| CPC G06Q 50/06 (2013.01) [G06Q 10/04 (2013.01)] | 15 Claims |
|
1. A method for performing multi-stage multi-level cooperative operation on a park integrated energy system containing a mineral industry user, comprising:
constructing a short-time scale dynamic pricing model, wherein the short-time scale dynamic pricing model comprises an upper-layer operation optimization model and a lower-layer user demand response model, the operation optimization model is constructed to minimize the sum of integrated operation costs at a day-ahead stage and an intraday stage, and the user demand response model is constructed to minimize the sum of an actual electricity-purchasing cost of the user and the electricity benefit loss in the demand response;
solving the short-time scale dynamic pricing model based on a Karush-Kuhn-Tucker condition and a Benders decomposition method to obtain day-ahead and intraday scheduling plans and a next-day electricity price; and
based on the day-ahead and intraday scheduling plans and the next-day electricity price, computing a strategy transfer probability distribution of each market subject within a preset deadline by improved regret degree matching mechanism, and simulating a decision-making process by a Monte Carlo method until each market subject reaches a correlated equilibrium state to obtain a pricing solution of the park integrated energy system within the preset deadline, wherein the operation optimization model comprises a first objective function; the first objective function comprises:
min Call=Ctrade+Cst+Ccoal+CoperateCFU-RTO+CCO2CFU-RTOCvam+CRT
wherein Call represents the sum of integrated operation costs at the day-ahead stage and the intraday stage, Ctrade represents a day-ahead market electricity-purchasing cost of the park integrated energy system and a net electricity-purchasing cost of the electricity selling income of the retail market, Cst represents a device start-stop cost, Ccoal represents a fuel cost of a coal-fired unit, CoperateCFU-RTO represents an operation and maintenance cost of the coal-fired unit and a ventilation air oxidation and power generation unit, CCO2CFU-RTO represents a carbon emission cost, Cvam represents a waste ventilation air penalty cost, and CRT represents an integrated operation cost at the intraday stage;
the user demand response model comprises a second objective function; the second objective function comprises:
 wherein Wuser represents the sum of the actual electricity-purchasing cost of the user and the electricity benefit loss in the demand response, Psellt and πsellt respectively represent an electric quantity and a price of the park integrated energy system participating in retail transaction at a period t, αSLt represents a benefit loss coefficient of a time transferable load at the period t, Ptt′, represents a load quantity transferred from the period t to a period t′, αILt and βSLt represent benefit loss coefficients of a reducible load, rILt represents a proportion of the reducible load at the period t, Pusert represents a total load of the user at the period t, PILt represents a reducible load value at the period t, and T represents the total number of periods;
the solving the short-time scale dynamic pricing model based on a Karush-Kuhn-Tucker condition and a Benders decomposition method comprises: representing the optimal solution of the user demand response model through the necessary and sufficient Karush-Kuhn-Tucker condition, and converting the short-time scale dynamic pricing model from double layers to a single layer; dividing the converted short-time scale dynamic pricing model into a master problem and subproblems, wherein related variables of the coal-fired unit and a ventilation air oxidation unit are taken as complex variables required to be optimized in the master problem, and each real-time scenario is taken as one subproblem; and sequentially solving the master problem and the subproblems, and computing upper and lower bounds of the original objective function and separating an optimal cut until a difference value between the upper and lower bounds meet a convergence condition to obtain the day-ahead and intraday scheduling plans and the next-day electricity price;
in the step of computing the strategy transfer probability distribution of each market subject within the preset deadline by the improved regret degree matching mechanism, the strategy transfer probability of each market subject in a next contract period is computed by the following formulas:
 wherein xn+1k(j) and xn+1k(i) respectively represent a probability of selecting a new decision j and still insisting on the current decision i in the (n+1)th decision by the market subject k, σ represents a weight factor of the current strategy preference degree, γ∈(0,¼), λ represents a proportionality coefficient and λ≥2MSk, M represents upper bound values of utility functions of all the strategies in a strategy set of the market subject k, Sk represents a cardinal number of the strategy set of the market subject k, Rnk(j,i) represents the regret degree of regretting selecting the current strategy i instead of selecting the strategy j by the market subject k after the nth decision, and Πk represents the strategy set of the market subject k; and
whether each market subject reaches the correlated equilibrium state is detected by the following formula:
 wherein μ represents a vector formed by the current strategy of each market subject, S represents a vector space replaced by all possible strategies, μk=j represents that the current strategy of the market subject k is j, μ−k represents a vector formed by the current strategy of other market subjects except the market subject k, ψ(μ) represents a probability corresponding to the strategy vector μ, Wk(i,μ−k) represents a benefit function value of the market subject k replacing the strategy j with the strategy i, and Wk(μ) represents a benefit function value of the market subject k adopting the original strategy.
|