Power system multi-resource collaborative planning method and system based on attribution analysis

Abstract

The application relates to a power system multi-resource collaborative planning method based on attribution analysis, and belongs to the technical field of power system planning. The method comprises the following steps: constructing a target function based on the load demand of a power system and the output of each unit, taking the minimum mismatch degree between supply and demand as the target, constructing a planning model based on set constraints and the target function; solving the planning model based on multiple sets of boundary condition values of the planning model to obtain corresponding planning schemes; analyzing and obtaining quantitative representations of the influence degree of each boundary condition and each constraint on the planning scheme; adjusting the boundary condition values based on the quantitative representations and solving the planning model, and selecting the planning scheme with the lowest planning cost in the solving result as the optimal planning scheme. The application determines the key factors influencing the planning scheme through attribution analysis on power planning, and determines the optimal planning scheme by adjusting the key factors, so that the scientificity and economy of the planning scheme are further improved.

Description

Technical Field

[0001] This invention relates to the field of power system planning technology, and in particular to a multi-resource collaborative planning method and system for power systems based on attribution analysis. Background Technology

[0002] In the process of green and low-carbon energy transformation, the integration of high-proportion renewable energy sources such as wind power and photovoltaics has become an inevitable development trend under the new power system model. During this transformation, the coordinated development of multi-timescale energy storage is an effective way to ensure supply and demand balance and renewable energy consumption. Therefore, resource planning decisions need to be made based on the coordinated operation of multiple technologies, which places more stringent requirements on power system planning methods.

[0003] Traditional planning models are all "black boxes" and cannot explain the physical logic of the decisions within the model. Planning decision-makers can indeed obtain the optimal solution in the sense of the objective function, that is, determine the optimal planning scheme of the power system under the premise of clear boundary conditions and constraints.

[0004] However, traditional planning models cannot clearly extract the dominant factors driving optimization results from the model, nor can they uncover the decision-making logic within the model. When faced with differences in resource demands across different regions (which could be small areas like a park or city, or large areas like an entire province or inter-provincial areas) or power grids of different sizes (such as microgrids or large-scale power grids), they lack a basis for formulating scientifically reasonable boundary conditions and constraints for each region. Furthermore, when facing different resource demands at different planning and construction stages, they similarly lack a basis for boundary conditions and constraints tailored to the different development characteristics of each stage (e.g., different demands for different types of energy storage resources at different planning stages, and different levels of stringency in carbon emission requirements at different planning stages). Consequently, traditional planning models cannot effectively provide scientific support for power system planning decisions across different regions, power grids of different sizes, and different planning and construction stages within the same region. Summary of the Invention

[0005] In view of the above analysis, the present invention aims to disclose a multi-resource collaborative planning method and system for power systems based on attribution analysis. When making power system planning decisions, attribution analysis is used to determine the degree of influence of each boundary condition or constraint on the planning scheme. The boundary conditions corresponding to the boundary conditions that have a greater impact on the planning scheme or the constraints that have a greater impact on the planning decision are adjusted according to the different degrees of influence. The optimal power system planning decision is determined based on the adjusted boundary conditions.

[0006] On the one hand, this invention provides a multi-resource collaborative planning method for power systems based on attribution analysis, specifically including the following steps:

[0007] An objective function is constructed based on the load demand of the power system and the output of each generating unit, with the goal of minimizing the supply-demand mismatch. A planning model is constructed based on power balance constraints, line flow constraints, output constraints of each generating unit, low-carbon constraints, and the objective function.

[0008] Based on the multiple sets of boundary condition values ​​of the planning model, the planning model is solved to obtain the planning scheme corresponding to each set of boundary condition values;

[0009] The SHAP method is used to analyze and obtain a quantitative representation of the influence of each boundary condition and each constraint on the planning scheme.

[0010] Based on the quantification, the corresponding boundary condition values ​​are adjusted, and the planning model is solved based on the multiple sets of adjusted boundary condition values. The planning scheme with the lowest planning cost among the solution results is selected as the optimal planning scheme.

[0011] Furthermore, the generating units include wind power, photovoltaic, and energy storage units; the objective function constructed based on the load demand of the power system and the output of each generating unit, with the goal of minimizing the supply-demand mismatch, includes:

[0012] When the sum of wind power output and photovoltaic power output exceeds the load demand, a power calculation formula for supply-demand mismatch is constructed based on wind power output, photovoltaic power output, load demand, and energy storage charging power.

[0013] When the sum of wind power output and photovoltaic power output is not greater than the load demand, a power calculation formula for supply-demand mismatch is constructed based on wind power output, photovoltaic power output, load demand and energy storage discharge power.

