Mine integrated energy system scheduling two-stage multi-form differential evolution method

Abstract

The invention discloses a two-stage multiform differential evolution method for dispatching a mine integrated energy system, and belongs to the technical field of intelligent dispatching. The method comprises the following steps: firstly, under a multi-form optimization framework, combining domain knowledge to construct an electricity-heat alternative task and an electricity-cold alternative task, and introducing a feasibility-driven knowledge migration mechanism to realize efficient migration of knowledge between the two tasks; secondly, designing a two-stage optimization strategy, proposing a neighborhood double-variation mechanism, in the first stage, performing parallel evolution on populations corresponding to two alternative tasks through a diversity-driven neighborhood double-variation strategy, and in the second stage, performing evolution on an original task through a convergence-driven neighborhood double-variation strategy to obtain an optimal feasible solution set; and finally, analyzing the time complexity for explaining the rationality of the proposed method for optimizing the scheduling of the mine integrated energy system. The method can deal with the limitation of an existing optimization method when strong coupling constraint and multiple uncertainties are processed, and the economical efficiency, the energy utilization rate and the scheduling stability of the system are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent scheduling, and in particular relates to a two-stage multi-form differential evolution method for scheduling a mine comprehensive energy system. Background Art

[0002] In recent years, the integrated mine energy system (CMIES) has attracted much attention because it can effectively alleviate the high energy consumption and high pollution problems in mines. CMIES fully utilizes coal mine associated energy, such as air heat, exhaust air, mine water and geothermal energy, based on the integration of traditional integrated energy systems (IESs) equipment, such as wind power, photovoltaics, gas turbines and energy storage devices. These associated energy resources are abundant, and if they are directly discharged, they will cause waste and even have a negative impact on the environment. Therefore, how to efficiently utilize these associated energy is an important issue to promote the transformation of mine energy structure and reduce carbon emissions.

[0003] Compared with traditional IESs, CMIES has more diverse energy types and more complex energy conversion processes, which increases the difficulty of system management and optimization. Especially in terms of scheduling optimization, the stable operation of CMIES depends on a reasonable scheduling strategy, which can not only improve the economic benefits of the system, but also improve the utilization efficiency of associated energy in mines. However, the scheduling of the system also needs to deal with the uncertainty of many associated energy outputs. For example, air heat and exhaust air volume fluctuate with the mine exhaust volume and production capacity, mine water flow is affected by climate and geological conditions, and geothermal resources also have seasonal changes. These uncertainties make it difficult for the scheduling scheme to accurately match the actual demand, which significantly affects the accuracy and effectiveness of optimization. Therefore, it is necessary to further consider uncertainty in CMIES scheduling, but it also greatly increases the complexity of optimization.

[0004] In order to deal with the uncertainty in scheduling optimization, a variety of methods have been developed in recent years, including stochastic programming, fuzzy programming, robust optimization and interval optimization. Among them, stochastic programming describes uncertain variables through probability distribution, aiming to obtain the optimal solution under expectation, but it relies on a large amount of historical data, has high computational complexity and the scheme is too optimistic; fuzzy programming uses fuzzy numbers to deal with uncertainty, allowing some constraints to be relaxed, but its results are limited by the definition of membership functions and are prone to large errors. In contrast, robust optimization and interval optimization do not need to rely on a large amount of data or pre-assume parameter distribution, and have more application advantages. Robust optimization focuses on optimization in the worst case and is a conservative decision-making method; interval optimization focuses on the overall performance of the target interval, and often converts uncertainty problems into deterministic problems through interval order relations or minimum and maximum regret criteria. However, this model conversion may lose some uncertainty information, and different conversion models may cause the same problem to be converted into different deterministic problems, affecting the accuracy of the optimization results and the stability of the system.

[0005] Therefore, it is particularly important to develop a direct method that does not rely on model conversion to deal with the uncertainty problem in CMIES scheduling, retain more uncertainty information, and improve the accuracy of scheduling solutions. Summary of the invention

[0006] The problem to be solved by the present invention is to provide a two-stage multi-form differential evolution method for scheduling a mine comprehensive energy system.

[0007] Comprehensively consider the multiple uncertainties of mining associated energy and load to improve the economic benefits and energy utilization efficiency of the system.

[0008] The present invention adopts the following technical solution: a two-stage multi-form differential evolution method for scheduling a mine comprehensive energy system, comprising the following steps:

[0009] Step 1: Construct alternative tasks: Under the multi-form optimization framework, two alternative tasks, electricity-heat and electricity-cooling, are constructed for the original model of the mine integrated energy system to solve the feasible domain positioning problem caused by the interval strong coupling constraint;

[0010] Step 2: Construct a knowledge transfer strategy based on two alternative tasks, and transfer the coupled variable parts of the feasible solutions in the two alternative tasks each time to improve the efficiency of locating feasible solutions;

[0011] Step 3: Construct a two-stage optimization framework: the first stage is used to search multiple feasible domains, and the second stage is used to perform a detailed search on the original task;

[0012] Step 4: Construct a differential evolution strategy based on neighborhood double mutation: Combine the DE / rand / 1 and DE / current-to-pbest / 1 operators to construct a double mutation mechanism that integrates the neighborhood mechanism, and optimize the corresponding tasks in the two stages respectively. The mutation strategy in the first stage is the neighborhood double mutation strategy driven by diversity, and the mutation strategy in the second stage is the neighborhood double mutation strategy driven by convergence.

