CN113836706A - Numerical control machining parameter optimization method and storage medium - Google Patents

Abstract

The invention discloses a numerical control machining parameter optimization method and a storage medium, and belongs to the field of numerical control workpiece machining parameter optimization. The method comprises the steps of constructing an applicability function according to a workpiece processing optimization model with constraint conditions, solving by using an orthogonal double-chain differential evolution algorithm, carrying out optimal solution search through a population of orthogonal double chains, producing filial chains after orthogonal chain cross variation, carrying out orthogonal double-chain updating by adopting 3 strategies of orthogonal double-chain search tending to a global optimal solution, orthogonal double-chain search tending to the filial chains and global search with the global optimal solution as a center, enhancing the global search capability of a classical differential evolution algorithm and accelerating the convergence speed, thereby obtaining the optimal numerical control processing parameter solution set and the adaptability value. The invention takes the actual processing cost, the cost required by workpiece loading and unloading operation and cutter idle running, the tool changing operation cost, the cutter abrasion cost, and the cutting depth of each pass of rough machining and finish machining as optimization targets, and is suitable for common numerical control processing parameter optimization.

Description

Numerical control machining parameter optimization method and storage medium

Technical Field

The invention relates to a numerical control machining parameter optimization method and a storage medium, in particular to a numerical control machining parameter optimization method and a computer program product based on an orthogonal double-chain differential evolution algorithm with consideration of global search and local search strategies, and belongs to the technical field of numerical control workpiece machining parameter optimization and group intelligent optimization.

Background

Numerical control processing is used as a basic industry of manufacturing industry, and becomes an important way for accelerating economic development and improving comprehensive national strength and position of countries in the world, and occupies a main position in the contemporary industrial system of China. The main purpose of researching the machining parameters is to produce workpieces with low cost and high quality, the machining cost can be reduced by proper cutting parameters, the machining efficiency and the machining quality are improved to improve the economic benefit, and popularization and application of the work can provide more convenience for manufacturing enterprises to quickly respond to the market. At present, the numerical control processing enterprises in China still have the current situation that data control parameters are selected through experiments by means of experience and reference manuals, and the optimization of numerical control parameters is difficult to realize. The invention relates to an optimization problem of numerical control machining parameters, namely how to select proper machining parameters under a given machining constraint condition to achieve the minimum machining cost.

The early methods for optimizing numerical control machining parameters mainly comprise two methods: the first is a test method, such as factor design and a corresponding curved surface method; and secondly, a mathematical processing method such as dynamic programming and linear or nonlinear programming algorithm is used for seeking the optimal solution of the problem, however, a large number of processing parameter optimization problems are complex problems of multi-constraint nonlinearity, and the optimal solution is difficult to obtain by using the traditional mathematical processing method. With the rapid development of meta-heuristic algorithms, many scholars apply the meta-heuristic algorithms to the processing parameter optimization problem in the field of computer integrated manufacturing and search the near-optimal solution of the problem by using the meta-heuristic algorithms. At present, genetic algorithms, particle swarm algorithms, ant colony algorithms and the like related to the prior art documents as described below are adopted in many cases:

(a)Chen，M.and K.Chen，Optimization of multipass turning operations with genetic algorithms：A note[J].International Journal of Production Research，2010.41(14)：p.3385-3388.

(b)Chen，M.C.and D.M.TSAI，A simulated annealing approach for optimization of multi-pass turning operations[J].International Journal of Production Research，2007.34(10)：p.2803-2825.

(c)Sankar R S，Asokan P，Saravanan R，et al.Selection of machining parameters for constrained machining problem using evolutionary computation[J].The International Journal of Advanced Manufacturing Technology，2007，32(9-10)：892-901.

(d)Srinivas J，Giri R，Yang S H.Optimization of multi-pass turning using particle swarm intelligence[J].International Journal of Advanced Manufacturing Technology，2009，40(1-2)：56-66.

(e) lie xinpeng, improved artificial bee colony algorithm and application thereof in the cutting parameter optimization problem research [ D ] science and technology university in china, 2013.

However, the above documents have a dilemma that the meta-heuristic algorithm falls into local optimum in the process of finding a solution.

Disclosure of Invention

In order to solve the above problems, the present invention aims to provide a method and a storage medium for optimizing numerical control machining parameters, which aims to solve the problems of resource waste, low optimization efficiency and low precision in the actual numerical control machining process due to the unreasonable selection of numerical control machining parameters. Specifically, the invention uses the orthogonal double chains to realize the global search and local search strategies, uses the orthogonal chain crossing strategy to perfect the differential evolution structure, solves the problem of local optimum trapping existing in the traditional differential evolution algorithm, effectively avoids local optimum trapping, improves the accuracy and accelerates the convergence speed.

