CN104881530A - Hobbing dry-cutting processing method based on optimized technical parameter - Google Patents

Abstract

The invention discloses a hobbing dry-cutting processing method based on an optimized technical parameter. The hobbing dry-cutting processing method is characterized in that the dry-cutting processing technical parameter is optimized according to the following main steps: (1) automatically grouping a living example set of a hobbing dry-cutting technique; (2) selecting an optimized group of the technical example of the hobbing dry-cutting based on similar characteristics; (3) structuring the optimized group network of the hobbing dry-cutting processing technique example; (4) Making a parameter optimization decision of the hobbing dry-cutting technique based on a webpage sequencing algorithm. The hobbing dry-cutting processing method has the following advantages: in the hobbing dry-cutting processing, the processing technique parameter is improved by using graph theory and webpage sequencing algorithm; the hobbing dry-cutting processing parameter is optimized; the method is quick in optimization of technique parameter and high in processing precision.

Description

The dry cutting processing method of a kind of gear hobbing based on Optimizing Process Parameters

Technical field

The present invention relates to gear machining technology, especially relate to the job operation in a kind of gear hobbing process process, technological parameter is optimized.

Background technology

To be that gear hobbing is dry to cut in applied research an important gordian technique to process parameter optimizing, and the cutting speed of cutting because gear hobbing is dry, the amount of feeding sharply raise, and the dry cutting process parameter optimization of gear hobbing has the complicacy being different from conventional cutting.The determination of cutting hobbing technological parameter is particularly important.But the relevant Cutting data accumulated in real process and craft embodiment less, making technologist when matching technological parameter, needing the cost plenty of time to collect required knowledge and data.In addition common process due to machining precision and working (machining) efficiency lower, the data accumulated and example can not indiscriminately imitate use.Therefore, the dry cutting process parameter of gear hobbing how effectively utilizing existing knowledge and data acquisition to be rationally suitable for, becomes problem demanding prompt solution.

It is undesirable that coupling case process scheme is applied to processing effect when decision-making technological problems, and occur that same process problem adopts multiple craft embodiments of different process scheme cannot the problem of decision-making because similarity value calculation is identical.Graph theory is as analyzing the important tool of relation between things and providing a good thinking based on the process parameter optimizing decision-making technique of webpage sorting for solving the problem, namely utilize existing craft embodiment to solve to be applicable to the technological parameter treating decision-making technological problems, the Cutting data and experimental knowledge that accumulate in craft embodiment can be made full use of.

Summary of the invention

For prior art above shortcomings, the object of the present invention is to provide a kind of dry cutting process parameter optimization method of gear hobbing based on graph theory and Algorithms for Page Ranking, this method is according to the dry cutting process example of accumulated gear hobbing, realize the optimization grouping of craft embodiment collection, building after optimizing group network adopts Algorithms for Page Ranking to carry out importance ranking to gear hobbing dry cutting process example Optimum Matching, draw high-quality craft embodiment, so just, provide set of parameter optimization method to new technology problem, improve calculating processing efficiency, optimizing machining technology effect.

In order to solve the problems of the technologies described above, in the present invention, have employed following technical scheme:

The dry cutting processing method of a kind of gear hobbing based on Optimizing Process Parameters, this method adopts the dry machine of cutting of numerical control gear hobbing to carry out gear hobbing process, it is characterized in that, during gear hobbing process, carry out the optimization of the gear hobbing process technological parameter based on graph theory and webpage sorting according to following steps, concrete steps are:

(1) Auto-grouping of gear hobbing dry cutting process example set is realized: set up existing example similar features according to gear hobbing dry cutting process example similarity feature and associate non-directed graph of having the right; By calculating the minimum spanning tree MST of non-directed graph of having the right; Crop the limit that weights in figure are greater than craft embodiment similarity threshold θ, θ generally gets 0.15, and craft embodiment similarity being greater than preset value is birdsed of the same feather flock together, and obtain n similar group, n is positive integer; Namely by the segmentation to non-directed graph minimum spanning tree, similar group of gear hobbing dry cutting process example is generated;

(2) realize choosing based on the dry craft embodiment optimization group of cutting of gear hobbing of similar features: the centre of similarity finding out each similar group according to the limit weights of the similarity of example, calculate respectively and treat the limit weights of decision-making technological problems and each centre of similarity, getting minimum similar group of limit weights is technique to be processed nearest craft embodiment optimization group;

