A marginal unit positioning method and positioning system based on dual simplex sensitivity analysis

Abstract

The present invention discloses a marginal unit positioning method and positioning system based on dual simplex sensitivity analysis. Based on the dual simplex algorithm and sensitivity analysis principle, the marginal unit positioning is realized, and it is suitable for various situations such as blocked sections, branches, and equipment within the limit and beyond the limit. It has universality, small calculation amount, and high accuracy. It breaks the limitations of traditional marginal unit positioning and provides a powerful analysis method for power spot market price analysis.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric power dispatching automation, and in particular relates to a marginal unit positioning method and a positioning system based on dual simplex sensitivity analysis. Background Art

[0002] At present, the first batch of 8 domestic electricity spot pilot units have successively completed the long-term uninterrupted settlement trial operation, further testing the integrity and adaptability of market rules and technical support systems. Some of the first batch of pilot units have transformed from bidding and clearing on the power generation side to centralized bidding and clearing on both the power generation and consumption sides, cultivating the awareness of market entities on both the power generation and consumption sides, and accumulating valuable experience for the formal operation of the spot market in the future.

[0003] Due to the differences in grid architecture and power technology characteristics of the first batch of spot pilot units, their market models, congestion management methods, pricing mechanisms, and analysis methods are different. By fully summarizing the operating experience of each pilot unit, it is found that the most sensitive, complex and difficult to analyze in the clearing results is the clearing price, and the difficulty of analyzing the node price ranks first among various price mechanisms. The node electricity price is the minimum increment of the total system cost caused by an increase of one unit in the load demand of a certain node under the condition of meeting certain constraints (such as system constraints, unit operation constraints, network security constraints, etc.). The typical definition of the node price includes three components: system energy price, congestion price and network loss price. The system energy price reflects the power supply and demand relationship of the entire market, the congestion price reflects the grid congestion situation at the node location, and the network loss price reflects the degree of influence of the node location on the transmission loss of the grid. Since the network loss cost is generally considered when the transmission and distribution price is approved, the network loss price does not need to be charged in the node electricity price, so the node price is composed of two components: system energy price and congestion price. The system energy price in the node price can also be regarded as the price of the system reference node. The system energy price in each node is the same, and the difference is the congestion price. Therefore, the system energy price reflects the minimum cost paid by the system to meet the system unit load increase under the current operating conditions, and the marginal unit refers to the set of units called for the current power generation to increase the unit power. Therefore, the positioning of the marginal unit is of great significance for analyzing the fluctuation of the system energy price, studying network congestion, and analyzing the operation status of the power market.

[0004] According to the definition of system energy price, when the system is not blocked, in the centralized spot clearing scenario with the minimum electricity purchase cost as the optimization goal, the units are cleared in order from low to high according to the declared price until the system electricity demand is met. The declared price of the last cleared unit is the marginal energy price of the system, and this unit is called the marginal unit. Since the system is not blocked, the blocking price is zero. According to the definition of node price, the node prices of the entire network are the same, which is equal to the system energy price. However, when the system is blocked, since the blocking price is not zero, the node prices of the entire network are no longer unified. The clearing order of the units is affected by the system blockage and is no longer cleared in order from low to high according to the declared price. The system energy price is jointly determined by the declared prices of multiple units. The number of marginal units in the system is no longer unique. The number of marginal units is equal to the number of blocked sections\branches\equipment plus one. It is impossible to locate the marginal units by sorting the declared prices. The inability to accurately locate the marginal units directly affects the accurate analysis of the fluctuation of the system energy price.

[0005] The existing method for locating priced units can only give the maximum possible set of marginal units, but cannot accurately locate marginal units, and can only provide auxiliary analysis and decision-making. The existing marginal unit search method based on node price definition requires that the blocking section, branch, and equipment of the pricing link system must operate within the limit and cannot be relaxed beyond the limit, so it is not universal. Some experts and scholars have proposed methods such as locating priced units and locating marginal units based on node price definition, but none of them are universal. Summary of the invention

[0006] Purpose of the invention: In order to solve the problems of inaccurate positioning of marginal units in the prior art and inability to locate in special circumstances, the present invention is based on the positioning method of the dual simplex algorithm of each pilot unit, and proposes a marginal unit positioning method and positioning system based on dual simplex sensitivity analysis, which can quickly and effectively locate marginal units under different circumstances, and provide a fast and effective positioning means for system energy price volatility analysis.