[0014] Based on the power calculation formula for supply-demand mismatch in each time period and the load demand in each time period, an objective function is constructed with the goal of minimizing the supply-demand mismatch.

[0015] Furthermore, the boundary conditions of the planning model include:

[0016] Load demand, wind power capacity factor, photovoltaic capacity factor, maximum carbon emissions, carbon emission intensity of thermal power units, maximum upward / downward flexible adjustment capability of thermal power units, upper limit of power of thermal power units, upper limit of power transmission line power, phase angle of transmission nodes, long-term energy storage charging / discharging efficiency, and short-term energy storage charging / discharging efficiency.

[0017] Furthermore, the SHAP method is used to analyze and quantify the impact of each boundary condition and constraint on the planning scheme, including:

[0018] A global interpretability analysis is performed on the planning model, and the marginal contribution (SHAP) value of each boundary condition is calculated as a quantitative representation of each boundary condition.

[0019] A local interpretability analysis is performed on the planning model, and the comprehensive relaxation value of each constraint is calculated as a quantitative representation of each constraint.

[0020] Furthermore, the overall relaxation value of each constraint is calculated using the following formula:

[0021] ;

[0022] Let i be the relaxation degree of constraint i at time t under different time resolutions; For the set of model constraints; This is the maximum output of the thermal power unit; The output of the thermal power unit at time t in the optimal solution of the model; Let i be the overall relaxation degree of constraint i during the evolution of the planning towards the optimal solution; To constrain the set of dual variables; These are the dual variables corresponding to each constraint in the locally interpretable model;

[0023] The local interpretability model of the planning model is represented as follows:

[0024] ;

[0025] To optimize variables, The optimization variable in the model is a 0 / 1 variable, which is the operating status of the thermal power unit. The optimization variables in the model are the solution variables. The coefficient column vector corresponding to the objective function; , , and This is the coefficient matrix of the variables under the corresponding constraints; , A constant column vector; This represents the inequality constraints in the planning model; These are the equality constraints in the planning model; This corresponds to the constraints containing 0 / 1 variables in the planning model.

[0026] Furthermore, the step of adjusting the boundary condition values ​​corresponding to the quantified representation, solving the planning model based on the multiple sets of adjusted boundary condition values, and selecting the planning scheme with the lowest planning cost among the solution results as the optimal planning scheme includes:

[0027] The SHAP values ​​are sorted from high to low, and the comprehensive relaxation values ​​are sorted from low to high.

[0028] The boundary conditions corresponding to the top-ranked SHAP values ​​are selected and their values ​​are adjusted within their respective ranges. And / or the boundary conditions included in the constraints corresponding to the top-ranked comprehensive relaxation values ​​are selected and their values ​​are adjusted within their respective ranges to obtain multiple sets of boundary condition values. The model is then solved based on these multiple sets of boundary condition values.

[0029] Based on the multiple solution results obtained, the corresponding planning costs are calculated respectively;

[0030] The planning scheme with the lowest planning cost is selected as the optimal planning scheme.

[0031] Furthermore, the objective function is expressed as:

[0032] ;

[0033] ;

[0034] in, The system's supply and demand mismatch. The system's power supply and demand are mismatched. , The wind power and solar power outputs at time t are respectively. The load demand at time t , These represent the charging and discharging power of energy storage, respectively.

[0035] Furthermore, the carbon emission constraints are expressed as follows:

[0036] ;

[0037] in, Indicates the area The set of thermal power units within the region; T represents the total duration; For thermal power units exist The effort required at that time; The carbon emission intensity of thermal power units; For time intervals; Maximum permissible carbon emissions.

[0038] Furthermore, the power balance constraint is expressed as:

[0039] ;

[0040] Among them, subscript express area; , , , Representing regions The set of thermal power units, new energy units, load nodes, and energy storage units within the system; express A collection of connecting lines between the region and other regions; For thermal power units exist The effort required at that time; , These are wind turbine units (S) and photovoltaic units (B / S). The output at time t; and They are respectively Total discharge power and total charging power of regional energy storage; for Time zone Links between other areas The transmission power; Indicates when t Regional load demand; and These represent the discharge power and charging power of short-term energy storage n at time t, respectively. and These represent the discharge power and charging power of long-term energy storage m at time t, respectively.

[0041] On the other hand, the present invention also provides a power system multi-resource collaborative planning system based on attribution analysis, comprising:

[0042] The planning model management module is used to construct an objective function based on the load demand of the power system and the output of each unit, with the goal of minimizing the supply-demand mismatch. The planning model is constructed based on power balance constraints, line flow constraints, output constraints of each unit, low-carbon constraints, and the objective function.