(3) structure of gear hobbing dry cutting process example optimal group network is realized: gear hobbing dry cutting process example optimal group network is a digraph, the craft embodiment optimization group obtained in step (2) is abstracted into digraph, in optimization group, each craft embodiment is mapped as the node of figure, mutual adduction relationship in group between craft embodiment is mapped as corresponding directed edge in figure, set between any two summits and have direct path, shift by original blue direction with probability d at each Nodes, with probability 1-d by red direction transfer; Wherein d is damping factor; Random and the non-zero PR value of given each node one is as the initial value of webpage sorting;

(4) based on the gear hobbing dry cutting process parameter optimization decision-making of Algorithms for Page Ranking: first, according to the gear hobbing dry cutting process example optimal group network that step (3) builds, obtain adjacency matrix P, transition probability matrix A, the proper vector of A is the initial value of PR value, is namely the initial value of webpage sorting value; Then, in conjunction with treating decision-making technological problems and the similarity optimizing each craft embodiment in group network, use webpage sorting innovatory algorithm to calculate the PR value of each craft embodiment, namely what PR value was the highest is and treats the dry cutting process example of the gear hobbing of decision-making technological problems Optimum Matching; Then complete dry the cutting of gear hobbing according to this example to process.

As optimization, the concrete steps that the gear hobbing dry cutting process example optimal group network in described step (3) builds comprise,

Step1 builds craft embodiment information node (b, J), and it is a tuple (L _iD), be used for preserving the craft embodiment information of user under technological problems k, comprise the numbering ID of craft embodiment, b, J, k, ID are positive integer;

Step2 is set damping factor d, PR value and will be calculated by iteration, and calculated amount is larger, for improving the speed of retrieval, PR value is calculated by the mode of off-line, thus first calculates PR value by conventional web sort algorithm, ratio of damping d is added again, the general value d=0.85 of d on the basis of simple formula;

The adduction relationship of Step3 factually between example, completes the weighted digraph of webpage sorting underground pipe net model;

Step4 sets threshold values z=0.15, crops the limit that limit weights are less than threshold values, completes the network struction of optimal set.

As optimization, the concrete steps of the dry cutting process of the gear hobbing based on the Algorithms for Page Ranking parameter optimization in described step (4) comprise,

Step1, according to digraph, defines adjacency matrix P, if example j quotes to example i, then and p _ij=1, on the contrary p _ij=0, p _ijfor the value in the i-th row jth row in matrix P, i, j are positive integer;

Step2 remember matrix P row and, row and.Row and: row and they sets forth craft embodiment j quoted number of times and its decision-making by other examples time quote the number of times of other craft embodiments;

Step3 Markov chain describes the state metastatic rule between each craft embodiment; Definition transition probability matrix $A=\left(a_{\mathrm{ij}}\right),a_{\mathrm{ij}}=\frac{p_{\mathrm{ij}}}{c_j},$ Wherein i, j=1,2 ... .n;

According to the fundamental property of Markov chain, for canonical Markov chain, there is stationary distribution in Step4, therefore meets α=(x ₁, x ₂... x _n) ^t, A α=α, then by when the eigenwert of matrix A is 1, characteristic of correspondence vector α, α are defined as the webpage sorting vector of craft embodiment, represent the PR value of i-th webpage;

Step5 draws the PR initial value f of each craft embodiment by α _i;

The result of Step6 to PageRank is evaluated, f _ivalue shows that more greatly this craft embodiment is more important, and the reference value treating decision-making technological problems is larger, sets a threshold value p, draws similar to technological parameter to be processed and that processing effect is comparatively stable craft embodiment, is designated as YJ=(YJ ₁, YJ ₂... YJ _s);

Step7 adopts one in conjunction with the PageRank innovatory algorithm of link analysis and craft embodiment technique content; This algorithm utilization inquiry treats that the information beyond decision-making technological problems theme improves the identification capability to optimizing group network, to reduce the generation of topic drift phenomenon; It calculates craft embodiment technique content and the correlativity treating decision-making technological problems, craft embodiment information node (b, J), and it is a tuple (L _iD), be used for preserving the craft embodiment information of user under technological problems k, calculate in conjunction with link structure, its computing formula is as follows:

${\mathrm{PR}}_q\left(p_i\right)=(1-d)\×{P^{\′}}_q\left(p_i\right)+d\underset{p_j\∈M\left(P_i\right)}{\mathrm{\Σ}}\frac{{\mathrm{PR}}_q\left(p_j\right)\×h_q(p_i,p_j)}{L\left(P_j\right)}$

P in formula ₁, p ₂, p ₃..., p _ncraft embodiment node, M (p _i) be craft embodiment p _ithe number of times be cited, L (p _j) be craft embodiment p _jquote the number of times of other examples, PR (p _j) be craft embodiment p _jpR value, d is damping factor, and being usually set to 0.85, d is ratio of damping, h _q(p _i, p _j) represent user under inquiry theme q from craft embodiment p _ijump to craft embodiment p _jpossibility.P' _q(p _i) represent when craft embodiment is not cited, jump to craft embodiment p _ipossibility.PR _q(p _j) represent craft embodiment p _itreating the PR value under decision-making technological problems inquiry theme q;

Step8 is for h _q(p _i, p _j) and P' _q(p _i) adopt inquiry theme q and craft embodiment p _irelated function R _q(p _i) calculate, show that formula is as follows:

${P^{\′}}_q\left(p_i\right)=\frac{R_q\left(p_i\right)}{\underset{k\∈W}{\mathrm{\Σ}}{R}_q\left(k\right)}{h}_q(p_i,p_j)=\frac{R_q\left(p_i\right)}{\underset{k\∈F_{P_j}}{\mathrm{\Σ}}{R}_q\left(k\right)}$

In formula, W represents the craft embodiment set optimized in group network, represent craft embodiment p _jthe example set that is cited;

Step9 calculates R _q(p _i) value: related function R _q(p _i) can be arbitrary in principle, but general value is at craft embodiment p _icraft embodiment information in inquire about the similarity degree value of theme q; $d(p_i,d_j)=\underset{q=1}{\overset{n}{\mathrm{\Σ}}}(w_q\left(p_i\right)\×w_q\left(d_j\right))/\sqrt{\left(\right(\underset{q=1}{\overset{n}{\mathrm{\Σ}}}{w}_q^2\left(p_i\right)\·(\underset{q=1}{\overset{n}{\mathrm{\Σ}}}{w}_q^2\left(d_j\right)}$ Quantitative description craft embodiment p _i, d _jbetween similarity degree.Wherein w _h(p _i) represent that q article of technological problems of i-th craft embodiment describes attribute.R _q(p _i) value is d (p _i, d _j) value;

Not only making full use of the link structure of webpage when transmitting webpage PR value, also considering the topic relativity between craft embodiment, making the transmission of PR value more accurate;

Step10 when needing decision-making until decision-making technological problems, by related function R _q(p _i) value substitution, draw YJ=(YJ ₁, YJ ₂... YJ _s) in the PR of each craft embodiment _q(p _i) value, the result of webpage sorting is evaluated, at YJ=(YJ ₁, YJ ₂... YJ _s) in again filter out PR _q(p _i) the maximum craft embodiment YJ of value _z, draw and processing effect preferably craft embodiment similar to technological parameter to be processed.

The present invention, by Graph-theoretical Approach and Algorithms for Page Ranking R. concomitans, can solve the dry cutting process Parametric optimization problem of new gear hobbing.The dry cutting processing method of gear hobbing of the present invention, gear hobbing based on a kind of uniqueness is dry cuts working process parameter optimization and realizes, this gear hobbing is dry cuts in process, catch the feature of the dry cutting process parameter of gear hobbing, carry out the Auto-grouping of the dry cutting process example set of gear hobbing, in similar group, choose the dry craft embodiment optimization group of cutting of gear hobbing, build gear hobbing dry cutting process example optimal group network, and complete the optimization of dry cutting process parameter; The highest technological parameter group of the sequence that finally obtains is utilized to carry out follow-up gear hobbing process.Algorithm speed is fast, and efficiency is high, and machining precision is high.

Accompanying drawing explanation

Fig. 1 is the schematic diagram based on the dry blanking method step of the gear hobbing of Optimizing Process Parameters in the specific embodiment of the invention;

Fig. 2 is the schematic diagram of gear hobbing dry cutting process example optimal group network building process in the specific embodiment of the invention;

Fig. 3 is the schematic diagram based on the high-speed dry cutting process parameter optimisation procedure of Algorithms for Page Ranking in the specific embodiment of the invention.