[0007] Technical solution: A method for locating marginal units based on dual simplex sensitivity analysis, comprising the following steps:

[0008] Step 1: Obtain basic spot market data;

[0009] Step 2: Establish a clearing model that complies with spot market rules;

[0010] Step 3: Based on the spot market basic data in step 1, solve the clearing model constructed in step 2 to obtain the unit combination and unit output in each period;

[0011] Step 4: Establish a node price calculation model based on the unit combination and unit output in each period;

[0012] Step 5: Solve the node price calculation model constructed in step 4 to obtain the Lagrange multipliers of the system load balance constraint, line / section flow constraint and system reserve demand constraint in each time period. Based on each Lagrange multiplier, obtain the node electricity price of node n in time period t;

[0013] Step 6: Based on the node price calculation model, the dual simplex sensitivity analysis method is used to locate the marginal units;

[0014] Wherein, step 6 specifically includes the following sub-steps:

[0015] S610: The node price calculation model is problem 1. Problem 2 is constructed based on problem 1. Based on problem 1 and problem 2, the corresponding dual problems are obtained, namely, dual problem 1 and dual problem 2. Dual problem 1 and dual problem 2 have exactly the same feasible domain.

[0016] Among them, question 1 is expressed as:

[0017]

[0018] In the formula, A represents the constraint coefficient matrix, b represents the constant term on the right side of the constraint, C represents the price coefficient, and X represents the decision variable;

[0019] Question 2 is expressed as:

[0020]

[0021] Where d represents the perturbation vector of the constant on the right side of the constraint condition;

[0022] S620: Assume that the optimal basis for problem 1 is B, B -1 is the optimal basis inverse matrix. Based on formula (29), the range θ of the disturbance vector d is calculated;

[0023] B -1 (b+d)≥0 (29)

[0024] S630: Ensure that the disturbance vector d is within the range θ, and calculate the system energy price through equation (31);

[0025]

[0026] Where, d j represents the constant change on the right side of the power generation and consumption balance constraint in period j, y * represents the optimal solution to the dual problem 1, represents the system energy price for time period j;

[0027] S640: Since B is the optimal basis for problem 1, we know that B -1b is the optimal solution to problem 1. When the perturbation vector d is within the range θ, problem 1 becomes problem 2, the optimal basis remains unchanged, and the optimal solution to problem 2 is:

[0028] B -1 (b+d)=B -1 b+B -1 d (32)

[0029] Among them, B -1 The basic variable corresponding to the non-zero term of d is the marginal unit.

[0030] The present invention also discloses a marginal unit positioning system based on dual simplex sensitivity analysis, comprising:

[0031] Basic data acquisition module, used to obtain basic data of spot market;

[0032] The clearing model building module is used to build a clearing model that complies with spot market rules;

[0033] The module for calculating the unit combination and unit output in each period is used to solve the clearing model constructed by the clearing model construction module according to the spot market basic data obtained by the basic data acquisition module, and obtain the unit combination and unit output in each period;

[0034] A node price calculation model building module is used to establish a node price calculation model based on the unit combination and unit output of each time period obtained by the unit combination and unit output calculation module of each time period;

[0035] The node electricity price calculation module is used to solve the node price calculation model constructed by the node price calculation model construction module, obtain the Lagrange multipliers of the system load balance constraints, line / section flow constraints and system reserve demand constraints in each time period, and obtain the node electricity price of node n in time period t based on each Lagrange multiplier;

[0036] The marginal unit positioning module is used to locate marginal units using the dual simplex sensitivity analysis method.

[0037] Beneficial effects: Based on the dual simplex algorithm and sensitivity analysis principle, the present invention proposes a marginal unit positioning method and positioning system based on dual simplex sensitivity analysis. The theoretical basis is mature, and it is suitable for various situations such as blocked sections, branches, and equipment within the limit and beyond the limit. It has universality, small calculation amount, and high accuracy. It breaks the limitations of traditional marginal unit positioning and provides a powerful analysis method for power spot market price analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a flow chart. DETAILED DESCRIPTION

[0039] The technical solution of the present invention is now further described in conjunction with the accompanying drawings and embodiments.

[0040] like Figure 1 A marginal unit location method based on dual simplex sensitivity analysis is shown, comprising the following steps:

[0041] Step 1: Obtain basic spot market data, which includes but is not limited to system data, unit data, interconnection line planning data, load data, unit group data and sensitivity data; wherein, system data includes time period information, system load and system reserve requirements; unit data includes basic unit information, unit calculation parameters, unit energy quotation, unit initial state, unit power constraint and unit ramp rate; interconnection line planning data includes basic interconnection line information and interconnection line planned power; load data includes bus load forecast; unit group data includes unit group power limit and unit group power limit; sensitivity data includes power generation transfer distribution factor of unit and load injection power to line and section flow, and sensitivity data is obtained by obtaining the latest power grid physical model and real-time operation mode data, and calculated by PQ decoupling method.

