阅读说明：本技术 一种基于自适应遗传算法的多金属多目标配矿方法 (A kind of more metal multiple target ore-proportioning methods based on self-adapted genetic algorithm ) 是由 顾清华 刘海龙 王迪 卢才武 冯治东 李学现 李俊飞 于 2019-08-30 设计创作，主要内容包括：一种基于自适应遗传算法的多金属多目标配矿方法,根据矿山配矿实际生产要求和指标,构建多金属多目标配矿模型,包括：品位偏差最小目标、采掘成本和运输成本最小目标、总产量偏差最小目标、矿石岩性的配比偏差最小目标、任务量约束、生产能力约束、氧化率约束、有害物质约束以及运输量约束；然后采用自适应遗传算法对多金属多目标配矿模型进行求解。本发明通过建立以采矿和运输成本、配矿后品位偏差、总产量偏差和配矿后矿石岩性配比偏差最小为目标的配矿模型,并采用自适应遗传优化算法,对多金属多目标配矿问题进行求解,实现了多金属多目标配矿的优化。(A kind of more metal multiple target ore-proportioning methods based on self-adapted genetic algorithm, summing target is wanted according to mine ore matching actual production, construct more metal multiple target ore matching models, comprising: grade deviation minimum target, cost of mining and transportation cost minimum target, total output deviation minimum target, the proportion deviation minimum target of ore lithology, task amount constraint, capacity constraint, oxygenation efficiency constraint, harmful substance constraint and freight volume constraint；Then more metal multiple target ore matching models are solved using self-adapted genetic algorithm.The present invention is dug up mine by establishing matches the ore matching model of the minimum target of deviation with ore lithology after grade deviation, total output deviation and ore matching after transportation cost, ore matching, and use Adaptive Genetic optimization algorithm, more metal multiple target ore matching problems are solved, the optimization of more metal multiple target ore matchings is realized.)

1. a kind of more metal multiple target ore-proportioning methods based on self-adapted genetic algorithm, which comprises the steps of:

Step 1, summing target is wanted according to mine ore matching actual production, it is as follows constructs more metal multiple target ore matching models:

1) grade deviation minimum target:

Wherein, x_ijFor ore locations i to the Ore Transportation amount by mine point j；g_ai, g_bi, g_ciRespectively a in ore locations i ore, b, c tri- Kind of metal for mine grade；g_aj, g_bj, g_ciRespectively by a in mine point j ore, the target grade of tri- kinds of metals of b, c；M, n difference For ore locations i and by the number of mine point j；

2) cost of mining and transportation cost minimum target:

Wherein, C_1ij, C_2ijRespectively indicate ore locations i unit cost of winning and ore locations i to the unit haul distance by mine point j at This；L_ijIndicate ore locations i to the haul distance by mine point j；

3) total output deviation minimum target:

Wherein, K indicates the target output of ore matching plan phase；

4) the proportion deviation minimum target of ore lithology:

Wherein, in order to solve simplicity, it is classified as one kind when the ore locations for belonging to same lithology are sorted, by different lithology classification Classification and ordination respectively, that is, ore locations 1 all belong to a class lithology to ore locations a, and ore locations a+1 to ore locations b all belongs to In b class lithology, and so on；u_aj, u_bj, u_cj... it is respectively after ore matching by a in mine point j ore, b, c ... class lithology Target proportion；

5) task amount constrains:

Wherein, Q_i, A_iRespectively indicate the minimum task amount and maximum task amount of ore locations i；q_j,p_jIt respectively indicates by mine point j most Small task amount and maximum task amount；

6) capacity constraint:

Wherein, M_i, N_jRespectively each ore locations and by mine point production and processing ability maximum limit the quantity；

7) oxygenation efficiency constrains:

Wherein, g_diIndicate the content of the ore oxidation rate of ore locations i；By the requirement of the ore oxidation rate of mine point after a expression ore matching Match threshold limit value；

8) harmful substance constrains:

Wherein, g_wfiIndicate the percentage composition of w kind harmful substance in the ore of ore locations i；b_wBy in mine point after expression ore matching The limit value of the w kind harmful substance contents ratio of ore；

9) freight volume constrains:

c≤x_ij≤ d (i=1,2,3 ... m；J=1,2,3 ... is n)

Wherein, c, d respectively indicate ore locations to the minimum by mine point, maximum freight volume requirement；

Step 2, more metal multiple target ore matching models are solved using self-adapted genetic algorithm.

