CN104484507A - Part remanufacturing method based on reverse engineering - Google Patents

Abstract

The invention discloses a part remanufacturing method based on reverse engineering. The part remanufacturing method based on the reverse engineering comprises the following steps: 1) obtaining a surface point cloud data model of a waste part; 2) obtaining the original computer aided design (CAD) model of the waste part; 3) matching the surface point cloud data model of the step 1) with the original CAD model of the step 2); 4) according to the matching result of the step 3), obtaining the maximum damage depth of the waste part; and 5) if the maximum damage depth obtained in the step 4) is lower than the threshold, using the maximum damage depth as the feed rate, and performing the reducing-type repair to the waste part, if the maximum damage depth obtained in the step 4) is higher than the threshold, performing the adding-type repair to the waste part.

Description

A kind of spare parts remanufacture method based on reverse-engineering

Technical field

The present invention relates to machinery and manufacture field again.

Background technology

Along with increasingly sharpening of shortage of resources and environmental problem, Rebuilding engineering receives to be paid close attention to widely.Manufacturing to make the value contained in waste and old resource obtain development and utilization to greatest extent again, alleviate the contradiction of shortage of resources and the wasting of resources, reduce a large amount of inefficacies, scrap products to the harm of environment, being the optimised form of worn-out machine tools resource and first-selected approach, is the important means economized on resources.

But the artificial participation of the existence of manufacture process again of current waste and old parts is many, and experience dependence is strong, and remediation efficiency is low, poor reliability, the problems such as repair process is irreversible.

Summary of the invention

The object of the invention is to solve in the manufacture process again of waste and old parts, repair mode and difficult parameters are with the problem of standardization and quantification.

The technical scheme adopted for realizing the object of the invention is such, and a kind of spare parts remanufacture method based on reverse-engineering, is characterized in that, comprise the following steps:

1) the surface point cloud data model of waste and old parts is obtained.Whether the mode obtaining body surface cloud data model is a lot, contact according to measuring sonde with measured surface, can be divided into contact type measurement and the large class of non-contact measurement two.Contact type measurement common equipment is three coordinate measuring machine (CMM), and non-contact measurement common equipment comprises laser scanner, structured light scanner, Industrial CT Machine etc.

2) the original cad model of described waste and old parts is obtained;

3) by step 1) surface point cloud data model and step 2) original cad model registration;

4) according to step 3) the result of registration, obtain the maximum lesion depths of described waste and old parts;

5) if step 4) the maximum lesion depths that obtains lower than threshold value, with described maximum lesion depths for the amount of feeding, formula reparation is subtracted to described waste and old parts.What deserves to be explained is, subtract formula reparation and refer to the repair mode removing material on origianl component matrix, namely parts injured surface is reprocessed, until surface damage is removed completely by machining modes such as car, milling, mills.

If step 4) the maximum lesion depths that obtains higher than threshold value, formula reparation is added to described waste and old parts.What deserves to be explained is, add the repair mode that formula reparation refers to adding material on waste and old parts matrix, the common formula renovation technique that adds has laser cladding, thermal spray, built-up welding etc., laser cladding process because of its material system be suitable for extensively, cladding layer with substrate combinating strength is high, matrix thermal deformation is little and technological process such as easily to be automated at the feature, be more and more applied to and manufacture in reparation again.

This patent proposes the waste and old spare parts remanufacture flow process framework based on reverse-engineering from the angle of system, this framework comprises to add formula reparation and subtract formula repairs two main lines, on the basis analyzing difference between waste and old component surface cloud data model and original cad model, launch to repair to waste and old parts.