[0043] The planning model solving module is used to solve the planning model based on multiple sets of boundary condition values ​​to obtain the planning scheme corresponding to each set of boundary condition values.

[0044] The attribution analysis module is used to analyze and obtain a quantitative representation of the influence of each boundary condition and each constraint on the planning scheme using the SHAP method;

[0045] The planning scheme formulation module is used to adjust the corresponding boundary condition values ​​based on the quantified representation, solve the planning model based on the multiple sets of boundary condition values ​​obtained after adjustment, and select the planning scheme with the lowest planning cost among the solution results as the optimal planning scheme.

[0046] The present invention can achieve at least one of the following beneficial effects:

[0047] Attribution analysis is used to determine the degree of influence of each boundary condition or constraint on the planning scheme. Based on the different degrees of influence, the boundary conditions corresponding to the boundary conditions that have a significant impact on the planning scheme or the constraints that have a significant impact on the planning decision are adjusted. The optimal power system planning scheme is then determined based on the adjusted boundary conditions. This approach enables the scientific development of optimal power system planning schemes suitable for different regions and planning stages, and saves construction costs and improves economic efficiency by selecting the planning scheme with the lowest planning cost.

[0048] By using the SHAP method to analyze and quantify the impact of each boundary condition or constraint on the planning scheme, the boundary conditions with the greatest impact on the planning scheme can be identified, which can guide practical applications. By improving the technology, the boundary conditions can be adjusted. For example, when it is found that the energy storage capacity has a significant impact on the planning scheme, the planning cost can be reduced by selecting energy storage equipment with a large energy storage capacity in the current plan. At the same time, it should promote the development of energy storage technology and expand energy storage capacity in future system construction.

[0049] Other features and advantages of the invention will be set forth in the following description, and some advantages may become apparent from the description or be learned by practicing the invention. Attached Figure Description

[0050] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0051] Figure 1 This is a flowchart of the multi-resource collaborative planning method for power systems according to the present invention;

[0052] Figure 2 This is the attribution analysis result of the boundary conditions in the embodiments of the method of the present invention;

[0053] Figure 3 This is the attribution analysis result of the constraint conditions in the embodiments of the present invention. Detailed Implementation

[0054] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0055] Method Implementation Examples

[0056] One embodiment of the present invention discloses a multi-resource collaborative planning method for power systems based on attribution analysis, which includes the following steps:

[0057] Step S01: Based on the load demand of the power system and the output of each unit, construct an objective function with the goal of minimizing the supply-demand mismatch. Construct a planning model based on power balance constraints, line power flow constraints, output constraints of each unit, low-carbon constraints, and the objective function.

[0058] Step S02: Based on the multiple sets of boundary condition values ​​of the planning model, solve the planning model to obtain the planning scheme corresponding to each set of boundary condition values;

[0059] Step S03: Use the SHAP method to analyze and obtain a quantitative representation of the degree of influence of each boundary condition and each constraint on the planning scheme;

[0060] Step S04: Adjust the boundary condition values ​​corresponding to the quantified representation, solve the planning model based on the multiple sets of boundary condition values ​​obtained after adjustment, and select the planning scheme with the lowest planning cost among the solution results as the optimal planning scheme.

[0061] The quantification representation includes the marginal contribution (SHAP) value of each boundary condition and the comprehensive relaxation value of each constraint. Adjusting the boundary condition values ​​based on the quantification representation includes: sorting the SHAP values ​​from high to low, and sorting the comprehensive relaxation values ​​from low to high; adjusting the corresponding values ​​of the boundary conditions corresponding to the top-ranked SHAP values, and / or adjusting the corresponding values ​​of the boundary conditions included in the top-ranked comprehensive relaxation values.

[0062] This embodiment discloses a multi-resource collaborative planning method for power systems based on attribution analysis. Attribution analysis determines the degree of influence of each boundary condition or constraint on the planning scheme. Based on the different degrees of influence, the boundary conditions corresponding to the boundary conditions that have a significant impact on the planning scheme or the constraints that have a significant impact on the planning decision are adjusted. Based on the adjusted boundary conditions, the optimal power system planning scheme is determined. This method can scientifically formulate optimal power system planning schemes suitable for different regions and planning stages, and save construction costs and improve economic efficiency by selecting the planning scheme with the lowest planning cost.

[0063] Specifically, in step S01, the power system refers to a power generation system composed of various generator sets. These generator sets include wind power, photovoltaic power, thermal power, and energy storage units; the objective function constructed based on the load demand of the power system and the output of each generator set, with the goal of minimizing the supply-demand mismatch, includes:

[0064] When the sum of wind power output and photovoltaic power output exceeds the load demand, a power calculation formula for supply-demand mismatch is constructed based on wind power output, photovoltaic power output, load demand, and energy storage charging power.