Embodiment

Thinking of the present invention is: utilize the minimal spanning tree algorithm of non-directed graph of having the right to obtain similar group of gear hobbing dry cutting process example, and then the dry craft embodiment optimization group of cutting of gear hobbing obtained based on similar features, gear hobbing dry cutting process example optimal group is utilized to build gear hobbing dry cutting process example optimal group network, use Algorithms for Page Ranking to complete the decision-making of gear hobbing dry cutting process parameter optimization, obtain optimum process example.

Below in conjunction with accompanying drawing, the invention will be further described, and gear hobbing process process parameter optimizing problem is described below:

Concrete steps of the present invention are as follows, understand with reference to Fig. 1-3:

(1) Auto-grouping of the dry cutting process example set of gear hobbing is realized; Set up existing example similar features according to gear hobbing dry cutting process example similarity feature and associate non-directed graph of having the right.By calculating the minimum spanning tree MST of non-directed graph of having the right.Crop the limit that weights in figure are greater than craft embodiment similarity threshold θ, θ generally gets 0.15, and is birdsed of the same feather flock together by craft embodiment larger for similarity, and obtain n similar group, n is positive integer.Namely by the segmentation to non-directed graph minimum spanning tree, similar group of gear hobbing dry cutting process example is generated;

(2) realize choosing based on the dry craft embodiment optimization group of cutting of gear hobbing of similar features; Find out the centre of similarity of each similar group according to the limit weights of the similarity of example, then calculate respectively and treat the limit weights of decision-making technological problems and each centre of similarity, getting minimum similar group of limit weights is technique to be processed nearest craft embodiment optimization group;

(3) structure of gear hobbing dry cutting process example optimal group network is realized; Gear hobbing dry cutting process example optimal group network is a digraph, the craft embodiment optimization group obtained in (2) is abstracted into digraph, in optimization group, each craft embodiment is mapped as the node of figure, mutual adduction relationship in group between craft embodiment is mapped as corresponding directed edge in figure, set between any two summits and have direct path, shift, with probability 1-d by red direction transfer (wherein d is for damping factor) by original blue direction with probability d at each Nodes.The random PR value (non-zero) of given each node one is as the initial value of webpage sorting;

(4) based on the gear hobbing dry cutting process parameter optimization decision-making of Algorithms for Page Ranking; First, according to the gear hobbing dry cutting process example optimal group network that (3) build, obtain adjacency matrix P, the proper vector of transition probability matrix A, A is the initial value of PR value, is namely the initial value of webpage sorting value; Then, in conjunction with treating decision-making technological problems and the similarity optimizing each craft embodiment in group network, use webpage sorting innovatory algorithm to calculate the PR value of each craft embodiment, namely what PR value was the highest is and treats the dry cutting process example of the gear hobbing of decision-making technological problems Optimum Matching.

The high-speed dry cutting process parameter apolegamy problem of hypothesis instance reasoning is designated as D _opt=(W, J, M, F).Wherein, W is the technique apolegamy problem of required decision-making.Existing craft embodiment collection J={J ₁, J ₂j _s, wherein $J_n=\left\{\right\{k_{n,1},k_{n,2}......k_{n,m_1}\},\{y_{n,1},y_{n,2},......y_{n,m_2}\},\{v_{n,1},v_{n,2},......v_{n,m_3}\left\}\right\},$ K _i,jrepresent the parameter between each example, y _i,jrepresent the attribute description of technological problems, v _i,jrepresent the processing stability evaluation attributes treating decision-making technique apolegamy example.Craft embodiment feature association is represented, M=(M (J with M ₁, J ₂), M (J ₁, J ₃) ... M (J ₁, J _s), M (J ₂, J ₃) ... M (J ₂, J _s) ... M (J _s-1, J _s)).Each high-speed dry cutting process example is mapped as proper vector v={fecture1, a w ₁; Fecture2, w ₂; Fecturen, w _ncan use $d(J_i,J_j)=\underset{h=1}{\overset{n}{\mathrm{\Σ}}}(w_h\left(J_i\right)\×w_h\left(J_j\right))/\sqrt{\left(\underset{h=1}{\overset{n}{\mathrm{\Σ}}}{w_h}^2\left(J_i\right)\right)\left(\underset{h=1}{\overset{n}{\mathrm{\Σ}}}{w_h}^2\left(J_j\right)\right)}$ Quantitative description craft embodiment J _i, J _jbetween similarity degree.Wherein w _h(J _i) represent that h article of technological problems of i-th craft embodiment describes attribute.F represents the arrangement sequence of optimal set PR value, F _max=min{F ₁, F ₂f _c.For avoid one independently craft embodiment not to be cited and the grade that produces is leaked, or in whole network one group is closely linked into the link of not going out between the craft embodiment of ring, adopts webpage sorting underground pipe net pattern, describes between example quote probability with k: e (u _i, u _j)=0 represents u _i, u _jbetween do not quote, d is damping factor.

