047] E store =Ev -E upload

[0048] When E store ≥E all-can-store hour:

[0049] When E store <E all-can-store Time: E store =E store

[0050] In the formula E store E is a new energy source that can be used for storage. all-can-release For the available capacity of all energy storage power stations, E all-can-store This refers to the storage capacity available for all energy storage power stations.

[0051] The operation process of the technical solution is as follows: Figure 1 As shown, the specific steps are described below:

[0052] New energy power generation model

[0053] The basic idea is to collect real-time power generation data from a renewable energy power plant, segment the data according to time series, combine and store load data from each time slice with important factor data such as weather data from the same time slice, use discrete analysis to determine the normal distribution of power generation data under the same conditions, identify key points of interest, formulate a value selection scheme, and establish a renewable energy power generation model. The purpose of this model is to obtain predicted power generation values ​​by inputting specified influencing factors. The model is used to predict future renewable energy power generation under the same conditions and to support decision-making.

[0054] For example:

[0055] 1. For example, basic data such as the installed capacity, commissioning time, and conversion rate of all photovoltaic cells in a certain photovoltaic power station have been obtained, as well as real-time power generation data of each unit in this photovoltaic power station for the past three years (this data changes continuously at 5-second intervals);

[0056] 2. Divide the time series into 5-second time slices (the time slice can only be greater than or equal to the minimum data sampling frequency; the smaller the time interval, the easier the data is to use). A time slice will have its basic attributes (e.g., date, season, time interval, etc.), some directly related attributes (e.g., meteorological attributes: weather, temperature, air pressure, sunshine, wind speed, wind direction, etc.), some indirectly related attributes (e.g., station latitude and longitude coordinates, battery pack failure rate, station power quality indicators, station high and low voltage ride-through capability, station anti-islanding capability, grid-connected inverter conversion rate, station grid-connected power consumption, station control system power consumption, etc.), and some additional attributes (e.g., whether it is a working period, whether it is a maintenance period, etc.).

[0057] 3. Now we have the real-time power generation corresponding to 3*365*24*60*(60 / 5) time slices, and each of these time slices has its own attributes.

[0058] 4. Data can be extracted and its distribution analyzed according to different attribute data (e.g., data can be sampled and analyzed according to four dimensions: latitude and longitude, season, weather, and time period). For example, a power generation analysis can be performed using a time slice centered at 31°N, 121°E, during winter, cloudy weather, and between 4-5 PM. Through filtering, N discretely distributed power generation data points are extracted. Mathematical methods such as central tendency, dispersion, and shape are used for data distribution analysis. The appropriate parameters for establishing a new energy power generation model are determined based on the actual data distribution. This model will have different forms depending on the data granularity and considerations. For ease of understanding, an abstract example is provided using the above four attribute inputs (l: latitude and longitude coordinates, s: season, t: temperature, w: time period) and a weighted average method. This model can be summarized as follows: In the form of , X is the power generation data sequence retrieved from the database according to the above search conditions, f is the power generation weight value, and β is the redundancy coefficient (usually less than 1). The previous analysis process was to determine the selection of its weight value and analyze how to select the redundancy coefficient.

[0059] 5. Once the model is established, input l, s, t, w to obtain f(l,s,t,w), which is the predicted power generation value under this condition; based on f(l,s,t,w), future conditions (such as weather forecasts) can be used as input parameters to obtain the predicted power generation value at future time points;

[0060] 6. By repeatedly calling this model, a power generation prediction curve for the next n hours can be generated. By plotting the prediction curve, the trend of future power generation changes can be analyzed, and strategies can be adjusted in a timely manner.

[0061] Energy storage and release model

[0062] Basic idea: Collect energy storage index data from all operational energy storage stations, perform cluster analysis based on the characteristics of energy storage types, determine the amount of electricity that the energy storage station can receive and store at a given time, and the amount of electricity that can be released to the grid. Establish an energy storage release model. The purpose of this model is to output the amount of electricity that can be stored and released, given the station type and current storage capacity, to aid in decision-making regarding whether to directly connect renewable energy generation to the grid or store it for later use. For ease of description, this model does not consider line losses in the transmission circuits from renewable energy stations to energy storage stations, nor does it consider energy losses caused during the storage of electricity at the energy storage stations.