Accompanying drawing explanation

Fig. 1 is the waste and old spare parts remanufacture flow process framework based on reverse-engineering;

Fig. 2 is waste and old component surface cloud data model acquisition step;

Fig. 3 is waste and old component surface cloud data model collection result;

Fig. 4 is waste and old component surface cloud data model preprocessing;

Fig. 5 is the damage boundary demarcation of waste and old component surface cloud data model;

Fig. 6 is the reconstruct of the original cad model of waste and old parts;

Fig. 7 is the extraction of waste and old parts deleted areas model;

Fig. 8 is surface point cloud data model and original cad model registration schematic diagram;

Fig. 9 is for improving ICP registration Algorithm flow process;

Figure 10 is waste and old component surface lesion depths schematic diagram;

Figure 11 is waste mold surface point cloud data collection steps;

Figure 12 is that waste mold subtracts formula reparation flow process;

Figure 13 is that traditional IC P and Revised ICP algorithm registration result contrast;

Embodiment

Below in conjunction with drawings and Examples, the invention will be further described, but should not be construed the above-mentioned subject area of the present invention and be only limitted to following embodiment.Without departing from the idea case in the present invention described above, according to ordinary skill knowledge and customary means, make various replacement and change, all should be included in protection scope of the present invention.

Embodiment 1:

Based on a spare parts remanufacture method for reverse-engineering, comprise the following steps:

1) the surface point cloud data model of waste and old parts is obtained.See in Fig. 2, figure, adopt ATOS optical three-dimensional scanning instrument to carry out data acquisition to waste and old parts, this equipment mobility is strong, and acquisition speed is fast, can obtain all kinds of component surface cloud data.In embodiment, by the surface spraying of tested waste and old parts white developer before gathering, to strengthen effect.The result gathered is as Fig. 3.

In embodiment, preferably, pre-service (filtration, stress release treatment, data compaction etc.) is carried out to the collection result shown in Fig. 3, has obtained waste and old component surface cloud data model as shown in Figure 4, the handling object of subsequent step can be it can be used as.

2) the original cad model (unworn) of described waste and old parts is obtained.In the present embodiment, the original cad model of described waste and old parts can directly obtain, and namely manufacturer or designer save this original cad model.When original cad model is lost, see the disclosed method obtaining original cad model of embodiment 2.

3) see Fig. 7, by step 1) surface point cloud data model and step 2) original cad model registration.That is, by two Unified Models in (being placed on) same coordinate system, the difference of two models can be compared, know the described lack part of surface point cloud data model and the maximum lesion depths of waste and old parts.

4) according to step 3) the result of registration, obtain the maximum lesion depths of described waste and old parts.In embodiment, described waste and old parts are parts of wearing and tearing, have some places abration position.Figure 10 shows these parts wherein place's abration position, wherein original contour of represented by dotted arrows parts, and circle points represents the point on original contour, and solid line represents the profile after part injury, and spider represents the point on damage profile, and its wearing depth is H.In these parts, the wearing depth of the darkest place's abration position that weares and teares is the maximum lesion depths hereafter mentioned.

5) if step 4) the maximum lesion depths that obtains lower than threshold value (being less than 5% of overall dimensions), with described maximum lesion depths for the amount of feeding, formula reparation is subtracted to described waste and old parts.By waste and old parts as blank, with the given amount of feeding, cut is carried out to it.

If step 4) the maximum lesion depths that obtains higher than threshold value (being greater than 5% of overall dimensions), formula reparation is added to described waste and old parts.Namely by the method at abration position adding material, described waste and old parts are repaired.

Embodiment 2

In the repair process of waste and old parts, the situation that the original cad file of parts is lost may be run into, now need to reconstruct its original cad model according to the residual, information in the surperficial cloud data collected.Different from traditional reverse modeling, when reconstructing the original cad model of waste and old parts, owing to comprising the damage field of parts in the cloud data that collects, the point cloud in this part cloud data and other intact region is discontinuous, the initial surface pattern of parts can not be reflected, therefore the cloud data using damage field should be avoided when Reverse reconstruction.

The key step of the present embodiment, with embodiment 1, just, needs by following steps obtaining step 2) described in original cad model.