[0013] Step 5, performing two-stage multi-form differential evolution for the scheduling of the mine integrated energy system: based on the alternative tasks constructed in step 1 and the two-stage optimization framework constructed in step 3, in the first stage, the populations corresponding to the two alternative tasks are evolved in parallel through the diversity-driven neighborhood double mutation strategy, knowledge is transferred between the two tasks through the knowledge transfer strategy, and the population finally obtained by the evolution is used as the initial population in the second stage, and the original task is optimized through the convergence-driven neighborhood double mutation strategy to obtain the optimal feasible solution set;

[0014] Step 6: Analyze the time complexity to illustrate the rationality of the proposed method for optimizing the scheduling of mine integrated energy systems.

[0015] Specifically, due to the strong coupling and uncertainty of constraints, the original CMIES scheduling problem is difficult to solve directly. To meet this challenge, the present invention draws on the idea of ​​multi-form optimization and formulates alternative tasks for the original tasks to reduce the difficulty of problem solving and achieve efficient optimization. In fact, methods such as changing constraints, changing the structure of the objective function, and reducing the number of control variables have achieved good results in the formulation of alternative tasks for multi-form optimization. The core of these methods is to simplify the tasks so that they are easy to solve, rather than blindly pursuing high quality solutions.

[0016] Preferably, in step 1, based on domain knowledge, two alternative tasks are constructed by decoupling the coupling relationship in the electric, cold and hot equation constraints. Specifically, the coupling space of the electric, cold and hot variables is divided into two low-dimensional subspaces: the electric-thermal subspace and the electric-cold subspace, thereby forming two alternative tasks: the electric-thermal task and the electric-cold task.

[0017] The electrical-thermal task is expressed as follows:

[0018]

[0019] The power-cooling task is expressed as follows:

[0020]

[0021] Among them, the optimization variables X are divided into three categories: electrical related variables X electric , heat-related variables X themal and the variable X related to cold cooling ; Function F represents the objective function; Function F1 represents the economic operation cost target; Function F2 represents the cost target of abandoned renewable energy and associated energy in mines; S represents the search space; R n Represents n-dimensional real number space.

[0022] In the electric-heat task, variables and constraints related to cooling are ignored; in the electric-cold task, variables and constraints related to heat are ignored. By decoupling the relationship between electricity, heat, and cooling, the variable dimension and constraint coupling strength of the original task are reduced.

[0023] Preferably, in step 2, a feasibility-driven knowledge transfer strategy is constructed to exchange feasible solutions between two alternative tasks and determine the intersecting feasible domain. The feasibility-driven knowledge transfer strategy performs knowledge transfer only when two tasks produce feasible solutions, so as to maximize the use of feasible information of both parties and explore the intersecting feasible domain.

[0024] Specifically, at the beginning of the iteration, the algorithm first searches for the feasible domain under each task. Since the feasible domains of the two alternative tasks are larger than those of the original tasks, the algorithm can quickly converge to the feasible solution space and then migrate the feasible solutions.

[0025] However, it should be noted that these two alternative tasks are heterogeneous tasks, because they involve different decision variables and cannot directly perform knowledge transfer. However, both tasks contain decision variables related to electricity. Considering that the electrical variables in the intersecting feasible domains are the same, the present invention only performs knowledge transfer on the electrical variables in the feasible solution and splices them with the hot or cold variables under each task to achieve information interaction. By contributing electrical variables, the two tasks can transfer effective information in the feasible domain to each other, enhance the ability to find intersecting feasible domains, and speed up the process of locating feasible solutions.

[0026] Specifically, in the optimization process of evolutionary algorithms, single-stage optimization is often prone to fall into local optimality, which is more obvious when dealing with the scheduling problem of mine integrated energy system. Therefore, the present invention reduces the risk of falling into the local optimal feasible domain by constructing a two-stage optimization strategy, combining extensive exploration and refined search.

[0027] Preferably, in step 3, a two-stage optimization framework is constructed, as follows: In the first stage, the algorithm simultaneously optimizes two alternative tasks (electric-heat task and electric-cooling task) to fully explore their respective feasible spaces and obtain the intersecting feasible domains of the two alternative tasks. Then, the second stage focuses on the optimization of the original task, focusing on a refined search of the intersecting feasible domains found in the first stage to obtain the optimal feasible Pareto front. This strategy reduces the complexity of the problem by decoupling the original task in the first stage, and effectively avoids the possibility of early convergence to the local optimum. In the second stage, deep optimization is performed within the confirmed feasible domain to ensure the globality and convergence of the solution.

[0028] In order to better allocate computing resources, the present invention sets a stage transition condition: when the number of iterations reaches 70% of the maximum number of iterations, the algorithm jumps from the first stage to the second stage.

[0029] Specifically, differential evolution is a simple and efficient evolutionary algorithm that has been widely used in various optimization problems, especially in the scheduling of integrated energy systems. Based on its superior performance in related fields, the present invention selects differential evolution as the core algorithm for CMIES scheduling optimization. However, the original differential evolution is prone to fall into local optimality when dealing with high-dimensional and strongly constrained problems.