The invention achieves the aim through the following technical scheme:

a numerical control machining parameter optimization method comprises the following steps:

step 1, establishing a target evaluation function with the minimum unit cost in the numerical control machining process, determining a target to be optimized and a machining constraint condition, and determining a parameter optimization model of the numerical control machining according to the target evaluation function and the machining constraint condition;

and 2, constructing a fitness function according to the numerical control machining parameter optimization model, and solving the parameter optimization model by adopting an orthogonal double-chain differential evolution algorithm to obtain an optimal parameter set and a corresponding fitness value.

Further, the orthogonal double-strand differential evolution algorithm comprises the following steps:

(1) based on the initial population X, the population of particles is initialized with an orthogonal double-stranded structure, which is sine double-stranded X _ sin1 and X _ sin2 generated with sine and cosine double-stranded X _ cos1 and X _ cos2 generated with cosine. X1orth _ chain population orthogonal chain 1 is formed by X _ sin1 and X _ cos 1; x2orth _ chain orthogonal chain 2 is formed by X _ sin2 and X _ cos2, and a group orthogonal double chain is formed by Xlorth _ chain and X2orth _ chain;

(2) calculating the fitness value of each individual in the orthogonal double-stranded population; preferentially evolving the sine double-chain into a sine chain X _ sin according to the fitness value with the constraint condition meeting the mark ismETAll; in the same way, the optimized cosine double chains form cosine chains X _ cos. Forming an optimized population orthogonal chain Xorth _ chain by X _ sin and X _ cos;

(3) using the orthogonal cross strategy based on X_{orth_chain}Completing the crossover to generate a filial chain X_newUpdating the offspring chain after mutation; at the same time, the optimal solution X of the current iteration is updated_{local_best}Optimization with current iterationFitness value J_{X_local_best}；

(4) According to the current iteration optimal fitness value J_{x_local_best}And global optimum fitness value J_XbestDetermining an orthogonal chain search strategy; according to the global optimum solution X_bestAnd the child chain X_newUpdating the population orthogonal double chains; at the same time, the global optimal solution X is updated_bestAnd global optimum fitness value J_Xbest(ii) a Judging whether a termination condition is reached, if so, outputting a result; otherwise, repeating the steps (2) to (4) to continue solving.

Further, the parameters of the differential evolution algorithm of the orthogonal double chains include: the maximum iteration number Gen, the number Np of differential population particles and the dimension Dim of numerical control machining parameters of the differential population particles, an orthogonal chain crossing factor CR, a differential variation factor F and a continuous trapping Local optimum number threshold Trap _ Local _ times.

Further, the initialization of the orthogonal double strand is: randomly generating a population of offspring X in solution space, i.e.

X∈{x_d ^i，g/d＝1，2，...，Dim；g＝1，2，...，Gen；i＝1，2，...，Np}

x_d ^i，gRepresenting the d-dimension individual solution of the ith particle of the g generation;

generating sine double-strand X _ sin and cosine double-strand X _ cos from X, i.e.

X_sin＝{X_sin1＝X_bd·sin(X)；X_sin2＝-X_bd·sin(X)}，

X_cos＝{X_cos1＝X_bd·cos(X)；X_cos2＝-X_bd·cos(X)}，

X_bdIs a search space boundary value;

construction of the orthogonal duplexes: x1_{orth_chain}＝{X_sin1，X_cos1}，X2_{orth_chain}＝{X_sin2，X_cos2}。

Further, calculating the fitness value of each particle band constraint condition in the orthogonal double chains; the sine chain x _ sin and the cosine chain y _ cos are preferably selected according to the fitness value, i.e.

X_sin^i，g＝argbest{J_{x_sin1} ^i，g，J_{x_sin2} ^i，g/i＝1，2，...，Np；g＝1，2，...，Gen}

X_cos^i，g＝argbest{J_{x_cos1} ^i，g，J_{x_cos2} ^i，g/i＝1，2，...，Np；g＝1，2，...，Gen}

J_{x_sin1} ^i，gRepresents the fitness value of the ith generation of the ith particle of the sine chain x _ sin 1;

J_{x_sin2} ^i，grepresents the fitness value of the ith generation of the ith particle of the sine chain x _ sin 2;

J_{x_cos1} ^i，grepresents the fitness value of the ith generation of the ith particle of the cosine chain x _ cos 1;

J_{x_cos2} ^i，grepresents the fitness value of the ith generation of the cosine chain x _ cos 2.

Preferred for argbest are: (1) jx 'satisfying all constraints is better than Jx' not satisfying all constraints;

(2) when the same constraint condition is satisfied (fully satisfied or not satisfied), argbest is argmin.