2. more metal multiple target ore-proportioning methods based on self-adapted genetic algorithm according to claim 1, which is characterized in that institute Self-adapted genetic algorithm is stated by improving to obtain to basic genetic algorithmic, development is as follows:

1) crossing-over rate and aberration rate formula are improved are as follows:

Wherein, P_cFor crossing-over rate, P_mAberration rate；f_maxFor the maximum adaptation angle value of all individuals in population；f_avgFor institute in population There is the average fitness value of individual；F ' is biggish fitness value in two individuals for participating in intersecting；F is the adaptation of variation individual Angle value；P_cmaxFor maximum crossing-over rate；P_cminFor the smallest crossing-over rate；P_mmaxFor maximum aberration rate；P_mminFor the smallest variation Rate；

2) mutation operation is improved are as follows:

Multiply the character number that individual dimension obtains each individual need variation first with aberration rate:

Gene (i)=d*P_m(i)

Wherein, gene (i) indicates the character number that i-th of body solution needs to make a variation in population；D refers to the dimension of individual, that is, String length；P_m(i) aberration rate of i-th of body in population is indicated；

Then change that above formula is calculated to be needed to make a variation at random in limits since last character of each individual The correspondence number of character, other characters do not have to become:

X "=X₁+1b+rand*(ub-1b)

Wherein, X " refers to the new individual that variation generates；X₁Refer to the part that variation is not needed in former individual；1b refers to each word in individual solution The minimum value of symbol；Ub refers to the maximum value of each character in individual solution；Rand indicates the random number on section (0,1)；

3) selection course is improved are as follows:

All solutions including female solution after a wheel cross and variation are compared, are left in the way of roulette solid Determine the individual of number, before keeping every wheel to start, the dimension of solution immobilizes.

3. more metal multiple target ore-proportioning methods based on self-adapted genetic algorithm according to claim 2, which is characterized in that adopt The process solved with self-adapted genetic algorithm to more metal multiple target ore matching models is as follows:

The parameter of algorithm, including population scale N, individual dimension d, maximum number of iterations t is arranged in Step 1_max, maximum, minimum to hand over Fork rate P_cmax、P_cminWith maximum, minimum aberration rate P_mmax、P_mmin；

Step 2 generates initial population at random, calculates the adaptive value of each individual；

Step 3, according to the crossing-over rate and aberration rate of the fitness value calculation individual of individual, and respectively with (0,1) that generates at random Between number be compared；

The individual that crossing-over rate in Step 3 is greater than random number is matched two-by-two at random, carries out crossover operation by Step 4, will The individual that aberration rate is greater than random number in Step 3 carries out mutation operation；

All solutions including female solution after a wheel cross and variation are compared, utilize the side of roulette by Step 5 Formula leaves individual identical with originally population number, and before keeping every wheel to start, the dimension of solution is all fixed and invariable；

Step 6, the solution that will be left behind goes to Step 2, until reaching termination condition, then terminates algorithm output result.

Technical field

The invention belongs to Mining system engineering and mine optimisation technique fields, in particular to a kind of to be calculated based on Adaptive Genetic More metal multiple target ore-proportioning methods of method.

Background technique

Ore matching is to guarantee the important means of grade of ore equilibrium and resource recycling in mine production, with resource The comprehensive reutilization of exploitation, more metal multiple target Blending optimization problems become one of the focus of mining industry circle common concern.Science Reasonable more metal multiple target ore matchings can effectively guarantee that ore matching grade is balanced, reduce the transportation cost of enterprise, significantly improve mine The comprehensive utilization ratio and economic benefit of stone.From the point of view of both at home and abroad, at present in the actual production of ore matching, most of ore-proportioning method does not have Consider to influence the factors such as ore lithology, oxygenation efficiency and harmful substance of ore dressing link, or to these factors the considerations of is insufficient, but These factors have important influence to ore dressing, consider that these factors can not only improve ore quality in ore matching, improve ore The rate of recovery, additionally it is possible to economize on resources, improve the performance of enterprises, meet the requirement of sustainable development.Therefore in order to meet current metal Actual demand of the opencut to more metal multiple target ore matchings, it is necessary to study a kind of more mesh of more metals for fully considering these factors Mark ore-proportioning method.