[0030] Preferably, in step 4, a neighborhood-based double mutation strategy (NDM) is proposed to improve the performance of differential evolution, combining the neighborhood technology and the DE / rand / 1 strategy to increase the diversity of the algorithm, and introducing the DE / current-to-pbest / 1 strategy to improve the feasibility and convergence speed of the algorithm.

[0031] In the two-stage optimization, the first stage focuses on diversity to explore more feasible domains, while the second stage focuses on feasibility and convergence to obtain a better Pareto front.

[0032] Furthermore, the present invention adjusts the performance of the dual mutation strategy at different stages by controlling the parameter Pa, thereby maximizing the overall effectiveness of the algorithm. The specific mutation strategy is as follows:

[0033] Phase 1 mutation strategy: Diversity-driven neighborhood double mutation strategy (NDM_div)

[0034]

[0035] Second stage mutation strategy: Convergence driven neighborhood double mutation strategy (NDM_con)

[0036]

[0037] Among them, rand i Represents a random number between 0 and 1; r1', r2' represents the i Two individuals randomly selected from the neighborhood of X, and r1'≠r2'; r1, r2 represent two individuals randomly selected from the current population, and r1≠r2; X best represents the optimal individual of the current population; X pbest represents one of the top q% excellent individuals in the population, q is set to 0.1; F i Randomly selected from [0.6, 0.8, 1.0]; Pa is a probability value used to control the performance emphasis of the mutation strategy in the two stages.

[0038] Specifically, after completing the mutation operation, selecting excellent individuals to enter the next generation is also one of the key steps of differential evolution. However, due to the uncertainty of the problem of the present invention, the target value and the constraint value are both interval numbers, and the traditional Pareto dominance relationship and crowding distance cannot effectively determine the dominance relationship and crowding degree between interval numbers.

[0039] Furthermore, the present invention adopts conditional interval dominance relationship and interval crowding distance to evaluate individual performance. Aiming at the problem of handling interval constraints, interval-based feasibility rules are introduced, and the above rules are combined with constraint dominance principle and improved epsilon to propose an improved interval environment selection strategy to select better individuals to enter the next generation.

[0040] Specifically, knowledge transfer occurs only in the first phase, that is, on two alternative tasks, to help locate the feasible solution space of the original task. The two alternative tasks have different optimization preferences, and the main goal is to explore the feasible spaces of the two alternative tasks and find the intersecting feasible domains of the two alternative tasks.

[0041] Preferably, in step 5, a two-stage multi-form differential evolution of the mine integrated energy system scheduling is performed, and the evolution process is divided into two stages, and different stages achieve different optimization goals respectively.

[0042] First, in the first stage, the population P corresponding to the two alternative tasks of electricity-heating and electricity-cooling H and P C , respectively, parallel evolution is performed through a diversity-driven neighborhood double mutation strategy, and individual evaluation is performed.

[0043] Then, knowledge is transferred between the two alternative tasks through the knowledge transfer strategy. Based on the different objectives and constraints of the two alternative tasks, the objectives and constraints of the migrated individuals are re-evaluated. An improved interval epsilon constraint processing mechanism is adopted to relax the constraints in the early stage of iteration and select potential individuals to enter the next generation.

[0044] Next, in order to make full use of the optimization results of the first stage, the population finally obtained in the first stage of evolution is used as the initial population P of the second stage.

[0045] Finally, in the second stage, the initial population P is evaluated under the original task, and the original task is optimized through the convergence-driven neighborhood double mutation strategy to obtain the optimal feasible solution set.

[0046] Preferably, in step 6, the time complexity is calculated as follows:

[0047] In the first stage, two alternative tasks are evolved in parallel. The evolution process of each population includes selection, crossover and mutation operations. Assume that the population size is N, the number of objective functions is M, and the maximum number of iterations is G. max , all individuals need to be evaluated and selected in each iteration, so the time complexity of a single iteration is approximately O(N).

[0048] Since the two populations evolve independently and in parallel in the first stage, the total time complexity is O(2NG max )=

[0049] O(NG max ).

[0050] In addition, the knowledge transfer operation is performed once per generation, involving the screening and re-evaluation of feasible individuals. Since this process only affects a part of the individuals, its time complexity is O(N) per generation, so the total complexity of knowledge transfer is O(NG max ), this extra operation does not significantly increase the overall time complexity.

[0051] In the second stage, the evolution of population P is based on the optimization results of the first stage. The total time complexity of the second stage is O(NGmax ).

[0052] In addition, two key calculations are involved: the dominance relationship based on the conditional interval and the interval crowding distance, which are respectively calculated in O(MN 2 ) and O(MNlogN) time complexity for calculation.

[0053] The environment selection mechanism involves applying the interval constraint dominance principle and the improved interval epsilon constraint processing mechanism. The complexity of this process is O(MN 2 ).

[0054] Combining the two phases and considering all operations, the overall time complexity of the algorithm is expressed as follows:

[0055] O(NG max +MN 2 +MNlogN)=O(NG max +MN 2 ).

[0056] The technical solution of the present invention also provides: an electronic device, comprising:

[0057] one or more processors;

[0058] a storage device having one or more programs stored thereon;

[0059] When the one or more programs are executed by the one or more processors, the one or more processors implement any of the above-mentioned two-stage multi-form differential evolution methods for scheduling a comprehensive energy system in a mine.