[0065] When the sum of wind power output and photovoltaic power output is not greater than the load demand, a power calculation formula for supply-demand mismatch is constructed based on wind power output, photovoltaic power output, load demand and energy storage discharge power.

[0066] Based on the power calculation formula for supply-demand mismatch in each time period and the load demand in each time period, an objective function is constructed with the goal of minimizing the supply-demand mismatch.

[0067] Furthermore, the objective function is expressed as:

[0068] ;

[0069] ;

[0070] in, The system's supply and demand mismatch. The system's power supply and demand are mismatched. , The wind power and solar power outputs at time t are respectively. The load demand at time t , These represent the charging and discharging power of energy storage, respectively.

[0071] Specifically, in step S01, the power balance constraint is expressed as:

[0072] ;

[0073] Among them, subscript express area; , , , Representing regions The set includes thermal power units, new energy units (wind turbines and photovoltaic units), load nodes, and energy storage units. express A collection of connecting lines between the region and other regions; For thermal power units exist The effort required at that time; , These are wind turbine units (S) and photovoltaic units (B / S). The output at time t; and They are respectively Total discharge power and total charging power of regional energy storage; for Time zone Links between other areas The transmission power; Indicates when t Regional load demand; and These represent the discharge power and charging power of short-term energy storage n at time t, respectively. and These represent the discharge power and charging power of long-term energy storage m at time t, respectively.

[0074] Specifically, in step S01, the line power flow constraint is expressed as:

[0075] ;

[0076] in, for Time zone Links between other areas The transmission power; and They are respectively with the line The phase angle of connected nodes, For the line Reactance, This represents the upper limit of the line's transmission power.

[0077] Specifically, in step S01, the output constraints of each unit include new energy output constraints, thermal power output constraints, and energy storage operation constraints.

[0078] Furthermore, the power output constraint of new energy sources is expressed as follows:

[0079] ;

[0080] in, and These refer to the installed capacity of wind turbine units and photovoltaic units, respectively. and These are the wind power capacity factor and the photovoltaic capacity factor, respectively.

[0081] Furthermore, the thermal power output constraint is expressed as:

[0082] ;

[0083] in, For thermal power units exist The effort required at that time; and thermal power units Upper and lower limits of output; The variable is 0 / 1, representing the operating status of the thermal power unit. 1 indicates that the thermal power unit is running, and 0 indicates that the thermal power unit is shut down. and thermal power units The maximum upward and downward flexible adjustment capabilities at time t.

[0084] Furthermore, energy storage operation constraints include short-term energy storage constraints and long-term energy storage constraints.

[0085] Short-term energy storage constraints are expressed as:

[0086] ;

[0087] in, and These represent the maximum short-term energy storage capacity and maximum power, respectively, indicated by the subscript. ds represents charging and discharging. For short-term energy storage at time t, and These represent the capacities at the initial and final stages of short-term energy storage operation, respectively. and These are the charging efficiency and discharging efficiency for short-term energy storage, respectively. and These represent the charging and discharging power of short-term energy storage at time t, respectively. and These represent the maximum charging and maximum discharging power for short-term energy storage, respectively.

[0088] Long-term energy storage constraints are expressed as:

[0089] ;

[0090] in, , and These represent the capacity, charging power, and discharging power of long-term energy storage at time t, respectively. and For long-term energy storage, discharge efficiency and charging efficiency; For time intervals; and These represent the capacity at the initial and final stages of long-term energy storage operation, respectively. and These are the upper and lower limits of the long-term energy storage capacity-power ratio, respectively. , , These are the maximum long-term energy storage capacity, maximum discharge power, and maximum charging power, respectively.

[0091] Specifically, in step S01, the low-carbon constraint is expressed as:

[0092] ;

[0093] in, The carbon emission intensity of thermal power units; The time interval is T; the total duration is T. Maximum permissible carbon emissions.

[0094] Specifically, in step S01, a planning model is constructed based on power balance constraints, line power flow constraints, unit output constraints, low-carbon constraints, and the objective function.

[0095] Furthermore, the boundary conditions of the planning model include: load demand. Wind power capacity factor Photovoltaic capacity factor Maximum carbon emissions Carbon emission intensity of thermal power units Maximum upward / downward flexible adjustment capability of thermal power units and Upper limit of thermal power unit power Upper limit of line transmission power Transmission node phase angle and Long-term energy storage charge / discharge efficiency, short-term energy storage charge / discharge efficiency.

