Domain-based intelligent assessment method for safe and economic operation margin of power grid

Abstract

The invention discloses a domain-based power grid safety economic operation margin intelligent evaluation method, and belongs to the technical field of combination of a power system and artificial intelligence. Comprising the following steps: learning new energy output data, load data and meteorological observation data by using an artificial intelligence algorithm, and predicting a source and load node fluctuation range in the next 24 hours; based on source and load node prediction results, respectively constructing an economic operation domain and a safe operation domain as a power grid economic dispatching plan set and a safe operation interval in the next 24 hours; integrating the power grid economic dispatching plan set and the safe operation interval to obtain a day-ahead dispatching plan set meeting the safe and economic operation requirements of the power grid; and considering the uncertainty of source and load node prediction, and finally obtaining the safe and economic operation margin of the power grid. The method can provide day-ahead scheduling decision support for a power grid, thereby supporting safe and economical operation of a novel power system.

Description

Technical Field

[0001] The present invention belongs to the technical field of combining power systems with artificial intelligence, and specifically relates to a domain-based intelligent evaluation method for safe and economical operation margin of power grids. Background Art

[0002] As the proportion of renewable energy in the power grid gradually increases, its volatility and uncertainty have brought new challenges to grid dispatching. Traditional grid dispatching methods mainly rely on load forecasting and generator dispatching, but it is difficult to cope with the fluctuations of renewable energy generation and changes in load demand. The uncertainty of renewable energy, the real-time volatility of loads and the influence of various real-time factors have reduced the reliability of grid dispatching, and traditional methods are difficult to ensure the safe and stable operation of the power grid.

[0003] At present, there are some grid security and economic operation assessment schemes based on optimization models, probability theory, and robust optimization, but they still have limitations such as complex calculations and poor real-time response capabilities when facing problems such as fluctuations in renewable energy output and uncertainty in load demand. By introducing advanced technologies such as artificial intelligence, it is possible to efficiently predict renewable energy generation, load demand, and meteorological conditions, and improve the accuracy and real-time performance of scheduling, but it is still necessary to quantify the safety margin of the grid to ensure system stability.

[0004] Faced with a large number of uncertain factors, how to accurately evaluate the safety margin of the power grid based on artificial intelligence technology and dynamically adjust the dispatching decision on this basis is still a major challenge in the safe and economic operation of the power grid. Therefore, an intelligent evaluation method is urgently needed to solve the problem of balancing power grid safety and economy.

[0005] Patent document CN117973881A discloses a low-carbon economic operation analysis method for an electric-gas-hydrogen integrated energy system based on domain theory. By setting the fixed carbon emission operation domain and the economic fixed carbon emission operation domain of the electric-gas-hydrogen system, considering the uncertainty of hydrogen energy flexibility and renewable energy, a corresponding system model is established, and the uncertainty factors are reconstructed and simplified into a deterministic mathematical model. The boundary of the operation domain is constructed through point-by-point simulation, thereby evaluating the feasibility of low-carbon operation of the system. This method uses traditional optimization methods, which have high computational complexity and long computational time.

[0006] Patent document CN118630735A discloses a low-carbon economic dispatch method for hydrogen-containing virtual power plants based on deep reinforcement learning. First, a mathematical model of the virtual power plant is established, covering electrical, thermal and hydrogen energy equipment; then a low-carbon economic dispatch model is constructed and trained using a deep reinforcement learning algorithm. By setting appropriate states, actions and reward functions, the intelligent agent is trained to make dispatch decisions. This method has no constraints on safe operation, and there are certain safety risks in the power grid.