[0060] The technical solution of the present invention also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in any of the above-mentioned two-stage multi-form differential evolution methods for scheduling a mine comprehensive energy system are implemented.

[0061] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0062] 1. The method of the present invention comprehensively considers the multiple uncertainties of associated energy and load in mines, and proposes two alternative tasks of electricity-heat and electricity-cooling based on domain knowledge, which solves the problem of feasible domain positioning caused by strong coupling constraints of intervals and effectively reduces the complexity of the problem.

[0063] 2. The method of the present invention adopts a two-stage optimization framework. The first stage searches for multiple feasible domains, and the second stage performs a detailed search on the original task to improve the diversity of understanding and the distribution of the Pareto front.

[0064] 3. The method of the present invention constructs a knowledge transfer strategy based on two alternative tasks, and only transfers the coupled variable part of the feasible solutions in the two tasks each time, which effectively improves the efficiency of the algorithm in locating feasible solutions in complex scheduling problems.

[0065] 4. The method of the present invention combines the DE / rand / 1 and DE / current-to-pbest / 1 operators, proposes a double variation mechanism integrating the neighborhood mechanism, which improves the convergence performance of the algorithm in uncertainty problems. In addition, the constraint processing technology is improved by combining the interval dominance relationship and the interval constraint processing rules, and a new interval-based environment selection strategy is proposed, which can directly deal with interval problems and significantly improve the effectiveness of interval variable optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 This is a flow chart of the two-stage multi-form differential evolution method for scheduling the mine comprehensive energy system of the present invention;

[0067] Figure 2 This is a framework diagram of a mine integrated energy system used in an embodiment of the present invention;

[0068] Figure 3 A comparison diagram of Pareto front results obtained by using the constrained multi-objective optimization algorithm and the method of the present invention respectively in an embodiment of the present invention;

[0069] Figure 4 A comparison diagram of Pareto front results obtained by using the interval multi-objective optimization algorithm and the method of the present invention respectively in the embodiments of the present invention;

[0070] Figure 5 The following is a comparison chart of Pareto front results obtained by respectively using the interval-constrained multi-objective optimization algorithm and the method of the present invention in the embodiments of the present invention. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the application is further elaborated in detail below in conjunction with the accompanying drawings. The described embodiments are only a part of the embodiments involved in the present invention. All non-innovative embodiments of other researchers in the field on this embodiment belong to the protection scope of the present invention. At the same time, for the step numbering in the embodiment of the present invention, it is only set for the convenience of explanation, and the order between the steps is not limited in any way. The execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0072] The present invention proposes a two-stage multi-form differential evolution method for the scheduling of a mine integrated energy system. The method achieves efficient solution to the problem by introducing core technologies such as alternative task construction, two-stage optimization strategy, feasibility-driven knowledge transfer strategy, neighborhood-based double mutation differential evolution strategy, and improved interval constraint processing technology into the two-stage multi-form differential evolution method.

[0073] The evolution process of the present invention is divided into two stages, such as Figure 1 As shown, different stages achieve different optimization goals.

[0074] The purpose of the first stage is to optimize two constructed alternative tasks, namely the electric-heat task and the electric-cooling task. The population P corresponding to each task H and P C Evolution proceeds in parallel until the stage transition conditions are met. During the evolution process, the two populations generate offspring through the diversity-driven neighborhood double mutation strategy (NDM_div) and use the feasibility-driven knowledge transfer mechanism to transfer each other's effective information. In order to retain individuals with potential, an improved interval environment selection mechanism combining interval dominance relations and constraint processing mechanisms is adopted.

[0075] The goal of the second stage is to evolve the population P corresponding to the original task until the termination condition of the algorithm is met. In order to make full use of the optimization results of the first stage, the final population of the first stage is used as the initial population of the second stage, and the offspring is generated through the convergence-driven neighborhood double mutation strategy (NDM_con), and the improved interval environment selection mechanism is adopted to ensure that potential individuals enter the next generation.

[0076] The pseudo code of the two-stage multi-form differential evolution framework of the present invention is as follows:

[0077]

[0078]

[0079] Among them, the input population size is N; the maximum number of iterations is G max ; Phase transition parameter β, where β is set to 0.7.

[0080] First, the populations of the two alternative tasks are initialized and evaluated under each task. Then, the algorithm enters the main loop and controls the evolution through the number of iterations G. When G≤β*G max When , the algorithm is in the first stage, otherwise it enters the second stage.

[0081] Obtaining the optimal feasible solution set includes the following sub-steps:

[0082] Step 5.1, Offspring Generation: In the first stage, P H and PC The offspring are generated respectively through the NDM_div strategy; in the second stage, P generates offspring through the NDM_con strategy.

[0083] Step 5.2: Individual evaluation: The first stage is to evaluate P H and P C The objectives and constraint violation values ​​are calculated under each task respectively; in the second stage, P is evaluated under the original task.

[0084] Step 5.3, knowledge transfer: knowledge transfer only occurs in the first stage, and each generation and However, since the goals and constraints of the two tasks are different, the individuals after the transfer need to re-evaluate the goals and constraints.