[0042] Step 2: Establish a clearing model that complies with spot market rules, specifically expressed as:

[0043] The objective function of the spot market clearing model is:

[0044]

[0045] Where: N represents the total number of units; T represents the total number of time periods; P i,t represents the winning bid output of unit i in period t; C i,t (P i,t ) is the operating cost of unit i in period t; is the startup cost of unit i in period t; M b represents the penalty factor of the power generation and consumption balance constraint used for market clearing optimization; They represent the positive slack variable and the reverse slack variable of the power generation and consumption balance constraint period t respectively; M r represents the penalty factor of the system reserve demand constraint used for market clearing optimization; They represent the slack variables of the system's positive reserve demand constraint period t and negative reserve demand constraint period t respectively; M l represents the penalty factor for the network power flow constraint of the market clearing optimization line l; They represent the forward power flow relaxation variable and reverse power flow relaxation variable of line l respectively; NL is the total number of lines; M s represents the penalty factor of the network power flow constraint for market clearing optimization section s; They represent the forward flow relaxation variable and reverse flow relaxation variable of section s respectively; NS represents the total number of sections.

[0046] The constraints of the clearing model that conforms to the spot market rules include system power generation and consumption balance constraints, system reserve demand constraints, unit operation constraints, unit cost constraints, unit group constraints and network security constraints.

[0047] Among them, the system power generation and consumption balance constraint condition is expressed as:

[0048]

[0049] Where, T j,t represents the planned power of tie line j in time period t; NT is the total number of tie lines; D t is the system load during period t;

[0050] Among them, the system reserve demand constraints include positive reserve and negative reserve, and the expression is:

[0051]

[0052]

[0053] In the formula, ΔP i u is the maximum ramp rate of unit i, ΔP i D is the maximum down-slope rate of unit i; are the maximum output and minimum output of unit i in period t+1 respectively; The spinning reserve requirement is adjusted up and down for period t respectively.

[0054] Among them, the unit operation constraints include the upper and lower limit constraints of the unit output, the maximum number of start-up and shutdown times of the unit, the minimum start-up and shutdown time constraints of the unit, and the climbing and sliding constraints of the unit.

[0055] Among them, the upper and lower limit constraint expressions of the unit output are:

[0056]

[0057] In the formula, α i,t Represents the operating status of unit i in period t, which is a 0 / 1 variable; if the unit is in shutdown state, then α i,t = 0, through this constraint condition, the unit output can be limited to 0; if the unit is in the start-up state, then α i,t =1, this constraint is the conventional upper and lower limit constraint of output; are the maximum output and minimum output of unit i in period t respectively.

[0058] Among them, the maximum start-stop times constraint of the unit is expressed as:

[0059]

[0060] In the formula, are the maximum number of starts and stops of unit i respectively; η i,t Is the unit i switched to the startup state in period t, a 0 / 1 variable, 0 means no startup switching, 1 means startup switching; γ i,t Indicates whether unit i is switched to shutdown state during period t, a 0 / 1 variable, 0 represents no shutdown switching, and 1 represents shutdown switching.

[0061] Among them, the minimum start-up and shutdown time constraint of the unit is expressed as:

[0062]

[0063]

[0064] Where, T U , T D is the minimum continuous start time and minimum continuous shutdown time of the unit, α i,k Represents the operating status of unit i in time period k.

[0065] Among them, the unit climbing and sliding constraint is expressed as:

[0066]

[0067]

[0068] In the formula, ΔP i U is the maximum ramp rate of unit i, ΔP i D is the maximum ramp rate of unit i, P i,t-1 is the output of unit i in period t-1.

[0069] Among them, unit cost constraints include unit output constraints, unit operating cost constraints, unit start-up and shutdown cost constraints, and unit shutdown cost constraints.

[0070] Among them, the unit output constraint expression is:

[0071]

[0072]

[0073] Where, M is the total number of sections in the unit quotation, P i,t,m is the winning bid power of unit i in the mth output interval in period t, They are respectively the upper and lower bounds of the mth output range declared by unit i.

[0074] Among them, the unit operation cost constraint expression is:

[0075]

[0076] Where, M is the total number of sections in the unit quotation, C i,t,m It is the energy price corresponding to the mth output segment declared by unit i in period t.

[0077] Among them, the constraint expression of unit start-up and shutdown cost is:

[0078]

[0079] In the formula, is the single startup cost of unit i, is the downtime cost of unit i in period t.