[0096] Specifically, in step S02, the planning model is solved based on multiple sets of boundary condition values ​​to obtain planning schemes corresponding to each set of boundary condition values. The planning schemes include: planning results for multi-timescale energy storage. (Maximum capacity for long-term energy storage) (Maximum charging power for long-term energy storage) (Maximum discharge power for long-term energy storage) (Maximum capacity for short-term energy storage) and (Maximum power of short-term energy storage), and the installed capacity of wind and solar new energy units. and Operating status of each unit (thermal power units) exist (output at the time) (Wind turbine S in) (output at the time) (Photovoltaic units) (output at time t) (Discharge power of short-term energy storage n at time t) (Charging power of short-term energy storage n at time t) (Discharge power of long-term energy storage m at time t) (Charging power of long-term energy storage m at time t), and power of system supply-demand mismatch. .

[0097] Specifically, in step S03, based on the multiple planning schemes obtained in step S02, the SHAP method is used to analyze and obtain a quantitative representation of the influence of each boundary condition or constraint on the planning scheme, including:

[0098] A global interpretability analysis is performed on the planning model, and the marginal contribution (SHAP) value of each boundary condition is calculated as a quantitative representation of each boundary condition.

[0099] A local interpretability analysis is performed on the planning model, and the comprehensive relaxation value of each constraint is calculated as a quantitative representation of each constraint.

[0100] Furthermore, a global interpretability analysis is performed on the planning model. Under all possible combinations of boundary conditions (features), the Shapley Value of each boundary condition (feature) is calculated and a weighted average is taken. SHAP is used to achieve a global interpretation of the boundary conditions (features) and to capture the interactive coupling effect between different boundary condition (feature) parameters. The calculation process is as follows:

[0101] ;

[0102] in, This indicates the calculation of the SHAP value for the planning model; This indicates whether specific input boundary conditions affect the output planning result; To characterize the planning outcome of variable i in the planning model SHAP value of contribution; The reference value for a specific input boundary.

[0103] Among them, the marginal contribution SHAP value of each boundary condition The calculation method is as follows:

[0104] ;

[0105] Where M is the set of boundary conditions (feature inputs), is the number of feature inputs, and represents a subset of feature inputs. The importance of each input boundary is calculated by evaluating the difference in importance between its presence and absence in the planning scheme. This represents the model output when the i-th feature input occurs; This represents the model output when the i-th factor does not appear. It is all possible subsets of M The weighted average. If This indicates the positive impact of boundary feature input i on improving planning efficiency, thereby inferring the dominant factors driving planning schemes under different boundaries.

[0106] like Figure 2 The attribution analysis results for the boundary conditions are shown in the figure, which illustrates the marginal contribution (SHAP) value for each boundary condition.

[0107] Furthermore, a local interpretability analysis is performed on the planning model, and the comprehensive relaxation value of each constraint is used to reflect the degree of relaxation or tension of the individual model constraints under specific feature inputs.

[0108] Specifically, the compact form of the planning model can be represented as follows:

[0109] ;

[0110] in, To optimize variables, The optimization variable in the model is a 0 / 1 variable, which is the operating status of the thermal power unit. The optimization variables in the model are the solution variables. The coefficient column vector corresponding to the objective function; , , and This is the coefficient matrix of the variables under the corresponding constraints; , A constant column vector; This represents the inequality constraints in the planning model; These are the equality constraints in the planning model; The constraints in the planning model that correspond to 0 / 1 variables; These are the dual variables corresponding to each constraint in the local interpretability model.

[0111] The corresponding expression is:

[0112] .

[0113] It should be noted that the dual variable of a single constraint, namely the relaxation, changes with the constraint's time resolution. To measure the overall relaxation of a single constraint in the planning model, when the constraint is tightened, the relaxation will be closer to 0, and the value of the dual variable will not be 0. However, when the constraint is relaxed, the value of the dual variable will be 0.

[0114] Furthermore, the method for calculating the overall relaxation of each constraint is as follows: sum the absolute values ​​of the dual variables at each resolution to reflect the tightness of a specific constraint in the planning model. For example, after inputting the parameters of a specific boundary condition, the optimal operating conditions and planning schemes of each unit are obtained, and the distance between them and the upper and lower bounds of the constraints is calculated.