[0007] Traditional grid dispatching methods rely on static models and cannot fully cope with the volatility of renewable energy and real-time changes in load demand, resulting in a mismatch between dispatching plans and actual conditions, reducing grid safety and economy. The lack of a dynamic adjustment mechanism means that emergencies cannot be responded to in a timely manner, which can easily lead to overload and instability. In addition, traditional methods ignore the characteristics of renewable energy generation and fail to effectively integrate real-time meteorological data with load forecasts, limiting the accuracy and flexibility of grid dispatching and failing to meet the grid operation requirements dominated by renewable energy. Summary of the invention

[0008] In view of the above-mentioned technical problems existing in the prior art, the present invention provides a domain-based intelligent evaluation method for the safe and economical operation margin of a power grid, which can ensure the safe and economical operation of a new power system while ensuring the operating speed.

[0009] To achieve the above-mentioned purpose of the invention, an embodiment provides a domain-based intelligent evaluation method for safe and economic operation margin of a power grid, comprising the following steps:

[0010] The system uses real-time monitoring and data collection technology to obtain data from different sources, including new energy output data, load demand data and meteorological observation data. New energy output data comes from various renewable energy power generation units, such as solar energy, wind energy, geothermal energy, etc., reflecting the power output of various new energy sources; load data covers the real-time and predicted power demand of each node in the power grid; meteorological observation data includes temperature, humidity, wind speed, radiation and other meteorological information related to power production and demand. By combining historical data, recent data and real-time data, the system can accurately capture the fluctuation range of each node in the power grid and provide data support for the safe and economic operation of the power grid. These data will be collected in real time through the interface and will be verified and processed to ensure their accuracy and completeness.

[0011] In one embodiment, the method adopts an improved Transformer model to improve the accuracy of load forecasting and adaptability to fluctuation range through multi-head attention mechanism, position encoding and multi-quantile regression. The model can effectively capture the complex patterns and trends of load changes by learning new energy output data, load data and meteorological observation data. The multi-head attention mechanism enhances the focus on multiple features and improves the adaptability to load fluctuations; multi-quantile regression enables the model to recursively predict future loads and accurately grasp the changing trends in time series; position encoding helps the model understand the relationship between each moment in the time series, so as to more accurately predict future load demand.

[0012] In one embodiment, the economic operation domain is generated, starting with clarifying the economic operation goals of the power grid, that is, minimizing economic factors such as power generation costs, system losses, and dispatching operation costs. To achieve this goal, the system needs to reasonably configure the output of various types of new energy power generation units and reasonably allocate the load demand of the power grid. The system optimizes the output combination of different power generation units by combining the forecast and historical data of new energy output to ensure that the power generation cost is minimized while avoiding insufficient or excessive power supply caused by new energy fluctuations. In this process, the system integrates all feasible dispatching schemes, taking into account factors such as new energy fluctuations, load demand, and power generation costs, to form a set of all feasible dispatching plans within the next 24 hours. These schemes together constitute the economic operation space of the power grid, and ultimately help determine the optimal dispatching plan to ensure that the power grid achieves the lowest operating cost and the best energy utilization efficiency while meeting the load demand.

[0013] In one embodiment, to further generate a safe operation domain, first of all, it is necessary to ensure that the power grid does not have safety hazards such as overload, instability or power shortage during the dispatching process. Its core goal is to ensure the safety and stability of the power grid. It is necessary to constrain the key safety parameters of the power grid, especially electrical parameters such as voltage, current and power, to ensure that the electrical parameters of each node in the power grid are always kept within the safe range to avoid equipment damage or system instability caused by parameter exceeding the standard. Secondly, it is necessary to comprehensively consider the line transmission capacity of the power grid and the load limit of the substation to ensure that the load of each transmission line in the power grid does not exceed its carrying capacity to avoid line tripping or power outage due to overload. At the same time, the output range of new energy generator sets also needs to be taken into consideration. The output of new energy such as wind power and photovoltaic power generation has strong volatility. Therefore, it is necessary to reserve a certain amount of redundancy in the dispatching to avoid unstable power supply caused by fluctuations in new energy. In addition, it is also necessary to consider the redundant design when the equipment fails to ensure that the power grid can still maintain stable operation through backup power or backup equipment when the equipment fails. Through the comprehensive consideration of these constraints, a safe operation domain is finally generated to ensure the safety and stability of the power grid under various operating conditions.