[0080] Among them, the constraint expression of unit shutdown cost is:

[0081]

[0082] In the formula, is the single shutdown cost of unit i, is the downtime cost of unit i in period t.

[0083] Among them, the unit group constraints include unit group output constraints and unit group power constraints.

[0084] Among them, the output constraint expression of the unit group is:

[0085]

[0086] In the formula, are the maximum and minimum outputs of generator group j in time period t.

[0087] Among them, the power constraint expression of the unit group is:

[0088]

[0089] In the formula, is the upper limit of power consumption of unit group j on the clearing day.

[0090] Among them, network security constraints include line flow constraints and section flow constraints.

[0091] The line flow constraint expression is:

[0092]

[0093] Where Pl max is the power transmission limit of line l; G l-i is the generator output power transfer distribution factor of the node where unit i is located to line l; G l-j is the power transfer distribution factor of the generator output from the node of tie line j to line l; K is the number of nodes in the system; G l-k is the generator output power transfer distribution factor of node k to line l; D k,t is the bus load value of node k in period t.

[0094] The section flow constraint expression is:

[0095]

[0096] In the formula, are the minimum transmission limit and maximum transmission limit of the power flow in section s respectively; G s-i G is the generator output power transfer distribution factor of the node where unit i is located to section s; s-j G is the generator output power transfer distribution factor of the node of tie line j to section s; s-k is the generator output power transfer distribution factor for node k to section s.

[0097] Step 3: Based on the basic data in step 1, a commercial optimization engine is used to solve the spot market clearing model constructed in step 2 to obtain the unit combination and unit output in each period;

[0098] Step 4: Establish a node price calculation model based on the unit combination and unit output in each period;

[0099] Among them, the optimization objective expression of the node price calculation model is:

[0100]

[0101] In the formula, N represents the total number of units; T represents the total number of time periods; P i,t represents the winning bid output of unit i in period t; C i,t (P i,t ) is the operating cost of unit i in period t; M b represents the penalty factor of the power generation and consumption balance constraint used for market clearing optimization; They represent the positive slack variable and the reverse slack variable of the power generation and consumption balance constraint period t respectively; M r represents the penalty factor of the system reserve demand constraint used for market clearing optimization; They represent the slack variables of the system's positive reserve demand constraint period t and negative reserve demand constraint period t respectively; M lrepresents the penalty factor for the network power flow constraint of the market clearing optimization line l; They represent the forward power flow relaxation variable and reverse power flow relaxation variable of line l respectively; NL is the total number of lines; M s represents the penalty factor of the network power flow constraint for market clearing optimization section s; They represent the forward flow relaxation variable and reverse flow relaxation variable of section s respectively; NS represents the total number of sections.

[0102] The optimization objectives of the node price calculation model and the market clearing model are basically the same. The differences are as follows: (1) Since the unit combination in the node price calculation model has been determined, the startup and shutdown costs are no longer considered in the optimization objectives; (2) The penalty factors of the clearing model and the node price calculation model are different in magnitude. In the market clearing stage, in order to ensure the safe and stable operation of the power grid, the network constraint penalty factor is set very high, but it cannot be used for congestion price calculation. Therefore, the size of the penalty factor in the pricing stage will be different.

[0103] The constraints of the node price calculation model include formula (2) to formula (5), formula (9) to formula (13) and formula (16) to formula (19).

[0104] Since the unit combination in the node price calculation model is a known quantity, the constraints such as the minimum start-up and shutdown time, the maximum number of startups, and the minimum number of startups are no longer considered. Since the node price calculation link is based on the unit output of the market clearing link, the node price calculation link sets the unit output adjustment range to a small bandwidth of the clearing output.

[0105] Step 5: Solve the node price calculation model constructed in step 4 to obtain the Lagrange multipliers of the system load balance constraints, line / section flow constraints and system reserve demand constraints in each time period. Based on each Lagrange multiplier, the node electricity price of node n in time period t is obtained as follows:

[0106]

[0107] In the formula, λ t The Lagrange multiplier representing the system load balance constraint during time period t; is the Lagrangian multiplier of the maximum forward power flow constraint of line l in time period t. When the line power flow exceeds the limit, the Lagrangian multiplier is the penalty factor for the relaxation of the network power flow constraint; is the Lagrangian multiplier of the maximum reverse power flow constraint of line l in time period t. When the line power flow exceeds the limit, the Lagrangian multiplier is the penalty factor for the relaxation of the network power flow constraint; is the Lagrange multiplier of the maximum positive power flow constraint of section s in time period t. When the section power flow exceeds the limit, the Lagrange multiplier is the penalty factor for the relaxation of the network power flow constraint; is the Lagrange multiplier of the maximum reverse power flow constraint of section s in time period t. When the section power flow exceeds the limit, the Lagrange multiplier is the penalty factor for the relaxation of the network power flow constraint; G l-n is the generator output power transfer distribution factor of node n to line l; G s-n is the generator output power transfer distribution factor for node n to section s.

