Distributed resource intelligent optimization regulation and control system and terminal
Through the distributed resource intelligent optimization and regulation system, the use of artificial intelligence prediction and optimization modules, the shortcomings of distributed energy systems in operational coordination and profit distribution are solved, and efficient power generation utilization and system goals are achieved.
Patent Information
- Application Number
- CN202411032620.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-08
AI Technical Summary
The existing distributed energy system has shortcomings in equipment operation coordination, energy optimization management capabilities, source network load storage interaction and system profit distribution, resulting in low system operation efficiency, inability to achieve full consumption and local utilization of new energy, and lack of system Level regulation potential and income statistical distribution mechanism.
Adopt distributed resource intelligent optimization and control system, including prediction module, scheduling optimization module and income calculation module, predict power generation and load conditions through artificial intelligence, generate optimization and control strategies, and conduct profit analysis and calculation to achieve economic regulation and safety management of system goals. .
Through high-precision prediction and optimization strategies, scientifically evaluate distributed power generation power, rationally plan energy storage resources, realize local consumption and maximum utilization of distributed power generation, improve system operation efficiency, and support multi-time scale regulation and revenue statistical distribution.
Smart Images

Figure CN118920471A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of distributed resource scheduling, and in particular to a distributed resource intelligent optimization and control system and a terminal. Background Art
[0002] Distributed energy systems mainly include electric vehicles, energy storage, distributed photovoltaics, distributed wind power and natural gas distributed energy. Distributed energy systems have developed rapidly in recent years, especially in the field of distributed photovoltaics, where installed capacity has achieved significant growth, and the proportion of coordinated application of multiple types of resources such as photovoltaic storage and charging has continued to increase. Despite the rapid development of distributed energy, it still faces some problems and challenges, such as the need to strengthen the coordination of equipment operation, the need to improve the level of energy optimization management capabilities, and the need to strengthen the interaction between sources, grids, loads and storage.
[0003] With the development of the Internet of Things and communication technology, data interconnection between distributed resource system devices and between distributed resource systems and power systems has been achieved, providing a technical basis for intelligent power control. Using intelligent energy control technology to control distributed resource systems in real time can enhance clean energy absorption capacity, reduce electricity costs, and improve system operation efficiency, which will help improve system management level and participate in grid interactive response capabilities.
[0004] In the prior art, fixed strategies or logical judgments are often used to implement system operation management. For example, the operation of the energy storage device is controlled by setting a timed charging and discharging logic for the energy storage system. When the capacity or power of the system exceeds the limit, the total load data of the transformer is collected and judged, and then the energy storage or other adjustable equipment is adjusted to achieve power reduction to meet the boundary constraints of the system operation.
[0005] Or adopt the photovoltaic priority consumption mode, photovoltaic power generation is first supplied to the load, and the excess is charged or returned to the grid. Or adopt the energy storage mode, the energy storage equipment is used as backup power, and the energy storage is only discharged when the grid is abnormal. The photovoltaic and grid supply power to the load and charge the energy storage at the same time; when the grid is normal, the energy storage SOC always maintains a full battery state. Or adopt the microgrid mode, which is suitable for scenarios without grid, photovoltaic and energy storage constitute a pure off-grid system; photovoltaic power generation is first supplied to the load, and the surplus power is used to charge the energy storage.
[0006] It can be seen that the existing system operation has at least the following defects:
[0007] 1. Unable to predict and judge power generation and load conditions, resulting in no reference basis for system regulation;
[0008] 2. It is impossible to fully absorb and locally utilize new energy power generation;
[0009] 3. Each device in the system operates independently, lacks a coordination mechanism, and the system operation efficiency is low;
[0010] 4. The timing charging and discharging or logic judgment strategy is simple and cannot maximize the economic benefits of the system;
[0011] 5. Lack of system-level regulation potential assessment and management capabilities, unable to effectively respond to the regulation needs of the upper market and power grid;
[0012] 6. There is a lack of a system-level revenue statistics and distribution mechanism, and it is impossible to achieve a reasonable distribution of system revenue among different distributed resource stakeholders. Summary of the invention
[0013] The purpose of the present invention includes providing a distributed resource intelligent optimization and control system and terminal, which can reasonably plan adjustable resources such as energy storage, realize local consumption and maximum utilization of distributed power generation, and achieve multi-system goals such as economic regulation, safety management, and demand response.
[0014] The embodiments of the present invention can be implemented as follows:
[0015] In a first aspect, the present invention provides a distributed resource intelligent optimization and control system, the system comprising a prediction module, a scheduling optimization module, an evaluation module and a benefit calculation module, wherein the prediction module, the scheduling optimization module and the benefit calculation module are connected in sequence, and the scheduling optimization module and the benefit calculation module are both connected to the evaluation module;
[0016] The prediction module is used to output prediction data based on artificial intelligence, and the prediction data serves as the input and boundary condition of the scheduling optimization module; the scheduling optimization module generates optimization control strategies based on the prediction data and transmits them to the benefit calculation module; the evaluation module is used to change the input and target of the scheduling optimization module, so that the scheduling optimization module generates the optimal strategy under different evaluation conditions and transmits it to the benefit calculation module; the benefit calculation module is used to perform benefit analysis and calculation based on the strategy generated by the scheduling optimization module.
[0017] In an optional embodiment, the prediction module uses a deep neural network to build a prediction model, collects historical data, forms data feature samples, and constructs a training data set and a sample library, mines the data in the sample library, optimizes the association between different types of data, builds a model library, and calibrates the iterative prediction model based on real-time business data and market data.