[0115] Furthermore, the overall relaxation value of each constraint is calculated using the following formula:

[0116] ;

[0117] Let i be the relaxation degree of constraint i at time t under different time resolutions; For the set of model constraints; This is the maximum output of the thermal power unit; The output of the thermal power unit at time t in the optimal solution of the model; Let i be the overall relaxation degree of constraint i during the evolution of the planning towards the optimal solution; This is the set of constraint dual variables. By calculating the combined relaxation of each constraint, we measure the impact of each constraint on the optimal solution. The contribution of constraints to the planning objective is considered. Constraints with low even variable values ​​have a higher overall relaxation degree, indicating that the constraint has a large margin under the current optimal solution and its marginal contribution to the planning objective is small. Conversely, constraints with high even variable values ​​have a lower overall relaxation degree, indicating that the constraint is a tight constraint or close to a tight constraint and its marginal contribution to the planning objective is large. This helps identify tight constraints that significantly contribute to the planning results. For example... Figure 3 The above are the attribution analysis results of the constraints, and the figure shows the comprehensive relaxation value of each constraint.

[0118] Specifically, in step S04, the boundary condition values ​​are adjusted based on the quantified representation, and the planning model is solved based on the adjusted sets of boundary condition values. The planning scheme with the lowest planning cost among the solution results is selected as the optimal planning scheme, including:

[0119] The SHAP values ​​are sorted from high to low, and the comprehensive relaxation values ​​are sorted from low to high.

[0120] The boundary conditions corresponding to the top-ranked SHAP values ​​are selected and their values ​​are adjusted within their respective ranges. And / or the boundary conditions included in the constraints corresponding to the top-ranked comprehensive relaxation values ​​are selected and their values ​​are adjusted within their respective ranges to obtain multiple sets of boundary condition values. The model is then solved based on these multiple sets of boundary condition values.

[0121] Based on the multiple solution results obtained, the corresponding planning costs are calculated respectively;

[0122] The planning scheme with the lowest planning cost is selected as the optimal planning scheme.

[0123] For example, such as Figure 2As shown, the boundary conditions that contribute significantly to the planning model include load demand, maximum carbon emissions, the upward adjustment capacity of thermal power units, long-term energy storage discharge efficiency, and short-term energy storage charge and discharge efficiency; such as Figure 3 The constraints that significantly contribute to the planning model include power balance constraints, carbon emission constraints, and long-term energy storage capacity constraints. Therefore, the boundary conditions that may need adjustment include the maximum allowable carbon emissions, the upward adjustment capability of thermal power units, long-term energy storage discharge efficiency, short-term energy storage charge / discharge efficiency, and boundary conditions involved in the constraints, such as the maximum long-term energy storage capacity, maximum discharge power, and maximum charging power.

[0124] It should be noted that when selecting the boundary conditions corresponding to the top-ranked SHAP values ​​and / or the constraint conditions corresponding to the top-ranked comprehensive relaxation values, the corresponding boundary condition values ​​need to be adjusted according to the actual application. For example, such as... Figure 2 As shown, the boundary condition that contributes the most to the SHAP value of the planning model solution is load demand. In actual planning, load demand needs to meet the needs of production and daily life. Therefore, adjustments to load demand must be made while still meeting the planned load requirements. Similarly, as... Figure 2 The maximum carbon emissions that significantly contribute to the solution of the planning model also need to be adjusted in terms of boundary conditions, provided that the policy requirements of the corresponding planning stage in the relevant region are met. For boundary conditions that can be adjusted through technological improvements, such as the upward adjustment capacity of thermal power units and the charging and discharging efficiency of energy storage, the boundary condition values ​​can be adjusted based on the actual achievable results of the technological improvements during actual implementation (e.g., through appropriate selection).

[0125] This embodiment discloses a multi-resource collaborative planning method for power systems based on attribution analysis. Attribution analysis determines the degree of influence of each boundary condition or constraint on the planning scheme. Based on the different degrees of influence, the boundary conditions corresponding to the boundary conditions that have a significant impact on the planning scheme or the constraints that have a significant impact on the planning decision are adjusted. The optimal power system planning scheme is then determined based on the adjusted boundary conditions. This method can scientifically formulate optimal power system planning schemes suitable for different regions and planning stages, and save construction costs and improve economic efficiency by selecting the planning scheme with the lowest planning cost.

[0126] By using the SHAP method to analyze and quantify the impact of each boundary condition or constraint on the planning scheme, the boundary conditions with the greatest impact on the planning scheme can be identified, which can guide practical applications. By improving the technology, the boundary conditions can be adjusted. For example, when it is found that the energy storage capacity has a significant impact on the planning scheme, the planning cost can be reduced by selecting energy storage equipment with a large energy storage capacity in the current plan. At the same time, it should promote the development of energy storage technology and expand energy storage capacity in future system construction.