In embodiment, by calculating the Gaussian curvature of each point in cloud data, extract the point of cloud data mean curvature sudden change, the damage field boundary line approximate according to these some structures one, and check that whether the feature outlines reconstructed in reverse modeling is through damage field, guarantees modeling accuracy.Its key step is:

1) Point cloud curvature is estimated

The method of Point cloud curvature estimation is a lot, conventional method is as parabola fitting process, 3DShepard Surface Fitting, Gauss – Bonnet method etc., due to parabola fitting process, to the process of Noise cloud data, comparatively additive method is more accurate, therefore adopting parabola fitting process to estimate a curvature at cloud grid vertex place, the method approaches the point in set point and neighborhood thereof with the analytic surface of a second order, for the surface of second order expression formula of matching by shown in formula (1).

z＝f(x,y)＝a ₀+a ₁x+a ₂y+a ₃xy+a ₄x ²+a ₅y ²(1)

To a certain data point p in cloud data _i, get the k-neighborhood composition partial points cloud of this point, to point (x all in this partial points cloud _j, y _j, z _j), do least square fitting by formula (2), namely solve:

$\min \underset{j}{\mathrm{\Σ}}{(a_0+a_1{x}_j+a_2{y}_i+a_3{x}_j{y}_i+a_4{x_j}^2+a_5{y_j}^2-z_j)}^2,j\∈(0,k)---\left(2\right)$

After trying to achieve surface equation coefficient, surface equation (1) is rewritten as parametric equation form, as shown in formula (3).

$r(x,y)=\left\{\begin{array}{c}X(x,y)=x\\ Y(x,y)=y\\ z(x,y)=a_0+a_{1}x+a_{2}y{+a}_3\mathrm{xy}+a_4{x}^2+a_5{y}^2\end{array}\right.---\left(2\right)$

Obtain the partial differential of r (x, y) to x, y, xx, yy, xy respectively, be designated as r _x, r _y, r _xx, r _yy, r _xy, the unit normal vector of curved surface is then first fundamental form of surface parameter E=r _xr _x, F=r _xr _y, G=r _yr _y, second fundamental form of a surface parameter L=r _xxn, M=r _xyn, N=r _yyn.Substitute into the computing formula (4) of curved surface Gaussian curvature, the Gaussian curvature at each point place can be obtained.

$K=\frac{\mathrm{LN}-M^2}{\mathrm{EG}-F^2}---\left(4\right)$

2) damage field border is divided

According to each point Gaussian curvature value K that upper step calculates _i, by the threshold k of setting _e, namely work as K _i> K _etime, then judge that this point is curvature mutation point, finally simulate damage field boundary line by curvature mutation point, cloud data is divided into damage field and intact region, as shown in Figure 5, and by the cloud data in damage field border stored in set N.

3) original cad model is reconstructed

The reconstruct of the original cad model of waste and old parts is realized by the modeling approach of " point-line-face-body ", as shown in Figure 6, first on point cloud model, construct one group of cross section contour, by checking the point whether each cross section contour having and gathers in N, find out and delete through a cross section contour for cloud damage field, finally obtaining the feature outlines in the intact region of a cloud.The operations such as feature outlines is stretched, scan, covering and cutting obtain the surface model of parts, finally thicken surface model or namely hypostazation obtains original cad model.

Embodiment 3

The present embodiment key step is with embodiment 1, and further, open one is applicable to step 3) improvement method for registering.

What deserves to be explained is that the surface point cloud data model of waste and old parts and the registration of original cad model are the keys obtaining maximum lesion depths, registration between two models needs through pre-registration and smart registration two steps usually, pre-registration is that two models are adjusted to correct position substantially, for accuracy registration provides good initial value, improve the efficiency of accuracy registration, the method of pre-registration mainly contains principal component analysis (PCA), principal axes algorithm, 3 alignment methods etc.Accuracy registration is the position correcting two models on the basis of pre-registration further, makes difference between the two minimum.With iterative closest point (Iterative Clostest Point in accuracy registration algorithm, ICP) algorithm is the most ripe, the essence of this algorithm is the Optimum Matching method based on least square method, the iterative process that algorithm repeats " finding corresponding point---optimum rigid body translation between corresponding point ", until meet the convergence criterion of setting, its transformation relation formula and convergence criterion, as formula (5), shown in formula (6).