[0018] In an optional embodiment, the prediction module includes a data acquisition preprocessing unit, a data acquisition unit, an AI model training unit, an AI model adjustment unit, and a prediction data output interface connected in sequence;
[0019] The data collection and preprocessing unit is used to collect and preprocess historical data and add it to the AI prediction training data set; the data acquisition unit is used to acquire real-time data and prediction data and add them to the AI prediction training data set; the AI model training unit trains and debugs the prediction model based on the AI prediction training data set; the AI model adjustment unit is used to adaptively adjust the prediction model; the prediction data output interface is used to output prediction data.
[0020] In an optional implementation, the data acquisition preprocessing unit is used to collect corresponding historical data for different prediction application scenarios, and store the historical data in a unified format. After collecting and storing the historical data, bad data is identified, eliminated, and corrected for the historical data, and added to the AI prediction training data set.
[0021] In an optional implementation, the scheduling optimization module aims to minimize the system operating cost and meet the system safe operation boundary. The optimization objective function of the scheduling optimization module is as follows:
[0022] min(A*(transformer / line losses+BESS operation cost+EV charging control cost)+B*∑(V_load–V_ub)+C*∑(V_lb–V_load)+D*load variation metric)
[0023] Among them, A, B, C, and D are adjustable weight coefficients, which are configured according to the regulation target; transformer / linelosses is the transmission line loss cost; V_ub and V_lb are the upper and lower limits of the distribution network voltage, respectively, and V_load is the distribution network load voltage; BESS operation cost is the battery charging and discharging cost; EV charging control cost is the cost of controlling / adjusting the electric vehicle charging load; Load variation metric is the load fluctuation evaluation indicator.
[0024] In an optional embodiment, the evaluation module performs the following steps:
[0025] 1) Calculate the maximum response capability using the distributed resource system and equipment capacity boundaries;
[0026] 2) Divide the maximum response capacity into response sequences with multiple response ratios according to the invitation requirements;
[0027] 3) The response capabilities corresponding to the response sequences are made into winning events one by one and input into the system operation strategy for operation optimization, and the total station revenue and response volume after responding to different response sequences are counted as the scenarios to be evaluated;
[0028] 4) The expected response volume is made into a winning bid event and input into the system operation strategy for operation optimization, and the total site revenue and response volume after the response are counted as the benchmark scenario.
[0029] 5) Compare the revenue differences between the scenarios to be evaluated and the benchmark scenarios one by one, divide the revenue differences by the response volume of the corresponding scenarios to be evaluated to obtain the unit cost of the response, and form the corresponding quantity-price sequence.
[0030] In an optional implementation, the profit calculation strategy of the profit calculation module is as follows:
[0031] 1) Real-time photovoltaic power generation power P pv ≤Real-time load power P load
[0032] In this case, all the power generation output is supplied to the station load, and all the power generation is consumed locally. The energy storage is determined according to the current strategic state, including:
[0033] ① Assume that the power corresponding to the energy storage strategy during this period is P BESS , P BESS ≤0 and P BESS +P pv ≤P load At this moment, the energy storage battery generates electricity to the grid, and the total power of power generation and energy storage discharge is less than the load power. Both power generation and energy storage discharge power are absorbed by the load:
[0034] Load absorbed photovoltaic power = P pv , cost (photovoltaic income) = P pv *t*Price grid ;
[0035] Load absorbed energy storage power = P BESS , cost (energy storage benefit) = P BESS *t*Price grid ;
[0036] Load absorbs grid power = P load -(P BESS +P pv )=P grid , cost = P grid *t*Price grid .
[0037] ②P BESS ≤0 and P BESS +P pv >P load At this moment, the energy storage battery generates electricity to the grid. The total power of photovoltaic power generation and energy storage discharge is greater than the load power. All photovoltaic power generation is absorbed by the load locally, and part of the energy storage discharge power is absorbed by the load, and part is returned to the grid:
[0038] Load absorbed photovoltaic power = P pv ; Cost (PV revenue) = P pv *t*Price grid ;
[0039] Load absorbed energy storage power = P load -P pv ; Cost (storage benefit) = (P load -P pv )*t*Price grid ;
[0040] Energy storage grid power = P BESS -(P load -P pv )=-P grid , profit = -P grid *Price -grid ;
[0041] ③P BESS >0, at this moment, the energy storage battery acts as a load, and all photovoltaic power generation power is absorbed by the load, and the insufficient part and energy storage charging power are supplemented by the grid:
[0042] Load absorbs photovoltaic power = P pv , cost (photovoltaic income) = P pv *t*Price grid ;
[0043] Load absorbs grid power = P load -P pv ; Cost = (P load -P pv )*t*Price grid ;
[0044] Energy storage absorbs grid power = P BESS , cost = P BESS *t*Price grid .
[0045] In an optional implementation, the profit calculation strategy of the profit calculation module is as follows:
[0046] 2) Real-time photovoltaic power generation power P pv >Real-time load power P load
[0047] In this case, the photovoltaic output not only meets the total charging power demand, but also has surplus. The surplus photovoltaic output is supplied to the energy storage battery for charging or to the grid. It needs to be determined according to the energy storage power status at this time, including:
[0048] ① Assume that the power corresponding to the energy storage strategy during this period is P BESS , and P BESS ≤0, at this moment, the energy storage battery generates electricity to the grid, and the surplus output of photovoltaic power and energy storage are both supplied to the grid. In addition to fully meeting the user's load demand, the surplus output of photovoltaic power is fully connected to the grid:
[0049] Load absorbs photovoltaic power = P load ; Cost (PV revenue) = P load *t*Price grid ;
[0050] Photovoltaic grid-connected power = P pv -P load ; Revenue = (P pv -P load )*Price -grid ;
[0051] Energy storage grid power = -P BESS , profit = P BESS *Price -grid ;
[0052] ②P BESS >0 and P load +P BESS ≥P pv At this moment, the energy storage battery serves as the load. In addition to supplying the load, the photovoltaic system also supplies the energy storage charging in the station. However, the surplus output of the photovoltaic system other than the load is less than the total demand for energy storage charging, and the grid needs to provide partial power support. That is, in this scenario, the photovoltaic system fully supplies the charging users in the station and the energy storage battery charging:
[0053] Load absorbed photovoltaic power = P load ; Cost (PV revenue) = P load *t*Price grid ;
[0054] Energy storage absorbs photovoltaic power = P pv -P load ; Cost (PV revenue) = (P pv -P load )*t*Price grid
[0055] Photovoltaic grid-connected power = 0;
[0056] Energy storage absorbs grid power = P BESS -(P pv -P load )=P grid , cost = P grid *Price grid ;
[0057] ③P BESS >0 and P load +P BESS <P pv At this moment, the energy storage battery acts as a load. In addition to supplying load, photovoltaic power also charges the energy storage battery in the station, and in addition to meeting the energy storage battery charging and load requirements, there is also surplus output to connect to the grid. At this time, the energy storage battery and the charging user are the net power load:
[0058] Photovoltaic power absorbed by load = P load ; Cost (PV revenue) = P load *t*Price grid ;
[0059] Energy storage absorbs photovoltaic power = P BESS ; Cost (PV revenue) = P BESS *Price grid ;
[0060] Photovoltaic grid-connected power = P pv -P load -P BESS =-P gri . Profit = -P grid *Price -grid .