Further, the objective function and the constraint condition are simultaneously integrated into a fitness function:

wherein f (x) is an optimized objective function, G_i(x) And H_j(x) Penalty functions, r, corresponding to inequality and equality constraints, respectively_iAnd s_jCorresponding penalty factors. For solutions satisfying all constraints, G_i(x) And H_j(x) Is 0, so that Φ (X) ═ f (X). For infeasible solutions, their fitness value is increased by a penalty function.

Further, a flag variable isMeetAll is introduced, the value of the fitness function is calculated, when all constraint conditions are met, the value of the flag variable isMeetAll is 1, otherwise, the value of the flag variable isMeetAll is 0.

Further, orthogonal chain crossing is carried out according to random probability to generate new filial generationGroup X_newIs shown as

CR_randIs cross probability random number, ranging from 0-1, Dim_randIs a dimension random number and ranges from 1 to Dim.

Further, according to the change of the optimal fitness value with the flag quantity isMeetAll, determining that the orthogonal chain search strategy updates the orthogonal double chains, namely:

an applicability value of isomeetall ═ 1 is better than an applicability value of isomeetall ═ 0;

if the optimal fitness value J of the current iteration_{X_local_best}Is better or less than the global fitness value J_XbestWhen using X_newTrend to global optimal solution X_bestPartial search strategy 1, updating the orthogonal double strand, i.e.

X_sin1＝(X_best+sin((X_new-X_best)/abs(X_bd-X_bew))·X_bd·α)/2

X_sin2＝(X_best-sin((X_new-X_best)/abs(X_bd-X_new))·X_bd·α)/2

X_cos1＝(X_best+cos((X_new-X_best)/abs(X_bd-X_new))·X_bd·α)/2

X_cos2＝(X_best+cos((X_new-X_best)/abs(X_bd-X_new))·X_bd·α)/2

α＝power(2，times_eff)

times_effCounting for significant iterations, i.e. successive occurrences of J_{X_local_best}Is better than or less than J_XbestNumber of times of

At the same time, the global optimal solution X is updated_bestGlobal optimum fitness value J_XbestAnd times_eff。

If J_{X_local_best}＝J_XbestAnd times_trapIf < Trap _ Local _ times, X is used_bestTrend towards X_newPartial search strategy 2, updating the orthogonal double strand, i.e.

X_sin1＝(X_best/times_trap+sin((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

X_sin2＝(X_best/times_trap-sin((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

X_cos1＝(X_best/times_trap+cos((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

X_cos2＝(X_best/times_trap-cos((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

α＝power(2，times_eff)

times_trapFor successive trapping in locally optimum statistical times, i.e. successive occurrence of J_{X_local_best}＝J_XbestThe number of times.

Update times_trap。

If J_{X_local_best}＝J_XbestAnd times_trapWhen Trap _ Local _ times, an orthogonal double-chain global search strategy taking a global optimal solution as a center is adopted to update the orthogonal double-chain, namely the orthogonal double-chain is updated

X_sin1＝X_best+X_bd·sin(θ)

X_sin2＝X_best-X_bd·sin(θ)

X_cos1＝X_best+X_bd·cos(θ)

X_cos2＝X_best-X_bd·cos(θ)

Theta is a random number and has a value in the range of 0-2 pi, i.e., theta is { theta ═ theta }_d ⁱ/i＝1，2，...Np；d＝1，2，...，Dim}

times_trapAnd times_effCarry out initializationAnd (4) setting.

Iteration is carried out through the searching strategy, and once the end condition of the algorithm is reached, the result and the fitness value of the final numerical control machining parameter optimization are output.

A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the numerical control machining parameter optimization method as described above.

Compared with the prior art, the invention has the beneficial effects that:

(1) according to the actual situation of the numerical control machining process, according to the model with the minimum unit cost of actual workpiece machining as the optimization target, the model is solved by adopting an orthogonal double-chain differential evolution algorithm, the optimal solution search is carried out on the population of the orthogonal double chains, the filial chains are produced after the orthogonal chains are subjected to cross variation, and the orthogonal double-chain search tending to the global optimal solution, the orthogonal double-chain search tending to the filial chains and the global search centering on the global optimal solution are adopted to implement orthogonal double-chain updating, so that the global search capability of the original differential algorithm is enhanced, the convergence speed is accelerated, the understanding precision is improved, a penalty function is added, and the global optimal solution is finally obtained.

(2) The method is suitable for the characteristics of nonlinearity, multiple extreme values, multiple targets and multiple constraints of a numerical control workpiece processing optimization model.

(3) The invention achieves a certain height on the precision of solving the problem of optimization of numerical control workpiece processing, reduces the actual processing cost and realizes the parameter optimization of numerical control processing.

Drawings

FIG. 1 is a flow chart of the numerical control workpiece processing parameter optimization method of the invention.

FIG. 2 is a flow chart of the differential evolution algorithm based on orthogonal double strands of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below by taking the optimization of the outer circle cutting of the numerical control machine tool, which is common in the production process, as an example, in the above-mentioned documents (a) to (e).