[0014] By comprehensively considering the economic operation domain and the safe operation domain, a set of day-ahead dispatch plans that meet the requirements of safe and economic operation of the power grid is obtained.

[0015] Considering the uncertainty of the prediction results of source and load nodes, the fluctuation range of new energy output and load demand is analyzed. In this process, considering that new energy output and load demand are affected by real-time factors such as meteorological changes, the prediction results have certain uncertainties. By analyzing the internal mechanism of the model and future real-time factors, combined with a certain confidence interval, the safe and economical operation range of the power grid is further defined. The remaining interval is the safety margin of the power grid, which indicates the fluctuation range that the power grid can tolerate when facing prediction uncertainty, ensuring that the power grid can still achieve safe and economical operation in an uncertain environment.

[0016] In one embodiment, the confidence interval is dynamically adjusted based on future real-time meteorological data, load changes, and fluctuations in new energy output. When the prediction results are relatively stable and the uncertainty is low, the confidence interval can be appropriately narrowed to reduce the safety margin and improve the economic efficiency of power grid operation; when the uncertainty of the prediction results is large, the confidence interval is expanded to increase the safety margin and ensure that the power grid can still maintain safe and stable operation under greater uncertainty.

[0017] The safe and economical operation margin of the power grid obtained based on the above steps is output to provide support for the day-ahead dispatching decision of the power grid, so as to ensure that the power grid can achieve safe and economical operation when facing factors such as fluctuations in renewable energy output, load changes and meteorological uncertainties.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] The improved Transformer model is used to more accurately predict the output of renewable energy, load demand and meteorological changes, and improve the safety and economy of power grid dispatching. By effectively analyzing the volatility and uncertainty of source and load nodes, and combining the dynamically adjusted confidence interval, it can flexibly respond to the impact of renewable energy fluctuations and future real-time factors, and ensure the stable operation of the power grid under different forecasting conditions. The establishment of the safe operation domain takes into account the electrical parameters, equipment redundancy and fault protection of the power grid, and enhances the stability and emergency response capabilities of the system. The above effects enhance the economy and reliability of the power grid under the dominance of renewable energy, and have strong practicality and broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1It is a flow chart of a domain-based intelligent evaluation method for safe and economical operation margin of a power grid provided in an embodiment. DETAILED DESCRIPTION

[0022] To make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation methods described herein are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0023] like Figure 1 As shown, the domain-based intelligent evaluation method for safe and economic operation margin of a power grid provided in the embodiment includes the following steps:

[0024] S110, the system obtains data from different sources, including renewable energy output data, load demand data, and meteorological observation data, through real-time monitoring and data collection technology. Furthermore, the improved Transformer algorithm is used to learn the above data and predict the fluctuation range of source and load nodes in the next 24 hours.

[0025] Among them, the new energy output data comes from various renewable energy power generation units. In this embodiment, solar energy and wind energy are considered to reflect the power output of various new energy sources. These data are recorded as P renewable (t) = [P solar (t),P wind (t)], where P solar (t) represents the solar power generation at time t, P wind (t) is the wind power generation. The load data includes the real-time and predicted power demand of each node in the power grid, which is recorded as L(t) = [L1(t), L2(t),…, L n (t)], where L i (t) represents the load demand of the i-th node at time t. Meteorological observation data include temperature, humidity, wind speed and radiation intensity, which affect power production and demand, and are recorded as M(t) = [T(t), H(t), V(t), R(t)], where T(t) is temperature, H(t) is humidity, V(t) is wind speed, and R(t) is radiation intensity. Through appropriate interfaces, the system collects these data in real time, verifies and processes them to ensure the accuracy and integrity of the data, and obtains the processed data P ′ , L ′ and M ′ , for ease of expression, it is denoted as x in the following text. Combining historical data, recent data and real-time data, the fluctuation range of each node in the power grid is further predicted.