In above-mentioned steps 1, gear hobbing dry cutting process exemplary construction building process is as follows:

Step1 according to craft embodiment storehouse, the similarity d (J of calculated examples _i, J _j), using craft embodiment as summit N=J, the similar features association between craft embodiment is as limit E={d (J _i, J _j) | J _i, J _j∈ J, i ≠ j}, by d (J _i, J _j) inverse as the weights on limit, build the association of example similar features and to have the right non-directed graph J' _d(J)=(N, E);

Step2 builds the abstract part J of craft embodiment collection ^*, wherein J ^*∈ J, j _j∈ J ^*, J _i≠ J _j, M (J _i, J _j) < θ, θ be similarity threshold, so J ^*represent similar group of a high-speed dry cutting process example.

In above-mentioned steps 2, the concrete steps that practical routine self-adaptation structure is done in gear hobbing comprise,

Step1 exists make gi=min{M (J _i, J _j)/J _i, J _j∈ J ^*.As M (g _i, W) and when getting minimum, gi is similar group of the nearest craft embodiment of W, established technology example optimal group;

Step2 adopts the prim algorithm of classical generation minimum spanning tree, solves G _d' minimum spanning tree MST;

Step3 build two tuple P=(J, M) wherein J be the set of all craft embodiments in network, M is the set on all cum rights limits in network.Arranging two new set U and T, wherein gathering U for depositing the summit in the minimum spanning tree of P, set T deposits the limit of the minimum spanning tree of P;

The initial value of Step4 order set U is U={u ₁, when supposing structure minimum spanning tree, from summit u ₁set out), the initial value of set T is T={ φ };

Step5 is from all u ∈ U, in the limit of v=J-U, choose the limit (u, v) with minimum weights, vertex v is added in set U, limit (u, v) is added in set T, so constantly repeatedly, until during U=J, minimum spanning tree structure is complete, now gathers all limits containing minimum spanning tree in T, namely obtains G _d' minimum spanning tree MST;

Step6 minimum spanning tree is divided into multiple subtree with different similarity, finds out each center obtain craft embodiment optimization group gi collection of birdsing of the same feather flock together.

In above-mentioned steps 3, as shown in Figure 2, concrete steps comprise gear hobbing dry cutting process example optimal group network building process,

Step1 builds craft embodiment information node (b, J), and it is a tuple (L _iD), be used for preserving the craft embodiment information of user under technological problems k, comprise the numbering ID of craft embodiment, b, J, k, ID are positive integer;

Step2 sets damping factor d, PR value will be calculated by iteration, calculated amount is larger, for improving the speed of retrieval, PR value should be calculated by the mode of off-line, thus first calculate PR value by being similar to conventional web sort algorithm, then on the basis of simple formula, add ratio of damping d, the general value d=0.85 of d;

The adduction relationship of Step3 factually between example, completes the weighted digraph of webpage sorting underground pipe net model;

Step4 sets threshold values z=0.15, crops the limit that limit weights are less than threshold values, completes the network struction of optimal set.

In above-mentioned steps 4, as shown in Figure 3, concrete steps comprise the dry cutting process parameter optimisation procedure of the gear hobbing based on Algorithms for Page Ranking,

Step1, according to digraph, defines adjacency matrix P, if example j quotes to example i, then and p _ij=1, on the contrary p _ij=0, p _ijfor the value in the i-th row jth row in matrix P, i, j are positive integer;

Step2 remember matrix P row and, row and.Row and: row and they sets forth craft embodiment j quoted number of times and its decision-making by other examples time quote the number of times of other craft embodiments;

Step3 Markov chain describes the state metastatic rule between each craft embodiment; Definition transition probability matrix $A=\left(a_{\mathrm{ij}}\right),a_{\mathrm{ij}}=\frac{p_{\mathrm{ij}}}{c_j},$ Wherein i, j=1,2 ... .n;