[0085] Step 5.4, environment selection: In the first stage, an improved interval epsilon constraint processing mechanism is used to relax the constraints in the early stage of iteration and select potential individuals to enter the next generation; in the second stage, the interval constraint dominance principle is adopted.

[0086] Step 5.5, stage transition: Once the next generation of individuals is generated, determine whether to enter the second stage based on the number of iterations G. When G>β*G max When the evolution enters the second stage.

[0087] Step 5.6, Final solution set: The algorithm continues to evolve P until the termination condition is met, and finally obtains the interval non-dominated feasible solution set P F .

[0088] In one embodiment of the present invention, a source-load uncertain integrated energy system for mines is constructed, such as Figure 2 As shown, the system provides heat, electricity and cold energy through various energy devices.

[0089] Specifically, thermal energy is provided by mine co-generation energy devices and gas turbines, electrical energy is provided by large power grids, wind and solar units, gas turbines and power storage devices, and cold energy is provided by electric refrigerators and absorption refrigerators.

[0090] In the mining process, associated energy such as exhaust air, air heat, mine water and geothermal heat are used to extract heat energy through exhaust air thermal storage oxidation devices, air source heat pumps, water source heat pumps and ground source heat pumps. In addition, the associated energy devices in mines are not only suppliers of heat energy, but also consumers of electric energy.

[0091] System loads are mainly divided into electrical load, thermal load and cooling load. Electrical load includes rigid electrical load and uncertain electrical load; thermal load includes fixed thermal load and uncertain thermal load.

[0092] At present, the prediction of uncertain parameters mainly includes probability prediction and interval prediction. Probability prediction requires a large amount of data to obtain the probability density function, while interval prediction only needs to determine the value range of the uncertainty parameter and does not require an accurate probability distribution model. Therefore, this embodiment uses interval numbers to describe the uncertainty of renewable energy, mining associated energy, electric load and thermal load.

[0093] Furthermore, the effectiveness of the two-stage multi-form differential evolution method of the present invention in solving scheduling problems is verified by constructing a source-load uncertain mine integrated energy system.

[0094] The scheduling problem model solved in this embodiment mainly includes two parts: objective function and constraints.

[0095] The objective function is as follows:

[0096] (1) Economic operating costs

[0097] The economic operation cost of the mine comprehensive energy system is F1(P,P I ), including: the power purchase cost of the power grid C e , Gas purchasing cost of gas grid C g , Operation and maintenance costs of renewable energy devices Operation and maintenance costs of associated energy devices The charging and discharging cost of the energy storage device C es and the operation and maintenance cost of the cooling device C c .

[0098]

[0099] C e,t =α grid,t P grid,t

[0100] C g,t =α gt,t P gt,t

[0101]

[0102] C es,t =α es (x es_out,t P es_out,t +x es_in,t P es_in,t )

[0103] C c,t =α ec Q ec,t +α ac Q ac,t

[0104] Among them, α is the cost coefficient; P, H, Q are the electrical, thermal, and cooling output powers of each device respectively; P I ,H I are the electrical and thermal output power ranges of each device respectively; the subscript t represents time t; x es_out,t and x es_in,t They respectively represent the output and input states of the electric storage device; the subscript grid corresponds to the power purchasing device of the power grid; the subscript gt corresponds to the gas purchasing device of the gas grid; the subscript wt corresponds to the renewable energy device; the subscripts vod, wshp, ashp, and gshp correspond to the exhaust air oxidation device, the water source heat pump device, the gas source heat pump device, and the ground source heat pump device, respectively; the subscripts es_out and es_in correspond to the output and input of the electric storage device, respectively; the subscripts ec and ac represent electric refrigerators and absorption refrigerators, respectively.

[0105] (2) Costs of abandoned renewable energy and mining associated energy

[0106] In order to make full use of renewable energy and associated energy, reduce the pollution of the environment caused by the power purchase of large power grids and the emission of associated energy, the objective function is to minimize the cost of abandoned renewable energy and mine associated energy.

[0107]

[0108] Among them, β is the penalty cost coefficient for abandoned energy; P preI ,H preI Predict the power range for renewable energy and mining associated energy electricity and heat; P I ,H I It is the actual output power range of electricity and heat from renewable energy and mining associated energy.

[0109] The constraints are as follows:

[0110] (1) Electric power balance constraints

[0111] The electric load includes fixed electric load and uncertain electric load.

[0112]

[0113] Where P load0,t is the fixed electric load power at time t; is the predicted power range of the uncertain electric load at time t.

[0114] 2) Thermal power balance constraints

[0115]

[0116] In the formula, H load0,t is the fixed heat load power at time t; is the power range predicted for the uncertain heat load at time t.

[0117] 3) Cold power balance constraint

[0118] Q ec,t +Q ac,t =Q load,t

[0119] In the formula, Q load,t is the cooling load power at time t.

[0120] 4) Constraints on renewable energy installations

[0121] The predicted power range of renewable energy can be expressed as:

[0122]

[0123] In the formula, the superscripts - and + represent the upper and lower limits of the prediction interval respectively; the actual power consumption range of renewable energy can be expressed as:

[0124]

[0125] And both must satisfy the following constraints:

[0126]

[0127] 5) Constraints on associated energy devices

[0128] The predicted power range of associated energy can be expressed as:

[0129]

[0130] The actual power consumption range of associated energy can be expressed as:

[0131]

[0132] And both must satisfy the following constraints:

[0133]

[0134]

[0135] In the formula, H preI =[H pre- ,H pre+ ] is the predicted power range of the associated energy in the mine; H I =[H - ,H + ] is the actual power consumption range of mining associated energy; η is the electric-thermal conversion coefficient of mining associated energy equipment.