[0063] For example:

[0064] 1. For example, basic data such as energy storage type, installed capacity, commissioning time, conversion rate, rated power, and maximum output power of all energy storage stations in a certain city have been obtained, as well as the current energy storage data of each energy storage module in these stations (this data changes continuously at 5-second intervals);

[0065] 2. According to energy storage type, it can be classified into pumped hydro storage, battery energy storage, hydrogen energy storage, compressed air energy storage, etc.

[0066] 3. Based on the commonalities (such as installed capacity, rated power, and maximum output power) and characteristics (such as pumped hydro storage not requiring inverters, battery energy storage experiencing time-varying degradation, and different conversion rates for different types), conduct corresponding analyses and calculate the amount of electricity that can be stored based on the indicators of each energy storage type. and the amount of electricity that can be released

[0067] Where Erate is the rated capacity of the storage unit, P is the current percentage of power in the storage unit, and Ez is the minimum reserved capacity of the storage unit.

[0068] 4. Summarize all storable electricity to derive a city-wide storable electricity model. and

[0069] 5. When making peak-shaving auxiliary decisions for energy storage power stations, the upper limit of electricity available for grid connection and the amount of electricity available for storage can be calculated based on real-time current storage data and new energy forecast data.

[0070] Decision Support for the Storage of Curbed Electricity from New Energy Power Plants

[0071] The decision-making process for storing or abandoning electricity is relatively simple; the formula for the amount of electricity that can be stored is:

[0072] E store =E v -E upload

[0073] When E store ≥E all-can-store hour:

[0074] When E store <E all-can-store Time: E store =E store

[0075] In the formula E store E is a new energy source that can be used for storage. all-can-release For the available capacity of all energy storage power stations, E all-can-store This refers to the storage capacity available for all energy storage power stations.

[0076] Currently, the dispatching strategy for renewable energy relies on a short-term power forecasting system (15 minutes to 4 hours) provided by renewable energy power plants, combined with real-time grid load dispatching of renewable energy output. The time resolution of the forecast is only 15 minutes. Firstly, this forecast granularity is too large; due to the volatility of renewable energy, significant changes can occur within 15 minutes. Secondly, this forecast only reflects the output of renewable energy and does not comprehensively consider the changing trends of electricity load or the current status of energy storage. If the accuracy of this forecast changes, it will lead to large fluctuations. Therefore, to ensure grid security, dispatching generally does not allow excessive participation from renewable energy, resulting in a large amount of renewable energy being wasted. If renewable energy is wasted, it represents a huge waste of energy; what could have been a way to save fossil fuels and reduce carbon dioxide emissions fails to produce the desired effect. Furthermore, the construction of renewable energy power plants requires significant investment. Our ultimate goal is to utilize renewable energy as much as possible. Currently, most of the electricity generated by new energy power plants is wasted if it cannot be connected to the grid. However, by comprehensively using new energy power generation models and energy storage release models, we can understand the grid-connected new energy situation at that time. Combined with the current percentage of energy storage capacity, we can store as much of the electricity that cannot be connected to the grid as possible in energy storage power stations. Only when the energy storage power station is fully full will the electricity be truly abandoned. In this way, we can maximize the use of electricity generated by new energy and achieve maximum benefits.

[0077] It's important to note that the actual model building process can be relatively simple, limited by data volume or low prediction accuracy requirements, or it can be very complex, requiring abundant data, numerous considerations, and high prediction accuracy. Due to space limitations and varying user needs, this section only provides a brief description, illustrating the thought process and steps. Ultimately, the more historical data collected, the finer the granularity, and the more relevant attributes considered, the more accurate the model will be, leading to more accurate predictions in later stages. This results in greater assistance for decision-making and higher reliability. Model building is an iterative process. A preliminary model can be built through historical data analysis. To verify the model's accuracy, machine learning techniques can be used to simulate the model using historical data, analyzing the differences between the summarized values, mean values, and actual data. The modeling parameters can be continuously adjusted to make the final model more representative.