[0108] Step 6: Locate marginal units based on the dual simplex sensitivity analysis method; specifically, it includes the following sub-steps:

[0109] First, the general form of the node price calculation model is given:

[0110]

[0111] In the formula: A represents the constraint coefficient matrix (constraint coefficients such as system balance, unit operation, and network security); b represents the constant term on the right side of the constraint (such as system load in power generation and consumption balance constraints, section limit in network security constraints, etc.); C represents the price coefficient (including unit declared price, relaxation constraint penalty coefficient, etc.); X represents the decision variable (including unit output, relaxation variable, etc.). The above model is problem 1, the model type is a linear programming model, and the calculation method adopts the dual simplex method.

[0112] In order to more clearly illustrate the necessity of sensitivity analysis, question 2 is constructed based on question 1:

[0113]

[0114] Where: d can be regarded as the disturbance vector of the constant on the right side of the constraint condition, for example, the power load disturbance on the right side is 1MW.

[0115] Next, write the dual problem of Problem 1 and Problem 2, calling them Dual Problem 1 and Dual Problem 2.

[0116] Dual Problem 1:

[0117]

[0118] Dual Problem 2:

[0119]

[0120] It can be seen that the dual problem 1 and the dual problem 2 have exactly the same feasible domain.

[0121] If we assume that y * is the optimal solution to the dual problem 1, and is also a feasible solution to the dual problem 2. According to the weak duality theorem, we know that:

[0122] maxZ2≤y *(b+d)=y * b+y * d=maxZ1+y * d (26)

[0123] Assume that the perturbation vector d has only the jth element d j Non-zero (can also be understood as the load disturbance d in period j j ), and the other elements are all zero, it can be further expressed as:

[0124]

[0125] The left side of the above equation represents the constant term perturbation d on the right side of question 1 j The change in unit cost is consistent with the definition of shadow price. Assuming the disturbance d j is the constant change on the right side of the power generation and consumption balance constraint in period j, then the unit cost change in the above formula can be understood as the shadow price of the balance constraint, that is, the system energy price (the system energy price reflects the minimum cost paid by the system to meet the system unit load under the current operating conditions). * As the optimal solution to the dual problem 1, is the shadow price of the balance constraint in problem 1, that is, the system energy price in time period j. Therefore, the unit cost change calculated by formula (27) cannot be strictly called the shadow price of the power generation and consumption balance constraint (that is, the system energy price).

[0126] To ensure that the unit cost change calculated by equation (27) is equal to the optimal solution of the dual problem 1 It is necessary to ensure that the optimal basis remains unchanged after the perturbation vector d in problem 1. Assuming that the optimal basis in problem 1 is B, the optimal judgment condition satisfies:

[0127]

[0128] Where: B is the optimal basis for problem 1; N is the matrix consisting of the remaining columns after removing the optimal basis from the matrix coefficient A; C N is the price coefficient of the non-basic variable; C B is the price coefficient of the base variable; B -1 is the optimal basis inverse matrix.

[0129] It can be seen from formula (28) that after the constant term perturbation d on the right side of problem 1 becomes problem 2, if we want to ensure that the optimal basis remains unchanged, we only need to satisfy formula (29):

[0130] B -1 (b+d)≥0 (29)

[0131] Since b is a constant term, the optimal basis inverse matrix B -1It has been obtained in the node price calculation in step 4, so the range θ of the disturbance vector d can be calculated by formula (29). As long as the disturbance d is within θ, the optimal basis can be guaranteed to remain unchanged. At this time, the following conclusion can be obtained:

[0132] maxZ2=y * (b+d)=y * b+y * d=maxZ1+y * d (30)

[0133] It can be further expressed as:

[0134]

[0135] So far, the above content proves that after the disturbance range θ is analyzed through sensitivity, as long as the disturbance d is within the range of θ, the system energy price can be calculated by formula (31), and then the marginal unit can be further located.

[0136] The system energy price reflects the minimum cost to meet the increase in system unit load under the current operating conditions, and the marginal unit refers to the set of units called upon to increase the current unit power of power generation.