The numerical control workpiece processing parameter optimization method specifically comprises the following steps:

step 1, establishing a cutting parameter optimization model according to a cutting process in numerical control machining.

Step 1.1, firstly, establishing a target evaluation function according to the minimized unit production cost UC, wherein the unit production cost is the actual cutting processing cost C_MCost C required for workpiece handling operation and idle running of tool_IAnd cost of tool changing operation C_rThe wear cost component C of the cutting tool_T。

The establishment process of each cutting parameter optimization target is as follows:

step 1.1.1, minimizing unit cost UC, UC ═ C_M+C_I+C_r+C_T。

Step 1.1.2 cost C of actually performing cutting_M，

Wherein k is₀For direct and indirect labor costs, D and L are the workpiece diameter and length.

Cost C required for work handling and tool idle running_I，C_I＝k₀[t_c+(h₁L+h₂)(n+1)]Wherein t is_cTime for loading and unloading work, h₁、h₂N is the number of rough cuts, and is a constant relating to the turning tool idle time and advance/retreat time.

Step 1.1.3, tool change operation cost C_R，

Step 1.1.4, cost of tool wear C_T，

Wherein k is_tWhich is the cost of the blade.

And step 1.2, simultaneously, determining a cutting parameter optimization model by taking parameter value ranges such as cutting speed, feed quantity and cutting depth, tool service life, machined surface roughness, cutting force and cutting power, stable cutting area, temperature on contact surfaces of chips and tools, surface roughness, cutting depth, and mutual restriction conditions between rough and finish machining parameters as constraint conditions.

Wherein, the 8 types of constraint conditions are as follows:

(1) the ranges of values of parameters such as cutting speed, feed amount and cutting depth are taken as constraints, factors such as the processing quality of a workpiece, the processing efficiency and the personal safety of operators are considered, and the ranges of values of the parameters such as the cutting speed, the feed amount and the cutting depth are limited between a given maximum value and a given minimum value. The expression of the constraint is shown as follows:

v_rL≤v_r≤v_rU

f_rL≤f_r≤f_rU

d_rL≤d_r≤d_rU

v_sL≤v_s≤v_sU

f_sL≤f_s≤f_sU

d_sL≤d_s≤d_sU

wherein v is_r、f_r、d_rRespectively representing the cutting speed, the feeding amount and the cutting depth in the rough machining process; v. of_s、f_s、d_sRespectively representing the cutting speed, the feeding amount and the cutting depth in the finish machining process; v. of_rL、v_rURespectively representing the lower and upper limits of the rough cutting speed; v. of_sL、v_sURespectively representing the lower and upper limits of the cutting speed of finish machining; f. of_rL、f_rURespectively representing the lower and upper limits of the rough machining feed rate; f. of_sL、f_sUThe lower limit and the upper limit of the finishing feed rate are shown; d_rL、d_rURespectively representing the lower and upper boundaries of the rough cutting depth; d_sL、d_sUThe upper and lower limits of the fine cut depth are shown, respectively.

(2) Constrained by tool life and machined surface roughness:

T_L≤T_r≤T_U

T_L≤T_s≤T_U

wherein, T_rAnd T_sRespectively, the life expectancy of the tool in rough machining and finish machining, T_LAnd T_UThe sub-tables represent the lower and upper bounds of tool life.

(3) With cutting force and cutting power as constraints:

F_r＝k₁(f_r)^μ(d_r)^v≤F_u

F_s＝k₁(f_s)^μ(d_s)^v≤F_u

wherein, F_r、F_sRespectively, cutting forces in rough machining and finish machining, F_uIs an upper bound on the allowable cutting force. k is a radical of₁Mu and v are constants in the expression; the cutting power in the course of machining must not exceed the machine power, P_r、P_sDenotes the cutting power in roughing and finishing, respectively, P_uTo the upper bound of the permissible cutting power, η is the efficiency of the machine tool.

(4) With the stable cutting area as constraint:

(v_r)^λf_r(d_r)^v≤SC

(v_r)^λf_s(d_s)^v≤SC

where λ and v are constants in the expression, and SC represents a stable cutting region constraint upper limit.

(5) With temperature on the chip and tool interface as a constraint:

Q_r＝k₂(v_r)^τ(f_r)^Φ(d_r)^δ≤Q_u

Q_s＝k₂(v_s)^τ(f_s)^Φ(d_s)^δ≤Q_u

wherein Q is_rAnd Q_sIndicating the temperature, Q, at the chip-tool contact surface_uIs the upper temperature range; k is a radical of₂Tau, phi and delta are expression constants,

(6) Surface roughness constraint, surface roughness being a factor in measuring the quality of a workpiece product:

wherein, SR_UUpper bound representing allowable roughness values:

(7) the cutting depth constraint, one of the important parameters in the cutting depth type cutting process, and the total cutting depth of the cutter are the sum of the cutting depths of a plurality of times of rough machining and a time of finish machining.