[0061] In a second aspect, the present invention provides a distributed resource intelligent optimization and control terminal, the terminal includes a processor and a display, the processor is integrated with a functional software module, the functional software module includes a prediction module, a scheduling optimization module, an evaluation module and a benefit calculation module;
[0062] The prediction module is used to output prediction data based on artificial intelligence, and the prediction data serves as the input and boundary condition of the scheduling optimization module; the scheduling optimization module generates optimization control strategies based on the prediction data and transmits them to the benefit calculation module; the evaluation module is used to change the input and target of the scheduling optimization module, so that the scheduling optimization module generates the optimal strategy under different evaluation conditions and transmits it to the benefit calculation module; the benefit calculation module is used to perform benefit analysis and calculation based on the strategy generated by the scheduling optimization module.
[0063] The distributed resource intelligent optimization and control system and terminal provided by the embodiments of the present invention have the following beneficial effects:
[0064] Through artificial intelligence prediction algorithms, high-precision predictions of distributed power generation and power load are provided, providing boundaries and references for the system's strategy generation and optimized operation, scientifically evaluating distributed power generation, rationally planning adjustable resources such as energy storage, and realizing local consumption and maximum utilization of distributed power generation. Through the distributed resource intelligent optimization and control system, the distributed resource equipment in the system is coordinated to carry out multi-time scale operation strategy generation, guide the system to optimize operation, achieve multi-system goals such as economic regulation, safety management, and demand response, carry out unified operation management, improve system operation efficiency, and combine the system and equipment operation costs to carry out scientific evaluation of the system's multi-time scale control capabilities and control costs, and feedback the system control response quantity and price evaluation pair for market declaration and resource selection on the cloud platform to support the control needs of the upper market and power grid. While generating multi-objective operation strategies, the intelligent optimization and control system can statistically analyze the operating costs and benefits of each stakeholder in the system based on the detailed operation data of the distributed resource system, combined with the marginal cost and price generation principle, and provide benefit analysis reports and inquiries to realize the statistical calculation of benefits of multiple stakeholders in the system. BRIEF DESCRIPTION OF THE DRAWINGS
[0065] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.
[0066] Figure 1 A schematic diagram of the composition of a distributed resource intelligent optimization and control system provided by an embodiment of the present invention;
[0067] Figure 2 It is the specific execution process of the evaluation module.
[0068] Icons: 1- Forecasting module; 2- Scheduling optimization module; 3- Evaluation module; 4- Benefit estimation module. DETAILED DESCRIPTION
[0069] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.
[0070] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. 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.
[0071] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.
[0072] It should be noted that, in the absence of conflict, the features in the embodiments of the present invention may be combined with each other.
[0073] Please refer to Figure 1 This embodiment provides a distributed resource intelligent optimization and control system (hereinafter referred to as: system), the system includes a prediction module 1, a scheduling optimization module 2, an evaluation module 3 and a benefit calculation module 4, wherein the prediction module 1, the scheduling optimization module 2 and the benefit calculation module 4 are connected in sequence, and the scheduling optimization module 2 and the benefit calculation module 4 are both connected to the evaluation module 3.
[0074] The prediction module 1 is used to output prediction data based on artificial intelligence as the input and boundary conditions of the scheduling optimization module 2. The scheduling optimization module 2 generates an optimization control strategy based on the prediction data and transmits it to the benefit calculation module 4. This optimization control strategy is the optimal control strategy based on the system operation target. If there is no change in the cloud or other operation targets, it will be operated as the optimal strategy. The evaluation module 3 is used to change the input and target of the scheduling optimization module 2, so that the scheduling optimization module 2 generates the optimal strategy under different evaluation conditions and transmits it to the benefit calculation module 4. The benefit calculation module 4 is used to perform benefit analysis and calculation based on the strategy generated by the scheduling optimization module 2 (including the optimization control strategy generated based on the prediction data and the optimal strategy under different evaluation conditions).
[0075] Specifically, the forecast data includes the forecasted power generation and load, and the forecast application scenarios of the forecast module 1 include charging load, production and living load, photovoltaic power generation, wind power, electric energy, etc. The time scale of the forecast service may include 15 minutes ago, 1 hour ago, 4 hours ago, 24 hours ago, 72 hours ago, and medium and long term.