Summary of the invention

In order to overcome the disadvantages of the above prior art, the more metal multiple target ore matchings of current Metal Open are solved the problems, such as, this Invention is designed to provide a kind of more metal multiple target ore-proportioning methods based on self-adapted genetic algorithm, establishes to dig up mine and transport Ore lithology matches the ore matching model of the minimum target of deviation after grade deviation, total output deviation and ore matching after defeated cost, ore matching, And propose a kind of Adaptive Genetic optimization algorithm, more metal multiple target ore matching problems are solved, the more mesh of more metals are realized The optimization of standard configuration mine.

To achieve the goals above, the technical solution adopted by the present invention is that:

A kind of more metal multiple target ore-proportioning methods based on self-adapted genetic algorithm, include the following steps:

Step 1, summing target is wanted according to mine ore matching actual production, constructs more metal multiple target ore matching models, comprising: product Position deviation minimum target, cost of mining and transportation cost minimum target, total output deviation minimum target, the proportion of ore lithology are inclined Poor minimum target, task amount constraint, capacity constraint, oxygenation efficiency constraint, harmful substance constraint and freight volume constraint；

Step 2, more metal multiple target ore matching models are solved using self-adapted genetic algorithm.

Compared with prior art, the present invention introduces ore lithology, oxygenation efficiency and has on the basis of existing ore matching model The influence factors such as evil substance construct more metal multiple target ore matching models of opencut, and using self-adapted genetic algorithm to mostly golden Belong to multiple target ore matching model solution.Than script are considered to influence factors such as ore lithology, oxygenation efficiency and harmful substances insufficient match Mine model more meets ore matching actual production requirement.Mostly use the method for linear programming to solve ore matching model, still originally Since ore matching complicated condition tends not to obtain actually active ore matching as a result, this self-adapted genetic algorithm proposed by the present invention Principle is simple, easy to accomplish, and has feasibility and superiority in processing challenge, to more metal multiple target ore matching models It carries out in solution procedure, can be quickly obtained and meet practical and effective ore matching result.

The present invention improves the utilization rate of ore to the quality for stablizing milling ore, improves ore matching efficiency, reduces transportation cost It has great significance.

Detailed description of the invention

Fig. 1 is in the present invention using the flow chart of adaptive more genetic algorithm solving models.

Specific embodiment

With reference to the accompanying drawing with example in detail embodiments of the present invention.

More metal multiple target ore matching problems of Metal Open can be described as: there are m ore locations on certain more Metal Open mountain, N is a by mine point, and mining area mainly produces three kinds of molybdenum, tungsten, copper metals；Lithology mainly has skarn type, saturating oblique angle rock, wollastonite angle Four kinds of rock, halleflinta lithology；Harmful substance is mainly calorize mud and sulfur dioxide, and the limit value that dressing plant requires is respectively 4% With 3%；The ore oxidation rate limit value that dressing plant requires is 16%；As the grade of each ore locations ore of the following table 1, ore lithology, The indexs such as oxygenation efficiency are all discrepant, and some indexs can't directly meet the requirement in dressing plant.Pass through reasonable distribution Each ore locations are to the yield x by mine point_ij, so that the requirement that can be met dressing plant by the indices of mine point ore is as follows Table 2 and 3, and make digging and transportation cost minimum.

1 part exploiting field stope grade of table, oxygenation efficiency, lithology situation

2 part of table is matched requirements of plan by the lithology of mine point

Grade requirements of plan of 3 part of table by mine point

Project	By mine point 1	By mine point 2	By mine point 3	By mine point 4
					Mo	0.110	0.098	0.073	0.110
Wo3	0.125	0.056	0.126	0.125
					Cu	0.011	0.006	0.014	0.011

A kind of more metal multiple target ore-proportioning methods of Metal Open of the present invention, include following steps:

1, it determines that summing target is wanted in surface mine ore matching actual production, summing target is wanted according to mine ore matching actual production, More metal multiple target ore matching models are constructed, need to comprehensively consider the production task requirement in mine, maximum productivity, target in the process The many factors such as grade requirement, mining and transportation cost, ore lithology, oxygenation efficiency and harmful substance.