Wherein d is_tDenotes the total depth of cut, n is the number of rough machining times and is a positive integer

(8) Constraint conditions for mutual restriction between rough machining parameters and finish machining parameters are as follows:

the cutting depth, the feed rate and the cutting depth in the rough machining and the finish machining have certain restricted relation:

v_s≥k₃v_r

f_r≥k₄f_s

d_r≥k₅d_s

wherein k is₃、k₄、k₅Is a constant of correlation.

Step 1.3, determining cutting parameters { v ] needing to be optimized_r、f_r、d_r、v_s、f_s、d_s}：

Cutting speed v of each rough machining_r；

Feed rate f per roughing_r；

Depth of cut d per roughing_r；

Finish machining cutting speed v_s；

Feed rate f for finishing_s；

Depth of cut d of finish machining_s。

In order to verify the implementation feasibility of the method, the technical method adopted according to the invention content obtains the machining process parameters of the machine tool according to the relevant documents (a) - (e) and the relevant design parameters of the numerical control machine tool, and completes the optimization objective and the constraint function.

TABLE 1 processing parameters

D＝50mm	L＝300mm	SC＝140	VrL＝50m/min	FrL＝0.1mm/rev
					DrL＝1mm	vrU＝500mm/rev	frU＝0.9mm/rev	DrU＝4mm	vsL＝50m/min
fsL＝0.1mm/rev	DsL＝0.5mm	vsU＝500m/min	frU＝0.9mm/rev	DsU＝2mm
					p＝5	q＝1.75	r＝0.75	v＝-1	μ＝0.75
v＝-1	η＝0.85	λ＝2	τ＝0.4	Φ＝0.2
					k0＝0.5$/min	k1＝108	k2＝132	k3＝1	k4＝2.5
k5＝1	kt＝2.5$/edge	δ＝0.105	R＝1.2mm	h1＝7*104
					h2＝0.3	C0＝6×10-11	te＝3.33×10-3min/edge	tc＝0.75min/piece	Pu＝15KW
TL＝25min	TU＝45min	FU＝200Kgf	QU＝1000℃	Sr＝10μm

And 2, solving the cutting parameter optimization model by using an orthogonal double-chain differential evolution algorithm.

Step 2.1, setting parameters of a differential evolution algorithm of the orthogonal double chains: maximum number of iterations g, number of population particles Np and dimension Dim (Dim 6, { v })_r、f_r、d_r、v_s、f_s、d_s}), an orthogonal chain crossing factor CR, a differential variation factor F and a threshold Trap _ Local _ times of continuous trapping Local optimum times; then, a difference population X is randomly generated in the solution space, i.e.:

X∈{x_d ^i，g/d＝1，2，...，Dim；g＝1，2，...，Gen；i＝1，2，...，Np}

x_d ^i，grepresents the d-dimension individual solution of the ith particle of the g generation.

Next, a sine double-strand X _ sin and a cosine double-strand X _ cos are generated by X, namely:

X_sin＝{X_sin1＝X_bd·sin(X)；X_sin2＝-X_bd·sin(X)}，

X_cos＝{X_cos1＝X_bd·cos(X)；X_cos2＝-X_bd·cos(X)}，

X_bdfor searching spaceBoundary value

Finally, the composition of the orthogonal duplexes: x1_{orth_chain}＝{X_sin1，X_cos1}，X2_{orth_chain}＝{X_sin2，X_cos2}。

2.2, a large number of constraint conditions exist in the cutting parameter model, the constraint conditions are nonlinear, and common differential evolution algorithms are random optimization algorithms aiming at the problem of no constraint, so that a penalty function method needs to be introduced into the algorithms, and a target function and the constraint conditions are simultaneously integrated into a fitness function.

Wherein f (x) is an optimized objective function, G_i(x) And H_j(x) Penalty functions, r, corresponding to inequality and equality constraints, respectively_iAnd s_jCorresponding penalty factors. For solutions satisfying all constraints, G_i(x) And H_j(x) Is 0, so that Φ (X) ═ f (X). For infeasible solutions, their fitness value is increased by a penalty function.

And (4) combining a penalty function, introducing a mark variable ismETAll, calculating the value of the fitness function, and if all constraint conditions are met, the value of the mark variable ismETAll is 1, otherwise, the value of the mark variable ismETAll is 0.