According to the fundamental property of Markov chain, for canonical Markov chain, there is stationary distribution in Step4, therefore meets α=(x ₁, x ₂... x _n) ^t, A α=α, then by when the eigenwert of matrix A is 1, characteristic of correspondence vector α, α are defined as the webpage sorting vector of craft embodiment, represent the PR value of i-th webpage;

Step5 draws the PR initial value f of each craft embodiment by α _i;

The result of Step6 to PageRank is evaluated, f _ivalue shows that more greatly this craft embodiment is more important, and the reference value treating decision-making technological problems is larger, sets a threshold value p, draws similar to technological parameter to be processed and that processing effect is comparatively stable craft embodiment, is designated as YJ=(YJ ₁, YJ ₂... YJ _s);

Step7 adopts one in conjunction with the PageRank innovatory algorithm of link analysis and craft embodiment technique content; This algorithm utilization inquiry treats that the information beyond decision-making technological problems theme improves the identification capability to optimizing group network, to reduce the generation of topic drift phenomenon; It calculates craft embodiment technique content and the correlativity treating decision-making technological problems, craft embodiment information node (b, J), and it is a tuple (L _iD), be used for preserving the craft embodiment information of user under technological problems k, calculate in conjunction with link structure, its computing formula is as follows:

${\mathrm{PR}}_q\left(p_i\right)=(1-d)\&times;{P^{\&prime;}}_q\left(p_i\right)+d\underset{p_j\&Element;M\left(P_i\right)}{\mathrm{\&Sigma;}}\frac{{\mathrm{PR}}_q\left(p_j\right)\&times;h_q(p_i,p_j)}{L\left(P_j\right)}$

P in formula ₁, p ₂, p ₃..., p _ncraft embodiment node, M (p _i) be craft embodiment p _ithe number of times be cited, L (p _j) be craft embodiment p _jquote the number of times of other examples, PR (p _j) be craft embodiment p _jpR value, d is damping factor, and being usually set to 0.85, d is ratio of damping, h _q(p _i, p _j) represent user under inquiry theme q from craft embodiment p _ijump to craft embodiment p _jpossibility.P' _q(p _i) represent when craft embodiment is not cited, jump to craft embodiment p _ipossibility.PR _q(p _j) represent craft embodiment p _itreating the PR value under decision-making technological problems inquiry theme q;

Step8 is for h _q(p _i, p _j) and P' _q(p _i) adopt inquiry theme q and craft embodiment p _irelated function R _q(p _i) calculate, show that formula is as follows:

${P^{\&prime;}}_q\left(p_i\right)=\frac{R_q\left(p_i\right)}{\underset{k\&Element;W}{\mathrm{\&Sigma;}}{R}_q\left(k\right)}{h}_q(p_i,p_j)=\frac{R_q\left(p_i\right)}{\underset{k\&Element;F_{P_j}}{\mathrm{\&Sigma;}}{R}_q\left(k\right)}$

In formula, W represents the craft embodiment set optimized in group network, represent craft embodiment p _jthe example set that is cited;

Step9 calculates R _q(p _i) value: related function R _q(p _i) can be arbitrary in principle, but general value is at craft embodiment p _icraft embodiment information in inquire about the similarity degree value of theme q; $d(p_i,d_j)=\underset{q=1}{\overset{n}{\mathrm{\&Sigma;}}}(w_q\left(p_i\right)\&times;w_q\left(d_j\right))/\sqrt{\left(\right(\underset{q=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_q^2\left(p_i\right)\&CenterDot;(\underset{q=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_q^2\left(d_j\right)}$ Quantitative description craft embodiment p _i, d _jbetween similarity degree.Wherein w _h(p _i) represent that q article of technological problems of i-th craft embodiment describes attribute.R _q(p _i) value is d (p _i, d _j) value;

Not only making full use of the link structure of webpage when transmitting webpage PR value, also considering the topic relativity between craft embodiment, making the transmission of PR value more accurate;

Step10 when needing decision-making until decision-making technological problems, by related function R _q(p _i) value substitution, draw YJ=(YJ ₁, YJ ₂... YJ _s) in the PR of each craft embodiment _q(p _i) value, the result of webpage sorting is evaluated, at YJ=(YJ ₁, YJ ₂... YJ _s) in again filter out PR _q(p _i) the maximum craft embodiment YJ of value _z, draw and processing effect preferably craft embodiment similar to technological parameter to be processed.