[0137] Since B is the optimal basis for Problem 1, we know that B -1 b is the optimal solution to problem 1. After the constant term perturbation d (within the range of θ) on the right side becomes problem 2, since the optimal basis remains unchanged, the optimal solution to problem 2 is:

[0138] B -1 (b+d)=B -1 b+B -1 d (32)

[0139] From formula (32), we can see that B -1 The basic variable corresponding to the non-zero term of d is the marginal unit.

[0140] At this point, step 6 gives the marginal unit location method based on dual simplex sensitivity analysis.

[0141] Step 7: Output the unit combination and unit output plan for each period of the market clearing in step 3, and send them to AGC for tracking and execution; output the congestion price and node electricity price calculated in step 5, and send them to the trading center to provide basic data for market calculation; output the marginal unit located in step 6 to provide a basis for the analysis of system energy price, congestion price, and node price fluctuations. The Lagrange multiplier of the line / section mentioned in step 5 is the congestion price.

[0142] Based on the above positioning method, a marginal unit positioning system based on dual simplex sensitivity analysis is proposed, including:

[0143] Basic data acquisition module, used to obtain basic data of spot market;

[0144] The clearing model building module is used to build a clearing model that complies with spot market rules;

[0145] The module for calculating the unit combination and unit output in each period is used to solve the clearing model constructed by the clearing model construction module according to the spot market basic data obtained by the basic data acquisition module, and obtain the unit combination and unit output in each period;

[0146] A node price calculation model building module is used to establish a node price calculation model based on the unit combination and unit output of each time period obtained by the unit combination and unit output calculation module of each time period;

[0147] The node electricity price calculation module is used to solve the node price calculation model constructed by the node price calculation model construction module, obtain the Lagrange multipliers of the system load balance constraints, line / section flow constraints and system reserve demand constraints in each time period, and obtain the node electricity price of node n in time period t based on each Lagrange multiplier;

[0148] The marginal unit positioning module is used to locate marginal units using the dual simplex sensitivity analysis method.

Claims

1. A method for locating marginal units based on dual simplex sensitivity analysis, characterized by: The following steps are involved: Step 1: Obtain basic spot market data; Step 2: Establish a clearing model that complies with spot market rules; Step 3: Based on the spot market basic data in step 1, solve the clearing model constructed in step 2 to obtain the unit combination and unit output in each period; Step 4: Establish a node price calculation model based on the unit combination and unit output in each period; Step 5: Solve the node price calculation model constructed in step 4 to obtain the Lagrange multipliers of the system load balance constraint, line / section flow constraint and system reserve demand constraint in each time period. Based on each Lagrange multiplier, obtain the node electricity price of node n in time period t; Step 6: Based on the node price calculation model, the dual simplex sensitivity analysis method is used to locate the marginal units; Wherein, step 6 specifically includes the following sub-steps: S610: The node price calculation model is problem 1. Problem 2 is constructed based on problem 1. Based on problem 1 and problem 2, the corresponding dual problems are obtained, namely, dual problem 1 and dual problem 2. Dual problem 1 and dual problem 2 have exactly the same feasible domain. Among them, question 1 is expressed as: In the formula, A represents the constraint coefficient matrix, b represents the constant term on the right side of the constraint, C represents the price coefficient, and X represents the decision variable; Question 2 is expressed as: Where d represents the perturbation vector of the constant on the right side of the constraint condition; S620: Assume that the optimal basis for problem 1 is B, B -1 is the optimal basis inverse matrix. Based on formula (29), the range of the perturbation vector d is calculated. B -1 (b+d)≥0 (29) S630: Ensure that the disturbance vector d is within the range The system energy price is calculated by formula (31); Where, d j represents the constant change on the right side of the power generation and consumption balance constraint in period j, y * represents the optimal solution to the dual problem 1, represents the system energy price for time period j; S640: Since B is the optimal basis for problem 1, we know that B -1 b is the optimal solution to problem 1. When the disturbance vector d is in the range After problem 1 becomes problem 2, the optimal basis remains unchanged, and the optimal solution of problem 2 is: B -1 (b+d)=B -1 b+B -1 d (32) where B -1 The basic variable corresponding to the non-zero term of d is the marginal unit; In step 2, the objective function of the spot market clearing model is: Where: N represents the total number of units; T represents the total number of time periods; P i,t represents the winning bid output of unit i in period t; C i,t (P i,t ) is the operating cost of unit i in period t; is the startup cost of unit i in period t; M b represents the penalty factor of the power generation and consumption balance constraint used for market clearing optimization; They represent the positive slack variable and the reverse slack variable of the power generation and consumption balance constraint period t respectively; M r represents the penalty factor of the system reserve demand constraint used for market clearing optimization; They represent the slack variables of the system's positive reserve demand constraint period t and negative reserve demand constraint period t respectively; M l represents the penalty factor for the network power flow constraint of the market clearing optimization line l; They represent the forward power flow relaxation variable and reverse power flow relaxation variable of line l respectively; NL is the total number of lines; M s represents the penalty factor of the network power flow constraint for market clearing optimization section s; They represent the forward flow relaxation variable and the reverse flow relaxation variable of section s respectively; NS represents the total number of sections; The constraints of the clearing model include system power generation and consumption balance constraints, system reserve demand constraints, unit operation constraints, unit cost constraints, unit group constraints and network security constraints; Based on the basic data in step 1, a commercial optimization engine is used to solve the clearing model constructed in step 2 to obtain the unit combination and unit output in each period; In step 4, the optimization objective expression of the node price calculation model is: In the formula, N represents the total number of units; T represents the total number of time periods; P i,t represents the winning bid output of unit i in period t; C i,t (P i,t ) is the operating cost of unit i in period t; M b represents the penalty factor of the power generation and consumption balance constraint used for market clearing optimization; They represent the positive slack variable and the reverse slack variable of the power generation and consumption balance constraint period t respectively; M r represents the penalty factor of the system reserve demand constraint used for market clearing optimization; They represent the slack variables of the system's positive reserve demand constraint period t and negative reserve demand constraint period t respectively; M l represents the penalty factor for the network power flow constraint of the market clearing optimization line l; They represent the forward power flow relaxation variable and reverse power flow relaxation variable of line l respectively; NL is the total number of lines; M s represents the penalty factor of the network power flow constraint for market clearing optimization section s; They represent the forward flow relaxation variable and the reverse flow relaxation variable of section s respectively; NS represents the total number of sections; The constraints of the node price calculation model include system power generation and consumption balance constraints, system reserve demand constraints, unit output upper and lower limit constraints, unit climbing and sliding constraints, unit output constraints, unit operating cost constraints, unit group output constraints, unit group power constraints and network security constraints.