[0127] System Implementation Examples

[0128] Another specific embodiment of the present invention discloses a multi-resource collaborative planning system for power systems based on attribution analysis, including a planning model management module, a planning model solving module, an attribution analysis module, and a planning scheme formulation module.

[0129] The planning model management module is used to construct an objective function based on the load demand of the power system and the output of each unit, with the goal of minimizing the supply-demand mismatch. It constructs a planning model based on power balance constraints, line flow constraints, output constraints of each unit, low-carbon constraints, carbon cost calculation formula, and the objective function.

[0130] The planning model solving module is used to solve the planning model based on multiple sets of boundary condition values ​​to obtain the planning scheme corresponding to each set of boundary condition values.

[0131] The attribution analysis module is used to analyze and obtain a quantitative representation of the degree of influence of each boundary condition or constraint on the planning scheme using the SHAP method.

[0132] The planning scheme formulation module is used to adjust the corresponding boundary condition values ​​based on the quantified representation, solve the planning model based on the multiple sets of boundary condition values ​​obtained after adjustment, and select the planning scheme with the lowest planning cost among the solution results as the optimal planning scheme.

[0133] Compared with the prior art, the beneficial effects provided by this embodiment are basically the same as those provided by the method embodiment, and will not be described in detail here.

[0134] It should be noted that the above embodiments are based on the same inventive concept, and any parts not described repeatedly can be referenced from each other.

[0135] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for multi-resource coordinated planning of a power system based on an attribution analysis, characterized in that, The method comprises the following steps: a target function is constructed based on the load demand and the output of each unit of the power system, with the lowest mismatch degree between supply and demand as the target, a planning model is constructed based on the power balance constraint, the line power flow constraint, the output constraint of each unit, the low-carbon constraint and the target function; a plurality of sets of boundary condition values are obtained by solving the planning model based on the plurality of sets of boundary condition values of the planning model, to obtain a planning scheme corresponding to each set of boundary condition values; The SHAP method is used to analyze the quantitative representation of the influence degree of each boundary condition and each constraint on the planning scheme, including: performing global explainability analysis on the planning model to calculate the marginal contribution SHAP value of each boundary condition as the quantitative representation of each boundary condition; performing local explainability analysis on the planning model to calculate the comprehensive slack value of each constraint as the quantitative representation of each constraint; and using the following formula to calculate the comprehensive slack value of each constraint condition: ; is the relaxation of constraint i at time resolution t; is the set of model constraints; is the maximum output of thermal power units; is the output of thermal power units at time t in the optimal solution of the model; is the integrated relaxation of constraint i in the evolution of the planning towards the optimal solution; is the set of constraint dual variables; is the dual variable corresponding to constraint i in the locally interpretable model; T is the total time length; the locally interpretable model of the planning model is represented as: ​ optimization variables are 0 / 1 variables, i.e., the operation states of thermal power units; optimization variables are 0 / 1 variables, i.e., the operation states of thermal power units; optimization variables are the solution variables of the model; is a coefficient column vector corresponding to the objective function; , , and are coefficient matrices of variables under corresponding constraint conditions; , is a constant column vector; represents the inequality constraints in the planning model; is the equality constraint in the planning model; corresponds to the constraint condition containing 0 / 1 variables in the planning model; the boundary condition values corresponding to the quantitative representation are adjusted based on the quantitative representation, the planning model is solved based on the plurality of sets of boundary condition values obtained after the adjustment, and the planning scheme with the lowest planning cost in the solving results is selected as the optimal planning scheme, comprising: the SHAP values are sorted from high to low, and the comprehensive slack values are sorted from low to high; a plurality of boundary conditions corresponding to the SHAP values ranked at the top are taken and the corresponding values are adjusted within the value range of the boundary conditions, and / or the boundary conditions included in the constraints corresponding to the comprehensive slack values ranked at the top are taken and the corresponding values are adjusted within the value range of the boundary conditions, to obtain a plurality of sets of boundary condition values, the model is solved based on the plurality of sets of boundary condition values; the planning cost corresponding to each set of boundary condition values is calculated based on the plurality of solving results; and the planning scheme corresponding to the lowest planning cost is selected as the optimal planning scheme.

2. The planning method of claim 1, wherein, The units include wind power, photovoltaic and energy storage units; the target function is constructed based on the load demand and the output of each unit of the power system, with the lowest mismatch degree between supply and demand as the target, comprising: when the sum of the wind power output and the photovoltaic output is greater than the load demand, a supply-demand mismatch power calculation formula is constructed based on the wind power output, the photovoltaic output, the load demand and the energy storage charging power; when the sum of the wind power output and the photovoltaic output is not greater than the load demand, a supply-demand mismatch power calculation formula is constructed based on the wind power output, the photovoltaic output, the load demand and the energy storage discharging power; the target function is constructed based on the supply-demand mismatch power calculation formula and the load demand of each time period, with the lowest mismatch degree between supply and demand as the target.