[0026] In the embodiment, an improved generative Transformer model is used, the core component of which is a multi-head attention mechanism. This mechanism optimizes feature representation by calculating the correlation between input data. First, the input data x is compared with the matrix W q , W k and W v Multiply them together to get the query vector q, key vector k and value vector v:

[0027] q=W q x

[0028] k=W k x

[0029] v=W v x

[0030] Next, the correlation between the query vector q and all key vectors k is calculated, and the scaled dot product calculation formula is used to obtain the value weight coefficient of each key. The weight coefficient is normalized by softmax, and finally the sequence of numerical vectors v is weighted and summed to obtain the final attention score vector a:

[0031]

[0032] Where d is the dimension of the vector, and att represents the calculation mechanism of the attention score. In order to improve the feature capture ability of the model, a multi-head attention mechanism is adopted to parallelly calculate multiple query vector sequences Q = [q1, q2, ..., q m ], key vector sequence K = [k1, k2, ..., k m ] and a sequence of numerical vectors V = [v1,v2,…,v m ], calculate the attention independently in each head and concatenate them into a final feature representation:

[0033]

[0034] in, Represents concatenated vectors in the same dimension. Compared with traditional RNN models, the multi-head attention mechanism has the following two advantages: first, it can capture the long-term dependency between any two data points in the input sequence; second, it can process input data in parallel, significantly improving computational efficiency.

[0035] Further introducing position coding, position coding is achieved by encoding sine and cosine functions of different frequencies. For the pth position in the data sequence, its position coding is:

[0036]

[0037] Among them, p represents the location of the data, d posrepresents the dimension of the position encoding vector, Through the above formula, the position of each data will be mapped to a dimension d pos This vector provides the model with the absolute position information of the data. In addition, through the following formula, we can get the vector with position a+b, which can be represented by the linear transformation of position a and position b, thus providing the model with relative position information:

[0038]

[0039] Through the above position encoding method, the model can effectively learn the position information of the input data, thereby considering the order and relative position relationship of the data when calculating attention, enhancing the expressive ability of the model.

[0040] In the embodiment, the multi-quantile regression method in the improved Transformer method is used to predict the new energy output and grid load fluctuations in the next 24 hours. This method can effectively capture the complex patterns and trends of load changes and new energy fluctuations by learning new energy output data, load data and meteorological observation data. The input data of the model is x, and its corresponding output is the prediction result y of new energy output and load fluctuations. First, the model adopts the conditional quantile regression theory to predict the new energy output and load demand at each moment by training the conditional quantile regression model. Assume that the data set contains N groups of data in the form of D={(x1,y1),(x2,y2),…,(x N ,y N )}, where x i is the input data, y i is the corresponding new energy output and load forecast result. The conditional cumulative distribution function (CDF) is defined as:

[0041] F(y|x i )=P(y i ≤y|x i )

[0042] Based on this, the model defines the α quantile as:

[0043]

[0044] To train the model, the pinball loss function is used, which has the form:

[0045]

[0046] Among them, ρ α is the pinball loss function, defined as:

[0047]

[0048] This loss function can effectively perform quantile prediction, thereby outputting the prediction interval of future renewable energy output and load. The specific prediction interval is expressed by the following formula:

[0049] C(x)=[q β / 2 (x),q 1-β / 2 (x)]

[0050] In order to further optimize the output and load forecast of renewable energy, a multi-quantile regression method based on generative Transformer is adopted, considering the pinball loss of multiple quantiles ranging from 1% to 99%. Its loss function is:

[0051]

[0052] in, represents the new energy output or load value at time t on the i-th day, Represents the j% quantile forecast value of the renewable energy output or load at time t on the i-th day. In this way, the model can provide a forecast of renewable energy output and load demand for the next 24 hours within multiple quantile ranges, and accurately grasp the changing trend in the time series through recursive forecasting. Finally, the generated multi-quantile forecast results are represented as a matrix, where each row corresponds to a quantile sequence and each column corresponds to a forecast value at a moment. The specific forecast values ​​are as follows:

[0053]

[0054] In this way, the model can generate different quantile forecasts of renewable energy output and load at each moment in the next 24 hours, provide uncertainty analysis for grid dispatching and renewable energy management, and help effectively deal with the impact of renewable energy fluctuations and load fluctuations.