2. The method for locating marginal units based on dual simplex sensitivity analysis according to claim 1 is characterized in that: In step 1, the spot market basic data includes system data, unit data, interconnection line planning data, load data, unit group data and sensitivity data.

3. The method for locating marginal units based on dual simplex sensitivity analysis according to claim 2 is characterized in that: The system data includes time period information, system load and system backup requirements; the unit data includes basic unit information, unit calculation parameters, unit energy quotation, unit initial state, unit power constraints and unit climbing rate; the interconnection line planning data includes basic interconnection line information and interconnection line planned power; the load data includes bus load forecast; the unit group data includes unit group power limit and unit group power limit; the sensitivity data includes the power generation transfer distribution factor of the unit and load injection power to the line and section flow.

4. The method for locating marginal units based on dual simplex sensitivity analysis according to claim 1 is characterized in that: In step 5, the node electricity price of node n in time period t is expressed as: In the formula, λ t The Lagrange multiplier representing the system load balance constraint during time period t; is the Lagrange multiplier of the maximum forward power flow constraint of line l in time period t. When the line power flow exceeds the limit, the Lagrange multiplier is the penalty factor for the relaxation of the network power flow constraint; is the Lagrangian multiplier of the maximum reverse power flow constraint of line l in time period t. When the line power flow exceeds the limit, the Lagrangian multiplier is the penalty factor for the relaxation of the network power flow constraint; is the Lagrange multiplier of the maximum positive power flow constraint of section s in time period t. When the section power flow exceeds the limit, the Lagrange multiplier is the penalty factor for the relaxation of the network power flow constraint; is the Lagrange multiplier of the maximum reverse power flow constraint of section s in time period t. When the section power flow exceeds the limit, the Lagrange multiplier is the penalty factor for the relaxation of the network power flow constraint; G l-n is the generator output power transfer distribution factor of node n to line l; G s-n is the generator output power transfer distribution factor for node n to section s.