[0076] Among them, the artificial intelligence prediction algorithm used in the prediction module 1 is based on advanced artificial intelligence algorithms, including deep neural network (DNN), long short-term memory network (LSTM), convolutional neural network (CNN), support vector machine (SVM) algorithm, etc., which can provide multi-time and spatial dimension prediction services for different business types. By building a reasonable artificial intelligence model architecture, collecting historical and real-time business data, geographic information system, historical weather information and forecast information, historical and real-time market information and other information to form effective data features (features), deeply mining massive historical data and optimizing the association between different types of data, and forming an effective training data set for different prediction services to train, debug and improve the prediction accuracy of the model. Thus, the charging load of the station and the region, other production and life loads, photovoltaic power generation, electric vehicle operation and other scenarios are predicted. The time scale of the prediction service may include 15 minutes ago, 1 hour ago, 4 hours ago, 24 hours ago, 72 hours ago, and medium and long term. The forecast data generated by the artificial intelligence algorithm provides input for other functional services. The scheduling optimization module 2 generates scheduling strategies of different time scales based on the forecast data. The forecast data is used as a reference in the generation of the scheduling strategy and the forecast data is used as a boundary condition. For example, in the energy storage control strategy at a certain moment in the future, the energy storage charging and discharging power cannot exceed the limits of power generation, load forecasting and power supply capacity of the power grid.
[0077] For different prediction application scenarios, by building a reasonable prediction model architecture, collecting massive historical data (including electrical and non-electrical data), forming effective data feature samples, and building a training data set and a sample library, we deeply mine the data in the sample library, optimize the relationship between different types of data, build a model library, and calibrate the iterative model based on real-time business data and market data to improve the prediction accuracy of the AI model.
[0078] In this embodiment, the advanced artificial intelligence algorithm uses a deep neural network (English name: Deep Neural Networks, abbreviated as: DNN). DNN is a discriminant model that can be trained using a back-propagation algorithm.
[0079] The learning process of a deep neural network is the process of minimizing a cost function, which usually includes an error term and a correction term. The error term is used to evaluate how well a neural network fits a data set. At the same time, the correction term prevents overfitting by controlling complexity.
[0080] The prediction module 1 has the advantages of strong adaptability and portability, and can be installed and deployed on the local site energy management edge controller and the cloud energy network system platform. The prediction data generated by the prediction module 1 can provide input for other functional services and has an independent display function for use by system operation and dispatch personnel, operation and maintenance personnel, and market trading personnel.
[0081] Specifically, the prediction module 1 includes a data collection preprocessing unit, a data acquisition unit, an AI model training unit, an AI model adjustment unit and a prediction data output interface which are connected in sequence.
[0082] The data collection and preprocessing unit is used to collect and preprocess historical data. Specifically, the data collection and preprocessing unit is used to collect corresponding historical data for different prediction application scenarios and store the historical data in a unified format. After collecting and storing the historical data, the bad data is identified, eliminated, and corrected, and added to the AI prediction training data set.
[0083] The data acquisition unit is used to acquire real-time data and predicted data. Specifically, real-time data includes real-time power data, meteorological data, etc., and predicted data includes weather forecast features, time features, etc. The data acquisition unit is used to configure the corresponding data acquisition interface according to different prediction application scenarios, collect real-time data and predicted data, and add them to the AI prediction training data set.
[0084] The AI model training unit trains and debugs AI models (prediction models) based on the AI prediction training data set. Specifically, the AI model training unit is used to use advanced artificial intelligence algorithms, especially supervised learning algorithms, to train a variety of AI intelligent prediction models for multiple time and space scales. This functional module will use multi-threaded training and hyperparameter adjustment to ensure the prediction accuracy requirements.
[0085] The AI model adjustment unit is used to adaptively adjust the AI model. Specifically, the AI model adjustment unit is used to regularly update the AI prediction training data set to regularly train and improve the AI model online to ensure that the prediction service is effective and reliable in the long term. The update frequency of the prediction model can be specified by the user, such as every day, every week, every two weeks, etc.
[0086] The prediction data output interface is used to output the prediction data. Specifically, the required data interface is configured according to the needs of various application scenarios to ensure seamless connection between the prediction results and subsequent functions.
[0087] The scheduling optimization module 2 is used to implement hybrid multi-objective optimization based on genetic algorithm and linear programming. Among them, the hybrid multi-objective optimization technology based on genetic algorithm and linear programming: the optimal scheduling of distributed resource systems and other smart loads is a complex nonlinear problem. Genetic algorithms are applied to find a set of optimal solutions for energy storage, scheduling of smart loads, and schedules for load management, and then these solutions are used as the starting point for linear programming to obtain a solution that is closest to the global optimal value. Compared with pure analytical optimization methods, this algorithm is more applicable in adapting to a variety of cost models for distributed resources.
[0088] In the daily operation of the system, the hybrid multi-objective optimization engine based on genetic algorithm and linear programming is used to take advantage of the distributed resource scheduling technology within the hour. Based on the forecast values of the day-ahead charging load, photovoltaic output, other loads, market information, etc., a day-ahead operation plan is formulated to determine the operation plan for each hour of the next 24 hours for adjustable resources such as battery energy storage, charging piles and smart loads. During the operation day, the battery energy storage, charging piles and smart loads are adjusted in real time according to the forecast values of real-time charging load, photovoltaic output, other loads and market information, so that the operation indicators of the station microgrid system fit the optimal solution and achieve the preset operation goals.
[0089] The optimization algorithm uses a joint hybrid local optimization engine: according to different needs, economic operation and safe operation are selected as joint optimization goals. In the case of inconsistency in the optimization of multiple objectives, a strategy based on Pareto efficiency is selected. In view of nonlinear factors such as battery capacity attenuation, charging and discharging losses, and system line losses in the optimization problem, the simulation software uses a genetic algorithm that is insensitive to nonlinearity to obtain multiple feasible and suboptimal solutions for energy storage and orderly charging scheduling, and then uses this set of solutions as the initial value of the linear programming algorithm to continue to obtain the global optimal solution.