Claims (3)

1. the dry cutting processing method of the gear hobbing based on Optimizing Process Parameters, this method adopts numerical control gear hobbing dry cutting gear-hobbing machine to carry out gear hobbing process, it is characterized in that, in gear hobbing process process, carry out the dry optimization of cutting working process parameter of gear hobbing according to following steps, concrete steps are:

(1) Auto-grouping of gear hobbing dry cutting process example set is realized: set up existing example similar features according to gear hobbing dry cutting process example similarity feature and associate non-directed graph of having the right; By calculating the minimum spanning tree MST of non-directed graph of having the right; Crop the limit that weights in figure are greater than craft embodiment similarity threshold θ, θ generally gets 0.15, and craft embodiment similarity being greater than preset value is birdsed of the same feather flock together, and obtain n similar group, n is positive integer; Namely by the segmentation to non-directed graph minimum spanning tree, similar group of gear hobbing dry cutting process example is generated;

(2) realize choosing based on the dry craft embodiment optimization group of cutting of gear hobbing of similar features: the centre of similarity finding out each similar group according to the limit weights of the similarity of example, calculate respectively and treat the limit weights of decision-making technological problems and each centre of similarity, getting minimum similar group of limit weights is technique to be processed nearest craft embodiment optimization group;

(3) structure of gear hobbing dry cutting process example optimal group network is realized: gear hobbing dry cutting process example optimal group network is a digraph, the craft embodiment optimization group obtained in step (2) is abstracted into digraph, in optimization group, each craft embodiment is mapped as the node of figure, mutual adduction relationship in group between craft embodiment is mapped as corresponding directed edge in figure, set between any two summits and have direct path, shift by original blue direction with probability d at each Nodes, with probability 1-d by red direction transfer; Wherein d is damping factor; Random and the non-zero PR value of given each node one is as the initial value of webpage sorting;

(4) based on the gear hobbing dry cutting process parameter optimization decision-making of Algorithms for Page Ranking: first, according to the gear hobbing dry cutting process example optimal group network that step (3) builds, obtain adjacency matrix P, transition probability matrix A, the proper vector of A is the initial value of PR value, is namely the initial value of webpage sorting value; Then, in conjunction with treating decision-making technological problems and the similarity optimizing each craft embodiment in group network, use webpage sorting innovatory algorithm to calculate the PR value of each craft embodiment, namely what PR value was the highest is and treats the dry cutting process example of the gear hobbing of decision-making technological problems Optimum Matching; Then complete dry the cutting of gear hobbing according to this example to process.

2. the dry cutting processing method of a kind of gear hobbing based on Optimizing Process Parameters as claimed in claim 1, is characterized in that,

As optimization, the gear hobbing dry cutting process example optimal group network in described step (3) builds as follows:

Step1 builds craft embodiment information node (b, J), and it is a tuple (L _iD), be used for preserving the craft embodiment information of user under technological problems k, comprise the numbering ID of craft embodiment, b, J, k, ID are positive integer;

Step2 is set damping factor d, PR value and will be calculated by iteration, and calculated amount is larger, for improving the speed of retrieval, PR value is calculated by the mode of off-line, thus first calculates PR value by conventional web sort algorithm, ratio of damping d is added again, the general value d=0.85 of d on the basis of simple formula;

The adduction relationship of Step3 factually between example, completes the weighted digraph of webpage sorting underground pipe net model;

Step4 sets threshold values z=0.15, crops the limit that limit weights are less than threshold values, completes the network struction of optimal set.

3. the dry cutting processing method of a kind of gear hobbing based on Optimizing Process Parameters as claimed in claim 1, is characterized in that,

As optimization, the gear hobbing dry cutting process parameter optimisation step based on Algorithms for Page Ranking in described step (4) is as follows:

Step1, according to digraph, defines adjacency matrix P, if example j quotes to example i, then and p _ij=1, on the contrary p _ij=0, p _ijfor the value in the i-th row jth row in matrix P, i, j are positive integer;

Step2 remember matrix P row and, row and.Row and: row and they sets forth craft embodiment j quoted number of times and its decision-making by other examples time quote the number of times of other craft embodiments;

Step3 Markov chain describes the state metastatic rule between each craft embodiment; Definition transition probability matrix A=(a _ij), wherein i, j=1,2 ... .n;