Specifically, summing target is wanted according to mine actual production, constructs more metal multiple target ore matching models, model is as follows:

1) grade deviation minimum target:

Wherein, x_ijFor ore locations i to the Ore Transportation amount by mine point j；g_ai, g_bi, g_ciRespectively a in ore locations i ore, Tri- kinds of metals of b, c for mine grade；g_aj, g_bj, g_ciRespectively by a in mine point j ore, the target grade of tri- kinds of metals of b, c；M, n Respectively ore locations i and the number by mine point j.

2) cost of mining and transportation cost minimum target:

Wherein, C_1ij, C_2ijThe unit cost of winning and ore locations i for respectively indicating ore locations i are to the unit haul distance by mine point j Cost；L_ijIndicate ore locations i to the haul distance by mine point j.

3) total output deviation minimum target:

Wherein, K indicates the target output of ore matching plan phase.

4) the proportion deviation minimum target of ore lithology:

Wherein, in order to solve simplicity, it is classified as one kind when the ore locations for belonging to same lithology are sorted here, by difference Lithology classification distinguishes classification and ordination, that is, ore locations 1 all belong to a class lithology, ore locations a+1 to ore removal to ore locations a Point b all belongs to b class lithology, and so on；u_aj, u_bj, u_cj... respectively by a, b, c ... in mine point j ore after ore matching The target of class lithology matches；

5) task amount constrains:

Wherein, Q_i, A_iRespectively indicate the minimum task amount and maximum task amount of ore locations i；q_j,p_jIt respectively indicates by mine point j Minimum task amount and maximum task amount.

6) capacity constraint:

Wherein, M_i, N_jRespectively each ore locations and by mine point production and processing ability maximum limit the quantity.

7) oxygenation efficiency constrains:

Wherein, g_diIndicate the content of the ore oxidation rate of ore locations i；By the ore oxidation rate of mine point after a expression ore matching It is required that proportion threshold limit value.

8) harmful substance constrains:

Wherein, g_wfiIndicate the percentage composition of w kind harmful substance in the ore of ore locations i；b_wBy mine after expression ore matching The limit value of the w kind harmful substance contents ratio of ore in point.

9) freight volume constrains.For each ore locations to the freight volume by mine point, if the task amount arranged is too small, that Just it is ready to receive without fleet, meets fleet to freight volume to the yield by mine point by each ore locations of reasonable arrangement It is required that:

c≤x_ij≤ d (i=1,2,3 ... m；J=1,2,3 ... is n) (11)

Wherein, c, d respectively indicate ore locations to the minimum by mine point, maximum freight volume requirement.

2, basic genetic algorithmic is improved to obtain a kind of self-adapted genetic algorithm, specific development is as follows:

1) improvement of crossing-over rate and aberration rate

Traditional genetic algorithm, crossing-over rate and aberration rate are fixed, so that the efficiency of genetic algorithm is relatively low, and And it is easy to fall into local optimum.Crossing-over rate and aberration rate should with the progress of operation, according to individual with solved in population and population Dispersion degree variation and change, efficiency of algorithm can be improved in this way, avoid falling into local optimum.

The intersection of improved self-adapted genetic algorithm and aberration rate formula below:

Wherein, P_cFor crossing-over rate, P_mAberration rate；f_maxFor the maximum adaptation angle value of all individuals in population；f_avgFor population In all individuals average fitness value；F ' is biggish fitness value in two individuals for participating in intersecting；F is variation individual Fitness value；P_{c max}For maximum crossing-over rate；P_{c min}For the smallest crossing-over rate；P_{m max}For maximum aberration rate；P_{m min}For minimum Aberration rate.