[0090] In this embodiment, the optimization objective function of the scheduling optimization module 2 is as follows:
[0091] min(A*(transformer / line losses+BESS operation cost+EV charging control cost)+B*∑(V_load-V_ub)+C*∑(V_lb-V_load)+D*load variation metric)
[0092] Among them, A, B, C, and D are adjustable weight coefficients, which are configured according to adjustment goals such as "minimum operating cost", "minimum peak-to-valley difference", and "charging demand priority";
[0093] transformer / line losses are transmission line loss costs;
[0094] V_ub and V_lb are the upper and lower limits of the distribution network voltage respectively, and V_load is the distribution network load voltage, which is obtained by the distribution network model state estimation; when there is no distribution network model, both B and C can be zero values;
[0095] BESS operation cost is the cost of charging and discharging the battery (depending on electricity price, losses, charging and discharging time and power);
[0096] EV charging control cost is the cost of controlling / adjusting the EV charging load (determined by factors such as electricity prices and user quotes);
[0097] Load variation metric is an evaluation index of load fluctuation, such as the difference between load peak and valley;
[0098] The scheduling optimization module 2 aims to minimize the system operation cost and meet the system safe operation boundary. The system operation cost can be divided into transmission line loss cost, BESS charging and discharging cost, EV or load charging and electricity cost. The cost is modeled according to the actual situation of the system, which is reflected in item A. The safe operation constraint needs to consider the impact of the upper and lower limits of the system voltage on the system operation, which is reflected in items B and C. At the same time, according to the system operation requirements, the load fluctuation requirements and the system fluctuation smoothing operation goals are considered, which is reflected in item D.
[0099] Charging pile power control, battery charging and discharging power control, system voltage control (PCS reactive power) and topology control (load segmentation switch) are all control variables that determine the objective function value.
[0100] Evaluation module 3 is used to evaluate the regulation potential and regulation costs. Figure 2 ,The specific execution steps of evaluation module 3 are as follows:
[0101] S1: Calculate the maximum response capability using the distributed resource system and device capacity boundaries;
[0102] S2: Divide the maximum response capacity into a response sequence with multiple response ratios according to the invitation requirements;
[0103] S3: The response capabilities corresponding to the response sequences are made into winning events one by one and input into the system operation strategy for operation optimization, and the total station revenue and response volume after responding to different response sequences are counted as the scenarios to be evaluated;
[0104] S4: The expected response volume is made into a winning event and input into the system operation strategy for operation optimization, and the total site revenue and response volume after the response are counted as the benchmark scenario.
[0105] S5: Compare the revenue differences between the scenarios to be evaluated and the benchmark scenarios one by one, divide the revenue differences by the response volume of the corresponding scenarios to be evaluated to obtain the unit cost of the response, and form the corresponding quantity-price sequence.
[0106] The revenue calculation module 4 is used to calculate the revenue based on the collected basic power data. Specifically, the basic power data refers to the basic data for the revenue calculation, and the basic power data includes:
[0107] (1) Real-time load power P load (including charging, equipment standby, loss and other loads, etc.), power consumption is positive, unit: kW;
[0108] (2) Real-time photovoltaic power generation power P pv (Based on the photovoltaic ECU data, the AC side grid-connected point power data), power generation is positive, unit: kW;
[0109] (3) Real-time energy storage battery charging and discharging power P BESS (Based on the node power on the AC side of the energy storage PCS), charging is positive, discharging is negative, unit: kW;
[0110] (4) Real-time grid power P grid (Based on the total meter data on the low-voltage side of the transformer), without considering transformer losses, power consumption is positive, and return is negative, unit: kW;
[0111] (5) Grid electricity price grid , return the grid electricity price Price -grid .
[0112] In various cases, the profit calculation strategy is as follows:
[0113] 1)P pv ≤P load
[0114] In this case, all power generation output is supplied to the station load, all power generation is consumed locally, and energy storage is determined based on the strategy status at this time.
[0115] ① Assume that the power corresponding to the energy storage strategy during this period is P BESS , P BESS ≤0 and P BESS +P pv ≤P load At this moment, the energy storage battery is also generating electricity to the grid. The total power of power generation and energy storage discharge is less than the load power, and both power generation and energy storage discharge are absorbed by the load:
[0116] Load absorbs photovoltaic power = P pv , cost (photovoltaic income) = P pv *t*Price grid ;
[0117] Load absorbed energy storage power = P BESS , cost (energy storage benefit) = P BESS *t*Price grid ;
[0118] Load absorbs grid power = P load -(P BESS +P pv )=P grid , cost = P grid *t*Price grid .
[0119] ②P BESS ≤0 and P BESS +P pv >P load At this moment, the energy storage battery is also generating electricity to the grid. The total power of photovoltaic power generation and energy storage discharge is greater than the load power. All photovoltaic power generation is absorbed by the load locally, and part of the energy storage discharge power is absorbed by the load, and part is returned to the grid:
[0120] Load absorbs photovoltaic power = P pv ; Cost (PV revenue) = P pv *t*Price grid ;
[0121] Load absorbed energy storage power = P load -P pv ; Cost (storage benefit) = (P load -P pv )*t*Price grid ;
[0122] Energy storage grid power = P BESS -(P load -P pv )=-P grid , profit = -P grid *Price -grid ;
[0123] ③P BESS >0, at this moment, the energy storage battery acts as a load, and all photovoltaic power generation power is absorbed by the load, and the insufficient part and energy storage charging power are supplemented by the grid:
[0124] Load absorbs photovoltaic power = P pv , cost (photovoltaic income) = P pv *t*Price grid ;
[0125] Load absorbs grid power = P load -P pv ; Cost = (Pload -P pv )*t*Price grid ;
[0126] Energy storage absorbs grid power = P BESS , cost = P BESS *t*Price grid ;
[0127] 2) P pv >P load
[0128] In this case, in addition to meeting the total demand for charging power, the photovoltaic output has surplus. In order to promote photovoltaic consumption, the surplus photovoltaic output needs to be supplied to energy storage batteries for charging or to the grid, which needs to be determined based on the energy storage power status at that time.