[0055] S120, solving the economic operation domain The goal of economic operation is to minimize the total cost of the power grid, mainly considering the output of new energy power generation, including the volatility of renewable energy such as wind power and photovoltaic power, as well as related system losses and scheduling operation costs. The objective function can be expressed as:

[0056] Minimize C total =C gen +C loss +C operation

[0057] Among them, C total is the total cost, C gen is the power generation cost, C loss is the system loss cost, C operationis the dispatching operation cost. The power generation cost mainly comes from the output of new energy, specifically the output power of wind power and photovoltaic power. It can be expressed as:

[0058]

[0059] Among them, c i is the unit power generation cost of the i-th renewable energy power generation unit (wind power and photovoltaic power are selected in this embodiment), P gen,i is the output power of the i-th power generation unit, and n is the total number of renewable energy power generation units. The system loss cost is usually related to the loss during the power flow process and can be expressed as:

[0060]

[0061] Among them, L k is the loss coefficient of the kth link, P k is the power output of this link. The dispatching operation cost is usually related to operations such as adjusting the output of renewable energy power generation units and adjusting load distribution, and can be expressed as:

[0062]

[0063] Among them, d i To adjust the operating cost of the output of the i-th renewable energy generation unit, ΔP gen,i is the adjustment of the output of the i-th power generation unit. An important constraint of power grid dispatch is to ensure that the load demand is met, which is usually expressed as:

[0064]

[0065] Among them, P load is the total load demand of the power grid. The output of each renewable energy power generation unit must be within its minimum and maximum output range, expressed as:

[0066]

[0067] Since new energy (only wind energy and solar energy are considered in this embodiment) is volatile, its output is affected by meteorological conditions (such as wind speed, radiation intensity, etc.), so these uncertain factors need to be considered. Wind energy output can be expressed as:

[0068] P wind =f wind (v wind ,θ wind )

[0069] Among them, v wind is the wind speed, θ wind is the angle of the wind turbine. Photovoltaic output can be expressed as:

[0070] Psolar =f solar (I solar ,θ solar )

[0071] Among them, I solar is the solar radiation intensity, θ solar is the angle of the solar panel. In order to ensure the stable operation of the power grid, it is also necessary to consider the constraints such as the frequency and node voltage of the power grid, which can be expressed as:

[0072]

[0073] Among them, V k is the voltage amplitude of the kth node, V min is the minimum requirement for voltage amplitude. Combining these objective functions and constraints, the optimization algorithm is used to solve them, and finally the optimal output combination of renewable energy power generation units and load distribution scheme are obtained, so as to ensure that the power grid can meet the load demand and achieve the economic dispatch target, especially in the environment dominated by renewable energy, to optimize the impact of its volatility on the economy and reliability of the power grid.

[0074] Based on the above objective function and constraints, the economic operation domain of the power grid can be defined as the set of all feasible solutions that meet the constraints. Specifically, the economic operation domain is expressed as:

[0075]

[0076] The economic operation domain includes all possible combinations of output combinations of different renewable energy power generation units and load distribution schemes, that is, the set of economic dispatch plans for the power grid in the next 24 hours.