2) improvement for the mode that makes a variation

General mutation operation is exactly that the character of the fixed numbers for the individual that will carry out mutation operation is carried out limitation model Enclose interior random variation, the number of characters and position to make a variation every time in this way be it is fixed, be not easy to generate multifarious individual.If The number of characters to make a variation every time and position are changed, it will be able to enrich the diversity of individual, it is easier to jump out locally optimal solution. The mutation operation of self-adapted genetic algorithm used in this problem are as follows:

Multiply the character number that individual dimension obtains each individual need variation first with aberration rate:

Gene (i)=d*P_m(i) (14)

Wherein, gene (i) indicates the character number that i-th of body solution needs to make a variation in population；D refers to the dimension of individual, It is exactly string length；P_m(i) aberration rate of i-th of body in population is indicated.

Then change the calculated needs of above formula at random in limits since last character of each individual The correspondence number of variation character, other characters do not have to become:

X "=X₁+1b+rand*(ub-1b) (15)

Wherein, X " refers to the new individual that variation generates；X₁Refer to the part that variation is not needed in former individual；1b refers to every in individual solution The minimum value of a character；Ub refers to the maximum value of each character in individual solution；Rand indicates the random number on section (0,1).

3) improvement of selection course

Traditional genetic algorithm is after the cross and variation of a wheel, without more directly being replaced with new explanation for former solution and new explanation Original solution, or after directly select in new explanation and former solution preferably, the former can reduce the efficiency of algorithm and be not easy to find optimal Solution, the latter can reduce the diversity of solution individual and be easily trapped into locally optimal solution.So if we compare original solution and new explanation Afterwards, it is retained more excellent solution maximum probability, and those more bad solutions can also have lesser probability to stay part, in this way may be used To increase the diversity of individual solution while improving efficiency of algorithm, it is easier to jump out local optimum.The selection that this problem is taken Process specifically:

All solutions (including female solution) after a wheel cross and variation are compared, are left in the way of roulette solid Determine the individual of number, before keeping every wheel to start, the dimension of solution is all fixed and invariable.

3, more metal multiple target ore matching models are solved using self-adapted genetic algorithm, specific solution procedure is as follows:

The parameter of algorithm, including population scale N, individual dimension d, maximum number of iterations t is arranged in Step 1_max, maximum, most Small crossing-over rate P_{c max}、P_{c min}With maximum, minimum aberration rate P_{m max}、P_{m min}, by multiple test of heuristics, the design parameter drafted Value such as the following table 4.

4 algorithm partial parameters value of table

Parameter	Occurrence
		Population scale N	100
Individual dimension d	m*n
		Maximum number of iterations tmax	1000
Maximum, minimum crossing-over rate Pc max、Pc min	0.9、0.6
		Maximum, minimum aberration rate Pm max、Pm min	0.5、0.1

Step 2 generates initial population at random, calculates the adaptive value (fitness function determined by objective function of each individual To determine).Since the solution of this problem constrains in section (c, d), can directly be given birth in MATLAB software using formula (16) Produce initial solution.

X (i, j)=e+ (f-e) * rand () (16)

Step 3 according to the crossing-over rate and aberration rate of the fitness value calculation individual of individual, and respectively with generate at random Number between (0,1) is compared.

Step 4 is matched the individual that crossing-over rate in Step 3 is greater than random number two-by-two at random, carries out crossover operation, The individual that aberration rate in Step 3 is greater than random number is subjected to mutation operation.

All solutions (including female solution) after a wheel cross and variation will be compared by Step 5, utilize the side of roulette Formula leaves individual identical with originally population number, and before keeping every wheel to start, the dimension of solution is all fixed and invariable.

The solution that Step 6 will be left behind goes to Step 2, until reaching termination condition, then terminates algorithm output result.

The present invention is directed to the more metal multiple target ore matching problems of opencut, mining minimum with grade deviation after ore matching and transport The minimum target of ore lithology percent deviation after cost minimization, total output deviation and ore matching, constructs more metal multiple targets Blending optimization model efficiently solves the problems, such as more metal multiple target ore matchings of opencut.For more metal multiple target Blending optimizations Model introduces self-adapted genetic algorithm, realizes the scientific and reasonable solution of more metal multiple target ore matching models.Above in conjunction with attached drawing The present invention is exemplarily described, it is clear that present invention specific implementation is not subject to the restrictions described above, as long as using this The improvement for the various unsubstantialities that the method concept and technical solution of invention carry out, or it is not improved by design and skill of the invention Art scheme directly applies to other occasions, within the scope of the present invention.