[0129] ① Assume that the power corresponding to the energy storage strategy during this period is P BESS , and P BESS ≤0, at this moment, the energy storage battery is also generating electricity to the grid, that is, the surplus output of photovoltaic power and energy storage are both supplying power to the grid. In addition to fully meeting the user's load demand, the surplus output of photovoltaic power is fully connected to the grid:
[0130] Load absorbs photovoltaic power = P load ; Cost (PV revenue) = P load *t*Price grid ;
[0131] Photovoltaic grid-connected power = P pv -P load ; Revenue = (P pv -P load )*Price -grid ;
[0132] Energy storage grid power = -P BESS , profit = P BESS *Price -grid ;
[0133] ②P BESS >0 and P load +P BESS ≥P pv At this moment, the energy storage battery serves as the load. In addition to supplying the load, the photovoltaic system also supplies the energy storage charging in the station. However, the surplus output of the photovoltaic system other than the load is less than the total demand for energy storage charging, and the grid needs to provide partial power support. That is, in this scenario, the photovoltaic system fully supplies the charging users in the station and the energy storage battery charging:
[0134] Load absorbs photovoltaic power = P load ; Cost (PV revenue) = P load *t*Pricegrid ;
[0135] Energy storage absorbs photovoltaic power = P pv -P load ; Cost (PV revenue) = (P pv -P load )*t*Price grid
[0136] Photovoltaic grid-connected power = 0;
[0137] Energy storage absorbs grid power = P BESS -(P pv -P load )=P grid , cost = P grid *Price grid ;
[0138] ③P BESS >0 and P load +P BESS <P pv At this moment, the energy storage battery acts as a load. In addition to supplying load, photovoltaic power also charges the energy storage battery in the station, and in addition to meeting the energy storage battery charging and load requirements, there is also surplus output to connect to the grid. At this time, the energy storage battery and the charging user are the net power load:
[0139] Photovoltaic power absorbed by load = P load ; Cost (PV revenue) = P load *t*Price grid ;
[0140] Energy storage absorbs photovoltaic power = P BESS ; Cost (PV revenue) = P BESS *Price grid ;
[0141] Photovoltaic grid-connected power = P pv -P load -P BE =-P grid . Profit = -P grid *Price -grid .
[0142] This embodiment also provides a distributed resource intelligent optimization and control terminal, the terminal includes a processor and a display, the processor as an edge computing unit, can be an embedded industrial computer, the processor is integrated with functional software modules, the functional software modules include prediction module 1, scheduling optimization module 2, evaluation module 3 and benefit calculation module 4, so that the processor can execute the functional steps of prediction module 1, scheduling optimization module 2, evaluation module 3 and benefit calculation module 4. The display is used to display the prediction data of prediction module 1, the final system operation strategy of scheduling optimization module 2, the evaluation results of evaluation module 3 and the benefit analysis calculation results of benefit calculation module 4. Of course, the terminal will also include a conventional communication module and a power module.
[0143] In this way, by collecting basic power data at high frequency, basic settlement data is formed to ensure the real-time and accuracy of the data. Based on the collected data, settlement statements and conclusion reports are generated to provide detailed electricity charges and energy usage information to all relevant parties.
[0144] Through artificial intelligence prediction algorithms, high-precision predictions of distributed power generation and power load are provided, providing boundaries and references for the system's strategy generation and optimized operation, scientifically evaluating distributed power generation, rationally planning adjustable resources such as energy storage, and realizing local consumption and maximum utilization of distributed power generation. Through the distributed resource intelligent optimization and control system, the distributed resource equipment in the system is coordinated to carry out multi-time scale operation strategy generation, guide the system to optimize operation, achieve multi-system goals such as economic regulation, safety management, and demand response, carry out unified operation management, improve system operation efficiency, and combine the system and equipment operation costs to carry out scientific evaluation of the system's multi-time scale control capabilities and control costs, and feedback the system control response quantity and price evaluation pair for market declaration and resource selection on the cloud platform to support the control needs of the upper market and power grid. While generating multi-objective operation strategies, the intelligent optimization and control system can statistically analyze the operating costs and benefits of each stakeholder in the system based on the detailed operation data of the distributed resource system, combined with the marginal cost and price generation principle, and provide benefit analysis reports and inquiries to realize the statistical calculation of benefits of multiple stakeholders in the system.
[0145] In addition, the system supports online report viewing and data export, which is convenient for users to obtain and analyze data at any time. Design reports in standard formats to display detailed electricity consumption, electricity prices and their corresponding costs and benefits. Configure and display the geographical location, number, name and resources included in the site, provide comprehensive site information to meet the needs of different sites. Reports and statements will show in detail the settlement results between the property and the power grid, load, photovoltaic, energy storage and other related parties, helping all parties to clearly understand their own electricity costs and benefits, and comprehensively display the electricity bill settlement and investment benefits of all related parties through indicators such as time-sharing electricity consumption, unit cost, and unit income.
[0146] The system can be used to assist in the settlement of electricity bills for multiple stakeholders and the analysis of the investment benefits of photovoltaic, energy storage and other facilities. Through the comprehensive management of the system, it can effectively improve the management efficiency and transparency of the park, provide accurate data support and decision-making basis for all stakeholders, and help achieve efficient use of energy and cost optimization. Provide accurate electricity bill settlement services to help all stakeholders in the park clarify their respective electricity bill expenditures and benefits. Through detailed reports and data analysis, evaluate the investment benefits of photovoltaic, energy storage and other facilities, and help users make wise investment decisions.