[0077] S130, solving the safe operation domain The core of safe operation of the power grid is to ensure that there are no potential safety hazards such as overload, instability or power shortage during the dispatching process, so as to ensure the safety and stability of the power grid system. First of all, the voltage of each node in the power grid must be within the safe range, that is, meet the following conditions:

[0078]

[0079] Among them, V k represents the voltage of the kth node, V min and V max are the minimum and maximum values ​​of the node voltage respectively. The current of each node in the power grid cannot exceed the line transmission capacity to ensure that the power grid will not be overloaded. For the kth line, its current constraint is:

[0080]

[0081] Among them, Iline,k represents the current of the kth line, is the maximum current transmission capacity of the line. The power demand and power generation capacity of each node in the power grid must be balanced and meet certain power margin requirements. For the kth node, its power constraint is:

[0082]

[0083] Among them, P load,k is the load demand of the kth node, P gen,k is the power generation capacity of the kth node, P max,k is the maximum available power generation of the kth node. The load of the substation must be controlled within its maximum load capacity to avoid damage or failure of substation equipment due to excessive load. Suppose the load constraint of the kth substation is:

[0084]

[0085] Among them, P substation,k represents the load of the kth substation, is the maximum load capacity of the substation. Renewable energy generation (such as wind power and photovoltaic power) is volatile and its output is affected by meteorological factors, so it needs to be dispatched within its expected output range. Suppose the output constraint of the i-th renewable energy generator set is:

[0086]

[0087] in, and are the minimum and maximum outputs of the i-th renewable energy generator set. In order to cope with possible equipment failures, the power grid needs to ensure that it can still maintain safe and stable operation of the system in the event of a failure. Assuming that when the k-th transmission line fails, the power grid still needs to maintain sufficient power and voltage margins. Therefore, the constraints after the failure are:

[0088]

[0089] in, is the power loss when a fault occurs, P load is the total load demand of the system. Finally, all constraints constitute the safe operation domain of the power grid, which is expressed as:

[0090]

[0091] The safe operation domain represents the set of all possible solutions that meet the safety constraints of power grid operation, ensuring that the power grid can operate stably without overload or power shortage under various disturbances and faults. Through real-time data monitoring and updating, the safe operation domain can be dynamically adjusted to cope with the ever-changing power grid operation conditions.

[0092] S140, determine the safety margin based on the confidence interval. In the embodiment, in order to ensure that the power grid can still achieve safe and economical operation in the face of uncertain factors, it is necessary to first comprehensively consider the fluctuation range of renewable energy output and load demand. This process involves uncertainty analysis of the prediction results, especially due to the influence of real-time factors such as meteorological changes, which leads to certain uncertainty in renewable energy output and load demand. To this end, the predicted values ​​of renewable energy output and load demand can be defined as and Its uncertainty can be expressed by a confidence interval. Set a confidence level α∈(0,1), and the corresponding prediction interval can be expressed by the following formula:

[0093]

[0094]

[0095] Among them, ΔP gen,i (t) and ΔP load,k (t) represent the fluctuation range of renewable energy output and load demand, respectively. These fluctuation ranges are related to the confidence intervals of historical data, real-time meteorological information and model predictions. Assuming that at a given future time t, the load demand and renewable energy output of the power grid are affected by uncertainty, the safe and economic operation range of the power grid can be obtained by analyzing these uncertainty ranges. This range can be expressed by the following formula:

[0096]

[0097] in, Represents the safe operation domain of the power grid under a given confidence interval α. Furthermore, when considering the economic efficiency of the power grid, the economic operation domain and the safe operation domain can be combined to obtain the final safe and economic operation domain. This domain not only takes into account the economic dispatching objectives of the power grid (such as minimizing the cost of power generation, loss and dispatching cost), but also takes into account the uncertainty of new energy output and load demand. The specific economic operation objectives of the power grid can be expressed as minimizing the cost function:

[0098]

[0099] When considering the uncertainty comprehensively, the safety margin S of the power grid is safe It can be expressed as:

[0100]

[0101] Ultimately, the safety margin of the power grid is expressed as the fluctuation range that the power grid can tolerate when facing fluctuations in renewable energy output, load demand, and forecast uncertainty. By dynamically adjusting the dispatching strategy of the power grid, the power grid can achieve safe and economical operation in an uncertain environment.