And 2.3, preferably selecting the orthogonal chains according to the fitness value of each individual in the orthogonal double chains with the constraint condition. Specifically, firstly, calculating the fitness of each individual in the orthogonal double chains with constraint conditions; then, the sine chain x _ sin and the cosine chain y _ cos are preferably selected according to the fitness value, namely:

X_{_}sin^i，g＝argbest{J_{x_sin1} ^i，g，J_{x_sin2} ^i，g/i＝1，2，...，Np；g＝1，2，...，Gen}

X_cos^i，g＝argbest{J_{x_cos1} ^i，g，J_{x_cos2} ^i，g/i＝1，2，...，Np；g＝1，2，...，Gen}

J_{x_sin1} ^i，gthe g-th generation of the sine chain x _ sin1Fitness value of the ith particle;

J_{x_sin2} ^i，grepresents the fitness value of the ith generation of the ith particle of the sine chain x _ sin 2;

J_{x_cos1} ^i，grepresents the fitness value of the ith generation of the ith particle of the cosine chain x _ cos 1;

J_{x_cos2} ^i，grepresents the fitness value of the ith generation of the cosine chain x _ cos 2.

Preferred for argbest are: (1) jx 'satisfying all constraints is better than Jx' not satisfying all constraints;

(2) when the same constraint condition is satisfied (fully satisfied or not satisfied), argbest is argmin.

Step 2.4, crossing the orthogonal chains according to the random probability to generate a new filial chain X_newExpressed as:

GR_randis cross probability random number, ranging from 0-1, Dim_randIs a dimension random number and ranges from 1 to Dim.

And 2.5, determining that the orthogonal chain search strategy updates the orthogonal double chains according to the change of the optimal fitness value with the constraint condition mark. In particular, the method comprises the following steps of,

an applicability value of isomeetall ═ 1 is better than an applicability value of isomeetall ═ 0;

if the optimal fitness value J of the current iteration_{X_local_best}Is better or less than the global fitness value J_XbestI.e. J_{x_local_best}＜J_XbestWhen using X_newTrend to global optimal solution X_bestThe local search strategy of (1) updates the orthogonal double-strand, namely:

X_sin1＝(X_best+sin((X_new-X_best)/abs(X_bd-X_new))·X_bd·α)/2

X_sin2＝(X_best-sin((X_new-X_best)/abs(X_bd-X_new))·X_bd·α)/2

X_cos1＝(X_best+cos((X_new-X_best)/abs(X_bd-X_new))·X_bd·α)/2

X_cos2＝(X_best+cos((X_new-X_best)/abs(X_bd-X_new))·X_bd·α)/2

α＝power(2，times_eff)

times_effcounting for significant iterations, i.e. successive occurrences of J_{X_local_best}Is better than or less than J_XbestNumber of times of

At the same time, the global optimal solution X is updated_bestAnd global optimum fitness value J_Xbest。

If J_{X_local_best}＝J_XbestAnd times_trapIf < Trap _ Local _ times, X is used_bestTrend towards X_newLocal search strategy of, updating orthogonal duplexes, i.e.

X_sin1＝(X_best/times_trap+sin((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

X_sin2＝(X_best/times_trap-sin((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

X_cos1＝(X_best/times_trap+cos((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

X_cos2＝(X_best/times_trap-cos((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

α＝power(2times_eff)

times_trapFor successive trapping in locally optimum statistical times, i.e. successive occurrence of J_{X_local_best}＝J_XbestThe number of times.

If J_{X_local_best}＝J_XbestAnd times_trapWhen Trap _ Local _ times, an orthogonal double-strand global search strategy with xtest as the center is adopted to update the orthogonal double-strand:

X_sin1＝X_best+X_bd·sin(θ)

X_sin2＝X_best-X_bd·sin(θ)

X_cos1＝X_best+X_bd·cos(θ)

X_cos2＝X_best-X_bd·cos(θ)

theta is a random number and has a value in the range of 0-2 pi, i.e., theta is { theta ═ theta }_d ⁱ/i＝1，2，...Np；d＝1，2，...，Dim}

Iteration is carried out through the search strategy, and once the end condition of the algorithm is reached, the final cutting parameter optimization result and the fitness value are output.

Will d_tThe results at 6.0mm are compared with the results of the algorithms employed in the above prior art documents (a) - (e), as shown in tables 2 and 3:

table 2 compares the optimal cutting parameters and fitness values with those of other optimization algorithms by using orthogonal double-chain differential evolution algorithm

As shown in table 2, the algorithm of the present invention has a minimum fitness value, obtains an optimal cutting parameter, and achieves a minimum unit cost.

TABLE 3 fitness value by orthogonal double-stranded differential evolution algorithm in comparison with other optimization algorithms

As shown in table 3, the fitness value obtained by the orthogonal double-stranded differential evolution algorithm is the minimum, and the average fitness value of multiple iterations is also the minimum, so that the method has fast and efficient parameter optimization performance compared with other algorithms. And document (e): compared with the MABC-inner algorithm, 11.79% of workpiece processing unit cost can be saved, and the cost saving is more remarkable compared with other algorithms.