[0136] 6) Gas turbine constraints

[0137]

[0138] In the formula, is the upper limit of the gas turbine output at time t; η is the gas turbine electric heat conversion coefficient; R up and R down They represent the upper and lower limits of the gas turbine climbing rate respectively.

[0139] 7) Constraints on electric energy storage devices

[0140]

[0141] In the formula, x es,t is the charge and discharge state variable of the power storage device at time t, x es,t =1 means the power storage device is in power storage state, x es,t =0 means the storage device is in the discharge state; and are the upper limits of the charging and discharging power of the energy storage device; S max and S min They are respectively the upper and lower limits of the amount of electricity that can be stored in the electric energy storage device.

[0142] 8) Cooling equipment output constraints

[0143]

[0144] Where P max is the maximum output power of the cooling equipment at time t; η is the electric cooling conversion coefficient.

[0145] 9) Grid output constraints

[0146]

[0147] In the formula, is the maximum output power of the large power grid at time t.

[0148] In order to verify the effectiveness of the two-stage multi-form differential evolution method proposed in this invention, it is applied to the scheduling problem of the mine integrated energy system with uncertain source and load at the same time with three types of comparison algorithms. The specific description of the three types of comparison algorithms is as follows:

[0149] The first category: 6 popular constrained multi-objective optimization algorithms, including the collaborative evolutionary constrained multi-objective algorithm (CCMO), the push-pull search algorithm (PPS), the evolutionary multi-task based constrained multi-objective algorithm (EMCMO), the evolutionary multi-task algorithm based on dynamic auxiliary tasks (MTCMO), the improved MTCMO algorithm (IMTCMO) and the multi-form optimization algorithm based on SPEA2 (MFOSPEA2).

[0150] The second category: an interval multi-objective optimization algorithm, namely, an interval multi-objective optimization algorithm based on elite genetic strategy (EG-IMOEA).

[0151] The third category: an interval-constrained multi-objective optimization algorithm, namely, the interval-constrained multi-objective evolutionary algorithm guided by the degree of interval constraint violation (DIC-MOEA / D).

[0152] It should be noted that the first type of algorithm lacks an interval comparison mechanism, and the second type of algorithm lacks a constraint processing mechanism, so these two types of algorithms cannot directly solve the problem of this embodiment. To this end, this embodiment combines the interval-based comparison strategy with the first type of algorithm, and combines the constraint dominance principle with the second type of algorithm to make up for these deficiencies.

[0153] In terms of parameter setting, the population size of the first two types of comparison algorithms is set to N = 100, and the number of iterations G max =5000. However, the third algorithm cannot find a feasible solution in the final iteration under this parameter setting. Therefore, for the third algorithm, the population size is adjusted to N = 500 and the number of iterations is adjusted to G. max = 2000. In addition, the parameters of other algorithms are kept consistent. To ensure the fairness of the comparison, all algorithms are run 20 times independently and the results are counted.

[0154] In this embodiment, the following three performance indicators are used to evaluate the performance of each algorithm in terms of convergence, diversity and uncertainty:

[0155] (1) Inter-generational distance (IGD):

[0156] IGD is used to measure the closeness between the Pareto front generated by the algorithm and the optimal Pareto front. It aims to evaluate the convergence and distribution of the algorithm. The smaller the IGD value, the better the algorithm performance.

[0157] (2) Interval Hypervolume (HV):

[0158] HV is another commonly used indicator for evaluating population diversity and convergence. It is calculated based on the reference point (1.1, 1.1) in the normalized target space. The higher the HV value, the better the performance of the algorithm.

[0159] (3) Uncertainty (IM):

[0160] This indicator reflects the performance of the population under uncertainty and is measured by calculating the sum of the interval lengths of all individual target values ​​in the Pareto optimal solution set. The smaller the uncertainty index, the better the algorithm performs in dealing with uncertainty.

[0161] Experiment 1: Comparison and analysis of the results of the first type of comparison algorithm and the algorithm of the present invention

[0162] This experiment compares the performance of TMFDE with the first type of algorithm to further verify its effect. The Pareto front comparison results of all algorithms are as follows: Figure 3 shown. Figure 3 It is shown that compared with other comparison algorithms, the method of the present invention obtains a more uniform and extensive Pareto frontier, which can better solve the source-load uncertain mining integrated energy system scheduling problem.

[0163] Experiment 2: Comparison and analysis of the results of the second type of comparison algorithm and the algorithm of the present invention

[0164] This experiment compares the performance of TMFDE with the second type of algorithm. The Pareto front comparison results are as follows: Figure 4 As shown in the figure, it can be seen that the uniformity and extensiveness obtained by the method of the present invention are better.

[0165] Experiment 3: Comparison and analysis of the results of the third type of comparison algorithm and the algorithm of the present invention

[0166] This experiment compares the performance of TMFDE with the third type of algorithm. The Pareto front comparison results are as follows: Figure 5 As shown in the figure, it can be seen that the uniformity and extensiveness obtained by the method of the present invention are better.