3. The planning method of claim 2, wherein, The boundary conditions of the planning model include: the load demand, the wind power capacity factor, the photovoltaic capacity factor, the maximum carbon emission, the carbon emission intensity of the thermal power unit, the maximum upward / downward flexible adjustment capability of the thermal power unit, the power upper limit of the thermal power unit, the transmission line transmission power upper limit, the transmission node phase angle, the long-term energy storage charging / discharging efficiency and the short-term energy storage charging / discharging efficiency.

4. The planning method of claim 3, wherein, The target function is represented as: ； ； wherein, is the system supply-demand mismatch degree, is the system supply-demand mismatch power, , are the wind power and photovoltaic power output at t, respectively, is the load demand at t, , are the charging and discharging power of the energy storage, respectively.

5. The planning method of claim 4, wherein, The maximum carbon emission is constrained by the following formula: ； wherein, represents a set of thermal power generating units in a region; represents a total time length; represents an output of the thermal power generating unit at represents a carbon emission intensity of the thermal power generating unit; represents a time interval; represents a maximum carbon emission amount allowed.​​ 6. The planning method of claim 5, wherein, The power balance constraint is represented as: ； wherein subscript represents the area; , , , respectively represent the set of thermal power units, the set of new energy units, the set of load nodes and the set of energy storage units in the area ; represents the set of tie lines between the area and other areas; is the output of the thermal power unit at time ; , are respectively the output of the wind turbine s and the photovoltaic unit at time t; and are respectively the total discharge power and the total charge power of the energy storage in the area ; is the transmission power of the tie line between the area and other areas at time ; represents the load demand of the area at time t; and respectively represent the discharge power and the charge power of the short-term energy storage n at time t, and respectively represent the discharge power and the charge power of the long-term energy storage m at time t.

7. An attribution analysis based multi-resource coordinated planning system for power systems, characterized in that, comprising: a planning model management module, configured to construct a target function based on the load demand and the output of each unit of the power system, with the lowest mismatch degree between supply and demand as the target, and to construct a planning model based on the power balance constraint, the line power flow constraint, the output constraint of each unit, the low-carbon constraint and the target function; a planning model solving module, configured to solve the planning model based on a plurality of sets of boundary condition values of the planning model, to obtain a planning scheme corresponding to each set of boundary condition values; An attribution analysis module is configured to analyze the SHAP method to obtain a quantitative representation of the influence degree of each boundary condition and each constraint on the planning scheme; the attribution analysis module comprises: performing global explainability analysis on the planning model to obtain a marginal contribution SHAP value of each boundary condition as the quantitative representation of each boundary condition; performing local explainability analysis on the planning model to obtain a comprehensive slack value of each constraint as the quantitative representation of each constraint; and using the following formula to calculate the comprehensive slack value of each constraint condition: ; is the relaxation degree of constraint i at time resolution t; is the set of model constraints; is the maximum output of thermal power units; is the output of thermal power units at time t in the optimal solution of the model; is the comprehensive relaxation degree of constraint i in the evolution process of the planning to the optimal solution; is the set of constraint dual variables; is the dual variable corresponding to each constraint in the locally interpretable model; T is the total time length; the locally interpretable model of the planning model is expressed as: ; is the optimization variable, the optimization variable in is a 0 / 1 variable, i.e., the operation state of a thermal power unit; the optimization variable in is a solution variable of the model; is the coefficient column vector corresponding to the objective function; , , and is the coefficient matrix of the variable under the corresponding constraint; , is the constant column vector; represents the inequality constraint in the planning model; is the equality constraint in the planning model; corresponds to the constraint condition containing the 0 / 1 variable in the planning model; The planning scheme formulation module is configured to adjust the boundary condition values corresponding to the quantized representation, solve the planning model based on the adjusted boundary condition values, and select the planning scheme with the lowest planning cost as the optimal planning scheme. The method comprises the following steps: sorting the SHAP values from high to low, and sorting the comprehensive slack values from low to high; adjusting the boundary conditions corresponding to the SHAP values at the top of the list within their value ranges, and / or adjusting the boundary conditions included in the constraints corresponding to the comprehensive slack values at the top of the list within their value ranges, to obtain multiple sets of boundary condition values; solving the model based on the multiple sets of boundary condition values; calculating the planning cost based on the multiple sets of solving results; and selecting the planning scheme corresponding to the lowest planning cost as the optimal planning scheme.

Images