According to the fundamental property of Markov chain, for canonical Markov chain, there is stationary distribution in Step4, therefore meets α=(x ₁, x ₂... x _n) ^t, A α=α, then by when the eigenwert of matrix A is 1, characteristic of correspondence vector α, α are defined as the webpage sorting vector of craft embodiment, represent the PR value of i-th webpage;

Step5 draws the PR initial value f of each craft embodiment by α _i;

The result of Step6 to PageRank is evaluated, f _ivalue shows that more greatly this craft embodiment is more important, and the reference value treating decision-making technological problems is larger, sets a threshold value p, draws similar to technological parameter to be processed and that processing effect is comparatively stable craft embodiment, is designated as YJ=(YJ ₁, YJ ₂... YJ _s);

Step7 adopts one in conjunction with the PageRank innovatory algorithm of link analysis and craft embodiment technique content; This algorithm utilization inquiry treats that the information beyond decision-making technological problems theme improves the identification capability to optimizing group network, to reduce the generation of topic drift phenomenon; It calculates craft embodiment technique content and the correlativity treating decision-making technological problems, craft embodiment information node (b, J), and it is a tuple (L _iD), be used for preserving the craft embodiment information of user under technological problems k, calculate in conjunction with link structure, its computing formula is as follows:

$P{R}_q\left(p_i\right)=(1-d)\&times;{P^{\&prime;}}_q\left(p_i\right)+d\underset{p_j\&Element;M\left(P_i\right)}{\mathrm{\&Sigma;}}\frac{{\mathrm{PR}}_q\left(p_j\right)\&times;h_q(p_i,p_j)}{L\left(P_j\right)}$

P in formula ₁, p ₂, p ₃..., p _ncraft embodiment node, M (p _i) be craft embodiment p _ithe number of times be cited, L (p _j) be craft embodiment p _jquote the number of times of other examples, PR (p _j) be craft embodiment p _jpR value, d is damping factor, and being usually set to 0.85, d is ratio of damping, h _q(p _i, p _j) represent user under inquiry theme q from craft embodiment p _ijump to craft embodiment p _jpossibility; P' _q(p _i) represent when craft embodiment is not cited, jump to craft embodiment p _ipossibility; PR _q(p _j) represent craft embodiment p _itreating the PR value under decision-making technological problems inquiry theme q;

Step8 is for h _q(p _i, p _j) and P' _q(p _i) adopt inquiry theme q and craft embodiment p _irelated function R _q(p _i) calculate, show that formula is as follows:

${P^{\&prime;}}_q\left(p_i\right)=\frac{R_q\left(p_i\right)}{\underset{k\&Element;W}{\mathrm{\&Sigma;}}{R}_q\left(k\right)}{h}_q(p_i,p_j)=\frac{R_q\left(p_i\right)}{\underset{k\&Element;F_{P_j}}{\mathrm{\&Sigma;}}{R}_q\left(k\right)}$

In formula, W represents the craft embodiment set optimized in group network, represent craft embodiment p _jthe example set that is cited;

Step9 calculates R _q(p _i) value: related function R _q(p _i) can be arbitrary in principle, but general value is at craft embodiment p _icraft embodiment information in inquire about the similarity degree value of theme q; $d(p_i,d_j)=\underset{q=1}{\overset{n}{\mathrm{\&Sigma;}}}(w_q\left(p_i\right)\&times;w_q\left(d_j\right))/\sqrt{((\underset{q=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_q^2\left(p_i\right)\&CenterDot;(\underset{q=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_q^2\left(d_j\right)}$ Quantitative description craft embodiment p _i, d _jbetween similarity degree; Wherein w _h(p _i) represent that q article of technological problems of i-th craft embodiment describes attribute.R _q(p _i) value is d (p _i, d _j) value;

Not only making full use of the link structure of webpage when transmitting webpage PR value, also considering the topic relativity between craft embodiment, making the transmission of PR value more accurate;

Step10 when needing decision-making until decision-making technological problems, by related function R _q(p _i) value substitution, draw YJ=(YJ ₁, YJ ₂... YJ _s) in the PR of each craft embodiment _q(p _i) value, the result of webpage sorting is evaluated, at YJ=(YJ ₁, YJ ₂... YJ _s) in again filter out PR _q(p _i) the maximum craft embodiment YJ of value _z, draw and processing effect preferably craft embodiment similar to technological parameter to be processed.