[0167] It can be seen that the two-stage multi-form differential evolution method of the present invention can not only effectively address the limitations of existing optimization methods in dealing with strong coupling constraints and multiple uncertainties, but also the results on the three performance indicators are better than the existing comparative algorithms, indicating its excellent performance in economy, energy utilization and scheduling stability.

[0168] In an embodiment of the present invention, an electronic device is also provided, including: one or more processors; a storage device on which one or more programs are stored; when the one or more programs are executed by the one or more processors, the one or more processors implement the two-stage multi-form differential evolution method for scheduling a comprehensive energy system for a mine described in any of the above embodiments.

[0169] In an embodiment of the present invention, a computer-readable storage medium is also provided, on which a computer program is stored. When the program is executed by a processor, the steps in any one of the two-stage multi-form differential evolution methods for scheduling a mine integrated energy system in the above embodiments are implemented.

[0170] The embodiments described above are only preferred specific implementation modes of the present invention, and the protection scope of the present invention is not limited thereto. Any simple changes or equivalent replacements of the technical solutions that can be obviously obtained by any technician familiar with the field within the technical scope disclosed in the present invention belong to the protection scope of the present invention.

Claims

1. A two-stage multi-form differential evolution method for scheduling of integrated energy systems in mines, characterized in that: The steps include: Step 1: Construct alternative tasks: Under the multi-form optimization framework, two alternative tasks, electricity-heat and electricity-cooling, are constructed for the original model of the mine integrated energy system to solve the feasible domain positioning problem caused by the interval strong coupling constraint; Step 2: Construct a knowledge transfer strategy based on two alternative tasks, and transfer the coupled variable parts of the feasible solutions in the two alternative tasks each time to improve the efficiency of locating feasible solutions; Step 3: Construct a two-stage optimization framework: the first stage is used to search multiple feasible domains, and the second stage is used to perform a detailed search on the original task; Step 4: Construct a differential evolution strategy based on neighborhood double mutation: Combine the DE / rand / 1 and DE / current-to-pbest / 1 operators to construct a double mutation mechanism that integrates the neighborhood mechanism, and optimize the corresponding tasks in the two stages respectively. The mutation strategy in the first stage is the neighborhood double mutation strategy driven by diversity, and the mutation strategy in the second stage is the neighborhood double mutation strategy driven by convergence. Step 5, performing two-stage multi-form differential evolution for the scheduling of the mine integrated energy system: based on the alternative tasks constructed in step 1 and the two-stage optimization framework constructed in step 3, in the first stage, the populations corresponding to the two alternative tasks are evolved in parallel through the diversity-driven neighborhood double mutation strategy, knowledge is transferred between the two tasks through the knowledge transfer strategy, and the population finally obtained by the evolution is used as the initial population in the second stage, and the original task is optimized through the convergence-driven neighborhood double mutation strategy to obtain the optimal feasible solution set; Step 6: Analyze the time complexity to illustrate the rationality of the proposed method for optimizing the scheduling of mine integrated energy systems.

2. The two-stage multi-form differential evolution method for scheduling a mine integrated energy system according to claim 1 is characterized in that: In step 1, an alternative task is constructed. By decoupling the coupling relationship in the electric, cold and hot equation constraints, the coupling space of the electric, cold and hot variables is divided into two low-dimensional subspaces: the electric-hot subspace and the electric-cold subspace, forming two alternative tasks: the electric-hot task and the electric-cold task, reducing the variable dimension and constraint coupling strength of the original task; The electrical-thermal task, ignoring variables and constraints related to cooling, is expressed as follows: The power-cooling task, ignoring heat-related variables and constraints, is expressed as follows: Among them, the optimization variables X are divided into three categories: variables X related to electricity electric , heat-related variables X themal and the variable X related to cold cooling ; Function F represents the objective function; Function F1 represents the economic operation cost target; Function F2 represents the cost target of abandoned renewable energy and associated energy in mines; S represents the search space; R n Represents n-dimensional real number space.

3. The two-stage multi-form differential evolution method for scheduling a mine integrated energy system according to claim 1 is characterized in that: The knowledge transfer strategy described in step 2 is used to improve the efficiency of transferring effective information between two alternative tasks, as follows: Step 2.1: knowledge transfer occurs in the first stage and is performed on two alternative tasks to help locate the feasible solution space of the original task; different optimization preferences are set for the two alternative tasks, with the goal of finding the intersecting feasible solution space of the two alternative tasks; Step 2.2: Exchange feasible solutions between two alternative tasks to speed up the efficiency of finding intersecting feasible domains; Step 2.3: Through the feasibility-driven knowledge transfer strategy, knowledge transfer is performed when two alternative tasks generate feasible solutions. The two alternative tasks both contain decision variables related to electricity. The electrical variables in the intersecting feasible domain are the same. Knowledge transfer is performed on the electrical variables in the feasible solution and they are spliced ​​with the hot or cold variables under each task to achieve information interaction.