[0147] The revenue calculation module 4 also adopts the marginal price mechanism of the high-low matching method. First, the highest buyer's charging price bid and the seller's lowest discharge price bid are compared. If the buyer's price bid is higher than the seller's price bid, the transaction is matched. Then, in the remaining unmatched buying and selling prices, the transaction is matched in the same way as above until there is no comparable bid price or the highest buyer's price is lower than the lowest seller's price. The average of the buyer's bid and the seller's bid in the last group of matching transactions is the market clearing price, and the settlement price of the power supply price of the power grid is the normal electricity price. The settlement price of distributed resources such as photovoltaics, loads and energy storage in the system is the market clearing price.
[0148] The distributed resource intelligent optimization and control system and terminal provided by the embodiments of the present invention have the following beneficial effects:
[0149] 1. Through artificial intelligence prediction algorithms, high-precision predictions of distributed power generation and power load are provided, providing boundaries and references for system strategy generation and optimized operation;
[0150] 2. Through high-precision forecasting data, scientifically evaluate the distributed generation power, rationally plan adjustable resources such as energy storage, and realize local consumption and maximum utilization of distributed generation;
[0151] 3. Through the Internet of Things communication technology and intelligent optimization and control systems and terminals, a distributed resource local management system is built to coordinate the distributed resource equipment in the system, carry out unified operation management, break the problem of independent operation of equipment, and improve the system operation efficiency;
[0152] 4. The intelligent optimization and control system integrates a multi-objective hybrid optimization engine, which can generate multi-time scale operation strategies, guide the system to optimize operation, and achieve multi-system goals such as economic control, safety management, and demand response;
[0153] 5. While generating multi-objective operation strategies, the intelligent optimization and control system can conduct scientific evaluation of the system's multi-time scale control capabilities and control costs in combination with the system and equipment operation costs, and provide feedback on the system's control response quantity and price evaluation, which is used for market declaration and resource selection on the cloud platform to support the control needs of the upper-level market and power grid;
[0154] 6. The intelligent optimization and control system can statistically analyze the operating costs and revenues of each stakeholder in the system during each period based on the detailed operating data of the distributed resource system and combined with the marginal cost and price generation principles, and provide revenue analysis reports and inquiries to realize the statistical calculation of the revenues of multiple stakeholders in the system.
[0155] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims
1. A distributed resource intelligent optimization and control system, characterized in that: The system comprises a prediction module (1), a scheduling optimization module (2), an evaluation module (3) and a revenue calculation module (4), wherein the prediction module (1), the scheduling optimization module (2) and the revenue calculation module (4) are connected in sequence, and the scheduling optimization module (2) and the revenue calculation module (4) are both connected to the evaluation module (3); The prediction module (1) is used to output prediction data based on artificial intelligence, and the prediction data serves as the input and boundary conditions of the scheduling optimization module (2); the scheduling optimization module (2) generates an optimization control strategy based on the prediction data, and transmits it to the benefit calculation module (4); the evaluation module (3) is used to change the input and target of the scheduling optimization module (2), so that the scheduling optimization module (2) generates the optimal strategy under different evaluation conditions, and transmits it to the benefit calculation module (4); the benefit calculation module (4) is used to perform benefit analysis and calculation based on the strategy generated by the scheduling optimization module (2).
2. The distributed resource intelligent optimization and control system according to claim 1 is characterized in that: The prediction module (1) uses a deep neural network to build a prediction model, collects historical data, forms data feature samples, and constructs a training data set and a sample library, mines the data in the sample library, optimizes the association between different types of data, builds a model library, and calibrates the iterative prediction model based on real-time business data and market data.
3. The distributed resource intelligent optimization and control system according to claim 1 is characterized in that: The prediction module (1) comprises a data collection preprocessing unit, a data acquisition unit, an AI model training unit, an AI model adjustment unit and a prediction data output interface which are connected in sequence; The data acquisition and preprocessing unit is used to collect and preprocess historical data and add it to the AI prediction training data set; the data acquisition unit is used to acquire real-time data and prediction data and add them to the AI prediction training data set; the AI model training unit trains and debugs the prediction model based on the AI prediction training data set; the AI model adjustment unit is used to adaptively adjust the prediction model; the prediction data output interface is used to output prediction data.
4. The distributed resource intelligent optimization and control system according to claim 3 is characterized in that: The data collection and preprocessing unit is used to collect corresponding historical data for different prediction application scenarios and store the historical data in a unified format. After collecting and storing the historical data, bad data is identified, eliminated, and corrected for the historical data, and added to the AI prediction training data set.
5. The distributed resource intelligent optimization and control system according to claim 1 is characterized in that: The scheduling optimization module (2) aims to minimize the system operation cost and meet the system safety operation boundary. The optimization objective function of the scheduling optimization module (2) is as follows: min(A*(transformer / line losses+BESS operation cost+EV charging control cost)+B*∑(V_load-V_ub)+C*∑(V_lb-V_load)+D*load variation metric) Among them, A, B, C, and D are adjustable weight coefficients, which are configured according to the regulation target; transformer / line losses are transmission line loss costs; V_ub and V_lb are the upper and lower limits of the distribution network voltage, respectively, and V_load is the distribution network load voltage; BESSoperation cost is the battery charging and discharging cost; EV charging control cost is the cost of controlling / adjusting the electric vehicle charging load; Load variation metric is an evaluation indicator of load fluctuation.
6. The distributed resource intelligent optimization and control system according to claim 1 is characterized in that: The evaluation module (3) performs the following steps: S1: Calculate the maximum response capability using the distributed resource system and device capacity boundaries; S2: Divide the maximum response capacity into a response sequence with multiple response ratios according to the invitation requirements; S3: The response capabilities corresponding to the response sequences are made into winning events one by one and input into the system operation strategy for operation optimization, and the total station revenue and response volume after responding to different response sequences are counted as the scenarios to be evaluated; S4: The expected response volume is made into a winning bid event and input into the system operation strategy for operation optimization, and the total station revenue and response volume after the response are counted as the benchmark scenario; S5: Compare the revenue differences between the scenarios to be evaluated and the benchmark scenarios one by one, divide the revenue differences by the response volume of the corresponding scenarios to be evaluated to obtain the unit cost of the response, and form the corresponding quantity-price sequence.