[0102] S150, dynamically adjust the confidence interval. In the embodiment, the confidence interval is dynamically adjusted according to future real-time meteorological data, load changes, and fluctuations in the output of new energy sources. When the prediction result is relatively stable and the uncertainty is low, the confidence level α is appropriately reduced, and the range of the confidence interval can be appropriately narrowed, thereby reducing the safety margin and improving the economy of the power grid operation; and when the uncertainty of the prediction result is large, the confidence level α is appropriately increased, and the range of the confidence interval is expanded to increase the safety margin, so as to ensure that the power grid can still maintain safe and stable operation under greater uncertainty.

[0103] The specific implementation methods described above provide a detailed description of the technical solutions and beneficial effects of the present invention. It should be understood that the above is only the most preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, supplements and equivalent substitutions made within the scope of the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A domain-based intelligent evaluation method for power grid safety and economic operation margin, characterized in that: The following steps are involved: Step 1: Use artificial intelligence algorithms to learn new energy output data, load data and meteorological observation data to obtain a model and use it to predict the fluctuation range of source and load nodes in the next 24 hours; Step 2: Based on the prediction results of source and load nodes, construct the economic operation domain and establish the economic dispatch plan set of the power grid; Step 3: Based on the prediction results of the source and load nodes and the operating status of the power grid, a safe operation domain is constructed to determine the safe operation range of the power grid; Step 4: Comprehensively utilize the economic operation domain and the safe operation domain to obtain a set of day-ahead dispatch plans that meet the safe and economic operation requirements of the power grid; Step 5: Utilize the uncertainty of the prediction results of source and load nodes to analyze the fluctuation range of source and load nodes and obtain the safe and economic operation margin of the power grid.

2. The domain-based intelligent evaluation method for power grid safety and economic operation margin according to claim 1 is characterized in that: The artificial intelligence method described in step one is an improved Transformer model, which includes a multi-head attention mechanism, position encoding, and multi-quantile regression to enhance the accuracy of load forecasting and its adaptability to the load fluctuation range.

3. The domain-based intelligent evaluation method for power grid safety and economic operation margin according to claim 1 is characterized in that: The new energy output data described in step one includes power output data from various types of renewable energy sources such as solar energy, geothermal energy, wind energy, ocean energy, biomass energy and nuclear fusion energy; the load data includes load demand data of each node in the power grid; the meteorological observation data includes temperature, humidity, wind speed, radiation and meteorological information related to power production and demand; the above data includes historical data, recent data and real-time data.

4. The domain-based intelligent evaluation method for power grid safety and economic operation margin according to claim 1 is characterized in that: The goal of economic operation in step two is to minimize the total cost of the power grid, mainly considering the output of new energy power generation, including the volatility of renewable energy such as wind power and photovoltaic power, as well as related system losses and dispatch operation costs.

5. The domain-based intelligent evaluation method for power grid safety and economic operation margin according to claim 1 is characterized in that: The goal of safe operation in step three is to ensure that the power grid will not have safety hazards such as overload, instability or power shortage during the dispatching process, thereby ensuring the safety and stability of the power grid system.

6. The domain-based intelligent evaluation method for power grid safety and economic operation margin according to claim 1 is characterized in that: In step 5, after obtaining the set of day-ahead dispatch plans that meet the requirements of safe and economical operation of the power grid, the confidence interval of safe and economical operation of the power grid is further delineated under a certain confidence level. The remaining interval is used as the safety margin of the power grid, which represents the tolerance of the power grid within the range of uncertainty.

7. The domain-based intelligent evaluation method for safe and economic operation margin of power grid according to claim 6 is characterized in that: The confidence level is dynamically adjusted based on future real-time meteorological data, load changes, and fluctuations in renewable energy output: when the forecast results are relatively stable, the forecast uncertainty is low. At this time, the range of the confidence interval can be appropriately narrowed, and the confidence level can be reduced, thereby reducing the safety margin range and improving the economy of power grid operation; when the uncertainty of the forecast results is large, the range of the confidence interval will increase accordingly, increasing the confidence level to increase the safety margin and improve the safety of power grid operation.

Images