4. The two-stage multi-form differential evolution method for scheduling a mine integrated energy system according to claim 1 is characterized in that: In step 3, a two-stage optimization framework is constructed to improve the diversity of solutions and the distribution of the Pareto front, which includes the following sub-steps: Step 3.1, in the first stage, optimize the two alternative tasks simultaneously, explore their respective feasible spaces, and obtain the intersecting feasible domains of the two alternative tasks; Step 3.2, set the stage transition condition, when the number of iterations reaches the preset maximum number of iterations, jump from the first stage to the second stage; Step 3.3: In the second stage, the original task is optimized, and the intersecting feasible domains found in the first stage are refined to obtain the optimal feasible Pareto frontier.

5. The two-stage multi-form differential evolution method for scheduling a mine comprehensive energy system according to claim 1 is characterized in that: In step 4, a differential evolution strategy based on neighborhood double mutation is constructed, combining the neighborhood method and DE / rand / 1 operator to increase diversity, introducing DE / current-to-pbest / 1 operator to improve feasibility and convergence speed, and adjusting the performance of the neighborhood double mutation strategy at different stages by controlling the parameter Pa; The diversity-driven neighborhood double mutation strategy is expressed as follows: The convergence-driven neighborhood double mutation strategy is expressed as follows: Among them, rand i Represents a random number between 0 and 1; r1', r2' represents the i Two individuals randomly selected from the neighborhood of X, and r1'≠r2'; r1, r2 represent two individuals randomly selected from the current population, and r1≠r2; X best represents the optimal individual of the current population; X pbest Indicates one of the top q% excellent individuals in the current population, q is set to 0.1; F i is a value randomly selected from [0.6, 0.8, 1.0]; Pa is a probability value used to control the performance emphasis of the mutation strategy in the two stages.

6. The two-stage multi-form differential evolution method for scheduling a mine comprehensive energy system according to claim 5 is characterized in that: In step 4, after the mutation operation, the conditional interval dominance relationship and interval crowding distance are used to evaluate individual performance, and the improved interval epsilon environment selection strategy is used to select excellent individuals to enter the next generation.

7. The two-stage multi-form differential evolution method for scheduling a mine integrated energy system according to claim 1 is characterized in that: In step 5, a two-stage multi-form differential evolution of the mine integrated energy system scheduling is performed, with an input population size of N and a maximum number of iterations of G. max , the stage transition parameter is β, and the evolution is controlled by the number of iterations G. When G≤β*G max When the evolution is in the first stage, otherwise it enters the second stage to obtain the optimal feasible solution set. The specific method is as follows: Step 5.1, offspring generation: In the first stage, the electric-heat task and the electric-cooling task correspond to the population P H and P C ,The two alternative tasks generate offspring through the diversity-driven neighborhood double mutation strategy respectively; Step 5.2: Individual evaluation: The first stage is to evaluate P H and P C Calculate the objectives and constraint violation values ​​under each task respectively, use the optimization results of the first stage as the initial population P of the second stage, and evaluate P under the original task in the second stage; Step 5.3, knowledge transfer: knowledge transfer only occurs in the first stage, and each generation is in the population P H and P C The corresponding feasible individuals and Based on the differences in the goals and constraints of the two alternative tasks, the goals and constraints of the transferred individuals are re-evaluated; Step 5.4, environment selection: In the first stage, an improved interval epsilon constraint processing mechanism is adopted to relax the constraints in the early stage of iteration and select potential individuals to enter the next generation. In the second stage, the interval constraint dominance principle is adopted; Step 5.5, stage transition: After the next generation of individuals is generated, determine whether to enter the second stage based on the state of the iteration number G. When G>β*G max When the evolution enters the second stage; Step 5.6, Final solution set: In the second stage, the initial population P generates offspring through the convergence-driven neighborhood double mutation strategy, and continues to evolve P until the termination condition is met, and finally obtains the interval non-dominated feasible solution set P F .

8. The two-stage multi-form differential evolution method for scheduling a mine comprehensive energy system according to claim 7 is characterized in that: The time complexity is analyzed in step 6, which includes the following sub-steps: Step 6.1: All individuals are evaluated and selected in each iteration. The time complexity of a single iteration is O(N). Since the two populations evolve independently in parallel in the first stage, the total time complexity is O(2NG max )=O(NG max ); Step 6.2: The knowledge transfer operation is performed once per generation, involving the screening and re-evaluation of feasible individuals. The total complexity of knowledge transfer is O(NG max ), will not significantly increase the overall time complexity; Step 6.3: In the second stage, the evolution of the initial population P is based on the optimization results of the first stage. The total time complexity of the second stage is O(NG max ); Step 6.4: Evolution also involves the dominance relationship based on the conditional interval and the interval crowding distance, and the time complexity is O(MN 2 ) and O(MNlogN); Step 6.5, the environment selection mechanism involves applying the interval constraint dominance principle and the improved interval epsilon constraint processing mechanism, with a complexity of O(MN 2 ); Step 6.

5. Considering all operations in the two stages, the overall time complexity of the algorithm is: OR(OF max +MN 2 +MNlogN)=O(NG max +MN 2 ) Where M is the number of objective functions, N is the population size; G max Indicates the maximum number of iterations.

9. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors implement the two-stage multi-form differential evolution method for scheduling a mine comprehensive energy system as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the program is executed by a processor, the steps in the two-stage multi-form differential evolution method for scheduling a mine comprehensive energy system as described in any one of claims 1 to 8 are implemented.

Images