The above results show that: in the aspect of solving the cutting parameter optimization problem, the orthogonal double-chain differential evolution algorithm has obvious advantages compared with other algorithms.

In conclusion, the cutting parameters are optimized through the differential evolution algorithm of the orthogonal double chains, the cutting parameter optimization model is established by taking the common numerical control machine tool excircle cutting optimization as an example, and finally, a reasonable and rapid numerical control workpiece processing parameter selection mode for guiding actual production is provided.

The above description is meant to be illustrative of the present invention and the technical principles employed, and not to limit the present invention, and any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. A numerical control machining parameter optimization method is characterized by comprising the following steps:

step 1, establishing a target evaluation function with the minimum unit cost in the numerical control machining process, determining a target to be optimized and a machining constraint condition, and determining a parameter optimization model of the numerical control machining according to the target evaluation function and the machining constraint condition;

and 2, constructing a fitness function according to the numerical control machining parameter optimization model, and solving the parameter optimization model by adopting an orthogonal double-chain differential evolution algorithm to obtain an optimal parameter set and a corresponding fitness value.

2. The numerical control machining parameter optimization method according to claim 1, wherein the orthogonal double-stranded differential evolution algorithm comprises the following steps:

(1) initializing a particle swarm by utilizing an orthogonal double-chain structure based on the initial population X, wherein the orthogonal double-chain structure is used for generating sine double-chains X _ sin1 and X _ sin2 by utilizing sine and generating cosine double-chains X _ cos1 and X _ cos2 by utilizing cosine; x1orth _ chain population orthogonal chain 1 is formed by X _ sin1 and X _ cos 1; x2orth _ chain orthogonal chain 2 is formed by X _ sin2 and X _ cos2, and a group orthogonal double chain is formed by X1orth _ chain and X2orth _ chain;

(2) calculating the fitness value of each individual in the orthogonal double-stranded population; preferentially evolving the sine double-chain into a sine chain X _ sin according to the fitness value with the constraint condition meeting the mark ismETAll; in the same way, optimizing cosine double chains to form cosine chains X _ cos; forming an optimized population orthogonal chain Xorth _ chain by X _ sin and X _ cos;

(3) using the orthogonal cross strategy based on X_{orth_chain}Completing the crossover to generate a filial chain X_newUpdating the offspring chain after mutation; at the same time, the optimal solution X of the current iteration is updated_{local_best}Optimal fitness value J with current iteration_{X_local_best}；

(4) According to the current iteration optimal fitness value J_{X_local_best}And global optimum fitness value J_XbestDetermining an orthogonal chain search strategy; according to the global optimum solution X_bestAnd the child chain X_newUpdating the population orthogonal double chains; at the same time, the global optimal solution X is updated_bestAnd global optimum fitness value J_Xbest(ii) a Judging whether a termination condition is reached, if so, outputting a result; otherwise, repeating the steps (2) to (4) to continue solving.

3. The numerical control machining parameter optimization method according to claim 2, characterized in that:

the parameters of the differential evolution algorithm of the orthogonal double chains comprise: the maximum iteration number Gen, the number Np of differential population particles and the dimension Dim of numerical control machining parameters of the differential population particles, an orthogonal chain crossing factor CR, a differential variation factor F and a continuous trapping Local optimum number threshold Trap _ Local _ times.

4. The numerical control machining parameter optimization method according to claim 3, characterized in that:

the initialization of the orthogonal double strand is as follows: randomly generating a population of offspring X in solution space, i.e.

X∈{x_d ^i,g/d＝1,2,…,Dim；g＝1,2,…,Gen；i＝1,2,…,Np}

x_d ^i,gRepresenting the d-dimension individual solution of the ith particle of the g generation;

generating sine double-strand X _ sin and cosine double-strand X _ cos from X, i.e.

X_sin＝{X_sin1＝X_bd·sin(X)；X_sin2＝-X_bd·sin(X)},

X_cos＝{X_cos1＝X_bd·cos(X)；X_cos2＝-X_bd·cos(X)},

X_bdIs a search space boundary value;

construction of the orthogonal duplexes: x1_{orth_chain}＝{X_sin1,X_cos1},X2_{orth_chain}＝{X_sin2,X_cos2}。

5. The numerical control machining parameter optimization method according to claim 4, characterized in that: in the step (2), calculating the fitness value of each particle band constraint condition in the orthogonal double chains; the sine chain x _ sin and the cosine chain y _ cos are preferably selected according to the fitness value, i.e.