7. The distributed resource intelligent optimization and control system according to claim 1 is characterized in that: The profit calculation strategy of the profit calculation module (4) is as follows: 1) Real-time photovoltaic power generation power P pv ≤Real-time load power P load In this case, all the power generation output is supplied to the station load, and all the power generation is consumed locally. The energy storage is determined according to the current strategic state, including: ① Assume that the power corresponding to the energy storage strategy during this period is P BESS , P BESS ≤0 and P BESS +P pv ≤P load At this moment, the energy storage battery generates electricity to the grid, and the total power of power generation and energy storage discharge is less than the load power. Both power generation and energy storage discharge power are absorbed by the load: Load absorbed photovoltaic power = P pv , cost (photovoltaic income) = P pv *t*Price grid ; Load absorbed energy storage power = P BESS , cost (energy storage benefit) = P BESS *t*Price grid ; Load absorbs grid power = P load -(P BESS +P pv )=P grid , cost = P grid *t*Price grid . ②P BESS ≤0 and P BESS +P pv >P load At this moment, the energy storage battery generates electricity to the grid. The total power of photovoltaic power generation and energy storage discharge is greater than the load power. All photovoltaic power generation is absorbed by the load locally, and part of the energy storage discharge power is absorbed by the load, and part is returned to the grid: Load absorbed photovoltaic power = P pv ; Cost (PV revenue) = P pv *t*Price grid ; Load absorbed energy storage power = P load -P pv ; Cost (storage benefit) = (P load -P pv )*t*Price grid ; Energy storage grid power = P BESS -(P load -P pv )=-P grid , profit = -P grid *Price -grid ; ③P BESS >0, at this moment, the energy storage battery acts as a load, and all photovoltaic power generation power is absorbed by the load, and the insufficient part and energy storage charging power are supplemented by the grid: Load absorbed photovoltaic power = P pv , cost (photovoltaic income) = P pv *t*Price grid ; Load absorbs grid power = P load -P pv ; Cost = (P load -P pv )*t*Price grid ; Energy storage absorbs grid power = P BESS , cost = P BESS *t*Price grid .
8. The distributed resource intelligent optimization and control system according to claim 7 is characterized in that: The profit calculation strategy of the profit calculation module (4) is as follows: 2) Real-time photovoltaic power generation power P pv >Real-time load power P load In this case, the photovoltaic output not only meets the total charging power demand, but also has surplus. The surplus photovoltaic output is supplied to the energy storage battery for charging or to the grid. It needs to be determined according to the energy storage power status at this time, including: ① Assume that the power corresponding to the energy storage strategy during this period is P BESS , and P BESS ≤0, at this moment, the energy storage battery generates electricity to the grid, and the surplus output of photovoltaic power and energy storage are both supplied to the grid. In addition to fully meeting the user's load demand, the surplus output of photovoltaic power is fully connected to the grid: Load absorbed photovoltaic power = P load ; Cost (PV revenue) = P load *t*Price grid ; Photovoltaic grid-connected power = P pv -P load ; Revenue = (P pv -P load )*Price -grid ; Energy storage grid power = -P BESS , profit = P BESS *Price -grid ; ②P BESS >0 and P load +P BESS ≥P pv At this moment, the energy storage battery serves as the load. In addition to supplying the load, the photovoltaic system also supplies the energy storage charging in the station. However, the surplus output of the photovoltaic system other than the load is less than the total demand for energy storage charging, and the grid needs to provide partial power support. That is, in this scenario, the photovoltaic system fully supplies the charging users in the station and the energy storage battery charging: Load absorbed photovoltaic power = P load ; Cost (PV revenue) = P load *t*Price grid ; Energy storage absorbs photovoltaic power = P pv -P load ; Cost (PV revenue) = (P pv -P load )*t*Price grid Photovoltaic grid-connected power = 0; Energy storage absorbs grid power = P BESS -(P pv -P load )=P grid , cost = P grid *Price grid ; ③P BESS >0 and P load +P BESS <P pv At this moment, the energy storage battery acts as a load. In addition to supplying load, photovoltaic power also charges the energy storage battery in the station, and in addition to meeting the energy storage battery charging and load requirements, there is also surplus output to connect to the grid. At this time, the energy storage battery and the charging user are the net power load: Photovoltaic power absorbed by load = P load ; Cost (PV revenue) = P load *t*Price grid ; Energy storage absorbs photovoltaic power = P BESS ; Cost (PV revenue) = P BESS *Price grid ; Photovoltaic grid-connected power = P pv -P load -P BESS =-P grid ; Profit = -P grid *Price -grid .
9. A distributed resource intelligent optimization and control terminal, characterized in that: The terminal comprises a processor and a display, wherein the processor is integrated with a functional software module, wherein the functional software module comprises a prediction module (1), a scheduling optimization module (2), an evaluation module (3) and a benefit estimation module (4); The prediction module (1) is used to output prediction data based on artificial intelligence, and the prediction data serves as the input and boundary conditions of the scheduling optimization module (2); the scheduling optimization module (2) generates an optimization control strategy based on the prediction data, and transmits it to the benefit calculation module (4); the evaluation module (3) is used to change the input and target of the scheduling optimization module (2), so that the scheduling optimization module (2) generates the optimal strategy under different evaluation conditions, and transmits it to the benefit calculation module (4); the benefit calculation module (4) is used to perform benefit analysis and calculation based on the strategy generated by the scheduling optimization module (2).
Citation Information
Cited By
Cloud side-end collaborative microgrid real-time simulation test platform and method
CN121580558A