5. A marginal unit positioning system based on dual simplex sensitivity analysis, characterized by: include: Basic data acquisition module, used to obtain basic data of spot market; The clearing model building module is used to build a clearing model that complies with spot market rules; The module for calculating the unit combination and unit output in each period is used to solve the clearing model constructed by the clearing model construction module according to the spot market basic data obtained by the basic data acquisition module, and obtain the unit combination and unit output in each period; A node price calculation model building module is used to establish a node price calculation model based on the unit combination and unit output of each time period obtained by the unit combination and unit output calculation module of each time period; The node electricity price calculation module is used to solve the node price calculation model constructed by the node price calculation model construction module, obtain the Lagrange multipliers of the system load balance constraints, line / section flow constraints and system reserve demand constraints in each time period, and obtain the node electricity price of node n in time period t based on each Lagrange multiplier; Marginal unit positioning module, used to locate marginal units using dual simplex sensitivity analysis method; The objective function of the clearing model is: Where: N represents the total number of units; T represents the total number of time periods; P i,t represents the winning bid output of unit i in period t; C i,t (P i,t ) is the operating cost of unit i in period t; is the startup cost of unit i in period t; M b represents the penalty factor of the power generation and consumption balance constraint used for market clearing optimization; They represent the positive slack variable and the reverse slack variable of the power generation and consumption balance constraint period t respectively; M r represents the penalty factor of the system reserve demand constraint used for market clearing optimization; They represent the slack variables of the system's positive reserve demand constraint period t and negative reserve demand constraint period t respectively; M l represents the penalty factor for the network power flow constraint of the market clearing optimization line l; They represent the forward power flow relaxation variable and reverse power flow relaxation variable of line l respectively; NL is the total number of lines; M s represents the penalty factor of the network power flow constraint for market clearing optimization section s; They represent the forward flow relaxation variable and the reverse flow relaxation variable of section s respectively; NS represents the total number of sections; The constraints of the clearing model include system power generation and consumption balance constraints, system reserve demand constraints, unit operation constraints, unit cost constraints, unit group constraints and network security constraints; The optimization objective expression of the node price calculation model is: In the formula, N represents the total number of units; T represents the total number of time periods; P i,t represents the winning bid output of unit i in period t; C i,t (P i,t ) is the operating cost of unit i in period t; M b represents the penalty factor of the power generation and consumption balance constraint used for market clearing optimization; They represent the positive slack variable and the reverse slack variable of the power generation and consumption balance constraint period t respectively; M r represents the penalty factor of the system reserve demand constraint used for market clearing optimization; They represent the slack variables of the system's positive reserve demand constraint period t and negative reserve demand constraint period t respectively; M l represents the penalty factor for the network power flow constraint of the market clearing optimization line l; They represent the forward power flow relaxation variable and reverse power flow relaxation variable of line l respectively; NL is the total number of lines; M s represents the penalty factor of the network power flow constraint for market clearing optimization section s; They represent the forward flow relaxation variable and the reverse flow relaxation variable of section s respectively; NS represents the total number of sections; The constraints of the node price calculation model include system power generation and consumption balance constraints, system reserve demand constraints, unit output upper and lower limit constraints, unit climbing and sliding constraints, unit output constraints, unit operating cost constraints, unit group output constraints, unit group power constraints and network security constraints.

6. The marginal unit positioning system based on dual simplex sensitivity analysis according to claim 5 is characterized in that: The basic data of the spot market include system data, unit data, interconnection line planning data, load data, unit group data and sensitivity data. The system data include time period information, system load and system backup requirements; the unit data include basic unit information, unit calculation parameters, unit energy quotation, unit initial state, unit power constraints and unit climbing rate; the interconnection line planning data include basic interconnection line information and interconnection line planned power; the load data include bus load forecast; the unit group data include unit group power limit and unit group power limit; the sensitivity data include the power generation transfer distribution factor of the unit and load injection power to the line and section flow.

7. The marginal unit positioning system based on dual simplex sensitivity analysis according to claim 5 is characterized by: The node electricity price of node n in time period t is expressed as: In the formula, λ t The Lagrange multiplier representing the system load balance constraint during time period t; is the Lagrange multiplier of the maximum forward power flow constraint of line l in time period t. When the line power flow exceeds the limit, the Lagrange multiplier is the penalty factor for the relaxation of the network power flow constraint; is the Lagrangian multiplier of the maximum reverse power flow constraint of line l in time period t. When the line power flow exceeds the limit, the Lagrangian multiplier is the penalty factor for the relaxation of the network power flow constraint; is the Lagrange multiplier of the maximum positive power flow constraint of section s in time period t. When the section power flow exceeds the limit, the Lagrange multiplier is the penalty factor for the relaxation of the network power flow constraint; is the Lagrange multiplier of the maximum reverse power flow constraint of section s in time period t. When the section power flow exceeds the limit, the Lagrange multiplier is the penalty factor for the relaxation of the network power flow constraint; G l-n is the generator output power transfer distribution factor of node n to line l; G s-n is the generator output power transfer distribution factor for node n to section s.

Images