X_sin^i,g＝argbest{J_{x_sin1} ^i,g,J_{x_sin2} ^i,g/i＝1,2,…,Np；g＝1,2,…,Gen}

X_cos^i,g＝argbest{J_{x_cos1} ^i,g,J_{x_cos2} ^i,g/i＝1,2,…,Np；g＝1,2,…,Gen}

J_{x_sin1} ^i,gRepresents the fitness value of the ith generation of the ith particle of the sine chain x _ sin 1;

J_{x_sin2} ^i,grepresents the fitness value of the ith generation of the ith particle of the sine chain x _ sin 2;

J_{x_cos1} ^i,grepresents the fitness value of the ith generation of the ith particle of the cosine chain x _ cos 1;

J_{x_cos2} ^i,grepresents the fitness value of the ith generation of the ith particle of the cosine chain x _ cos 2;

preferred for argbest are:

(1) jx 'satisfying all constraints is better than Jx' not satisfying all constraints;

(2) when the same constraint condition is satisfied (fully satisfied or not satisfied), argbest is argmin.

6. The numerical control machining parameter optimization method of claim 1, wherein the objective function and the constraint condition are simultaneously integrated into a fitness function:

wherein f (x) is an optimized objective function, G_i(x) And H_j(x) Penalty functions, r, corresponding to inequality and equality constraints, respectively_iAnd s_jIs the corresponding penalty factor; for solutions satisfying all constraints, G_i(x) And H_j(x) Is 0, such that Φ (X) ═ f (X); for unfeasible solutions, their fitness value is increased by the values of the penalty functions G (x) and H (x).

7. The numerical control machining parameter optimization method according to claim 6, characterized in that a flag variable isMeetAll is introduced, the value of the fitness function is calculated, and when all constraint conditions are satisfied, the value of the flag variable isMeetAll is 1, otherwise, it is 0.

8. The numerical control machining parameter optimization method according to claim 7, characterized in that: orthogonal chain crossing is carried out according to random probability to generate a new filial generation population X_newIs shown as

CR_randIs cross probability random number, ranging from 0-1, Dim_randIs a dimension random number and ranges from 1 to Dim.

9. The numerical control machining parameter optimization method according to claim 8, characterized in that: determining the orthogonal chain search strategy to update the orthogonal double chains according to the change of the optimal fitness value of the isometall with the mark quantity, namely:

an applicability value of isomeetall ═ 1 is better than an applicability value of isomeetall ═ 0;

if the optimal fitness value J of the current iteration_{X_local_best}Is better or less than the global fitness value J_XbestWhen using X_newTrend to global optimal solution X_bestPartial search strategy 1, updating the orthogonal double strand, i.e.

X_sin1＝(X_best+sin((X_new-X_best)/abs(X_bd-X_new))·X_bd·α)/2

X_sin2＝(X_best-sin((X_new-X_best)/abs(X_bd-X_new))·X_bd·α)/2

X_cos1＝(X_best+cos((X_new-X_best)/abs(X_bd-X_new))·X_bd·α)/2

X_cos2＝(X_best+cos((X_new-X_best)/abs(X_bd-X_new))·X_bd·α)/2

α＝power(2,times_eff)

times_effCounting for significant iterations, i.e. successive occurrences of J_{X_local_best}Is better than or less than J_XbestNumber of times of

At the same time, the global optimal solution X is updated_bestGlobal optimum fitness value J_XbestAnd times_eff；

If J_{X_local_best}＝J_XbestAnd times_trap<When Trap _ Local _ times, X is used_bestTrend towards X_newPartial search strategy 2, updating the orthogonal double strand, i.e.

X_sin1＝(X_best/times_trap+sin((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

X_sin2＝(X_best/times_trap-sin((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

X_cos1＝(X_best/times_trap+cos((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

X_cos2＝(X_best/times_trap-cos((X_new-X_best)·times_trap/abs(X_bd-X_new))·X_bd·α)/2

α＝power(2,times_eff)

times_trapFor successive trapping in locally optimum statistical times, i.e. successive occurrence of J_{X_local_best}＝J_XbestThe number of times of (c);

update times_trap；

If J_{X_local_best}＝J_XbestAnd times_trapWhen Trap _ Local _ times, an orthogonal double-chain global search strategy taking a global optimal solution as a center is adopted to update the orthogonal double-chain, namely the orthogonal double-chain is updated

X_sin1＝X_best+X_bd·sin(θ)

X_sin2＝X_best-X_bd·sin(θ)

X_cos1＝X_best+X_bd·cos(θ)

X_cos2＝X_best-X_bd·cos(θ)

Theta is a random number and has a value in the range of 0-2 pi, i.e.

times_trapAnd times_effCarrying out initialization setting;

iteration is carried out through the searching strategy, and once the end condition of the algorithm is reached, optimized numerical control machining parameters and the fitness value of the numerical control machining parameters are output.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that:

the program, when executed by a processor, implements the steps of a method of numerically controlled machining parameter optimization as claimed in any one of claims 1 to 9.

Images