[0078] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make several improvements and adjustments without departing from the principle of the present invention, and these improvements and adjustments should also be considered within the scope of protection of the present invention.

Claims

1. A new energy power station abandoned electricity storage auxiliary decision method, characterized in that, The auxiliary decision-making method The auxiliary decision-making method comprises the following processes: A new energy generation model and a storage release model are established, and Monte Carlo simulation is used for prediction; According to the prediction results of the new energy generation model and the storage release model, a new energy power station abandoned power storage auxiliary decision is obtained; The new energy power station abandoned power storage auxiliary decision is obtained according to the commonness and characteristics of various types of storage and in combination with the indexes of the storage types, and an auxiliary decision suggestion of the power station storable power is obtained; The auxiliary decision suggestion of the power station storable power is specifically as follows: the formula of the storable power is: E store = E v - E upload When E store ≥ E all-can-store then: When E store <E all-can-store E store = E store where E store is the new energy power available for storage, all-can-release is the total available capacity of all energy storage plants, all-can-store is the total available capacity of all energy storage plants; The specific process of establishing the new energy generation model and the storage release model is as follows: Historical data of power generation and power consumption of a new energy power station and a storage power station are obtained, and the historical data comprises new energy generation data, new energy power station generation equipment information and real-time monitoring data, storage power station storage equipment information and real-time monitoring data; The historical data are subjected to segmentation analysis to obtain a first constraint condition affecting new energy generation, and then the first constraint condition and obtained new energy generation data are used to obtain the new energy generation model by using a time series method and a discrete data sampling method; According to the characteristics of the storage types, clustering analysis is performed to determine the power that can be received and stored by the storage power station and the output that can be released to the power grid at a set time, and a storage release model is established; The new energy generation model is established as follows: The historical data are segmented into time slices in a set time unit according to a time sequence, and the segmented data are given a set additional attribute to obtain processed historical data; The processed historical data are subjected to data extraction according to different attributes, and the distribution thereof is analyzed to obtain the first constraint condition affecting new energy generation, and then the first constraint condition and obtained new energy output data are used to obtain the new energy generation model: where X i is the output data sequence retrieved from the database according to the additional attribute condition, fi is the new energy power generation weight value, β is the redundancy coefficient, and l, s, t, and w are additional attributes.

2. The new energy power station abandoned electricity storage auxiliary decision-making method according to claim 1, characterized in that, The storage release model is established as follows: The power generation process is divided into pumped storage, battery storage, hydrogen storage and compressed air storage according to the storage types; According to the characteristics of the storage types, clustering analysis is performed to determine the power that can be received and stored by the storage power station and the output that can be released to the power grid at a set time, and a storage release model is established.

3. The new energy power station abandoned electricity storage auxiliary decision method according to claim 2, characterized in that, The storage release model is as follows: According to the output results of the new energy generation model and the energy storage release model, the commonness and characteristics of various types of energy storage, and the index of the energy storage type, the storable electric quantity and the releasable electric quantity are calculated ​ wherein is the rated capacity of the storage unit, P i is the current percentage of charge of the storage unit, is the minimum reserve capacity of the storage unit; All the storable electric quantity is summarized to obtain the auxiliary decision suggestion of the storable electric quantity of the power station: and According to real-time current reserve data and new energy prediction data, the upper limit of the available online and the available power for storage are calculated, E all-can-release For all energy storage power stations, the releasable capacity is E all-can-store For all energy storage power stations, the releasable capacity is E 4. The new energy power station abandoned electricity storage auxiliary decision method according to claim 1, characterized in that, The additional attribute is l: latitude and longitude coordinates, s: season, t: temperature and w: time period.

5. The new energy power station abandoned electricity storage auxiliary decision-making method according to claim 1, characterized in that, The Monte Carlo simulation is used for prediction as follows: the Monte Carlo simulation is used to analyze the influence of the change of the first constraint condition on new energy generation, and a change trend curve is drawn.

Images