Q _j＝RP _i+T (5)

ε＝Σ||q _j-(Rp _i+T)|| ²→min (6)

Wherein P _iand Q _ibe 2 model data point sets, R is rotational transformation matrix, and T is translation transformation matrix, p _ifor model P _iin point, q _ifor model Q _iin point, meet convergence criterion when ε is minimum.

But manufacturing in practical application again, because waste and old component surface exists local damage, surface point cloud data model has certain difference compared with original cad model, according to traditional IC P algorithm, best-fit is implemented to two models, points all on model all will participate in registration computing, then the error of damage field can be homogenized, thus can not get maximum depth value accurately.Now by Fig. 8, this problem is described, Fig. 8 (a) is the situation before the original cad model of certain waste and old parts (thick dashed line) and surface point cloud data model (heavy line) registration, duty status according to these parts is learnt, parts damage concentration is in upper end (in encircled), and other positions are not then damaged or damaged minimum; Fig. 8 (b) is for adopting the result of traditional IC P algorithm registration, and the error distribution due to damage field gives damage little region, makes originally not damage or damage minimum region to have occurred error; Fig. 8 (c) is then desirable registration effect.

The resolving ideas of the present embodiment is for this reason: in traditional IC P method for registering, increase the screening process to two model corresponding point, by checking the distance of corresponding point and direction vector angle whether in the scope of setting, judge that whether corresponding point are the point of damage field, if be judged as the point of damage field, then reject these group corresponding point and regenerate corresponding point collection, carry out registration computing again, the registration Algorithm process flow diagram after improvement as shown in Figure 9.

Its key step is:

1) pre-registration: adopt 3 methods of aliging, with original cad model point set Q for reference data, effects on surface cloud data model point set P carries out pre-registration initial transformation, the surface point cloud data model point set P after original cad model point set Q and in advance registration ₀can be expressed as:

P ₀＝{p _i0|p _i0∈R ³，i＝1，2，…n}，Q ₀＝{q _j0|q _j0∈R ³，j＝1，2，…m}。

2) corresponding point are found: effects on surface cloud data model point set P _kin any point p _ik, find p _ikto the point that master pattern point set Q middle distance is nearest, with p in note master pattern point set Q _iknearest point is q _ik, composition corresponding point set Q _k={ q _ik| q _ik∈ R ³, i=1,2 ... n}, distance computing formula is d _ik=|| p _ik-q _ik|| → min, k are iterations.

3) transformation matrix is solved: to finding the corresponding point set P obtained _kwith Q _k, adopt optimum solution analysis method to calculate Σ || R _kp _ik+ T _k-q _ik|| ²→ min, tries to achieve rotational transformation matrix R during kth time iteration _kwith translation transformation matrix T _k.

4) relative position between Renewal model: with the transformation matrix obtained in step 3, effects on surface cloud data model point set P carries out rotation and translation conversion, obtains the position that surface point cloud data model is new, i.e. P _k+1=R _kp _k+ T _k.

5) corresponding point distance and orientation consistency is checked: check whether corresponding point spacing is less than the threshold value d of setting _e, i.e. d _i ^k< d _e, and calculate the direction vector of each group of corresponding point, check whether its angle theta is less than the threshold value θ of setting _e, i.e. θ < θ _e.

6) damage field corresponding point are rejected: if corresponding point spacing d _ik> d _eor θ _ik> θ _e, then judge that corresponding point are damage field point, damage field point rejected and generates new surface point cloud data model point set P ' and original cad model point set Q '.

7) iteration ends judges: between corresponding point, mean distance is less than given threshold value, namely wherein then iteration ends.

Embodiment 4

The present embodiment applies the method described in embodiment 1,2 or 3, carries out repair process to the hang-wheel on Gear Hobber frame of wearing and tearing.

Hang-wheel on Gear Hobber frame is one of key components and parts of hobbing machine, for installing equalizing gear, realizes straight-tooth processing driving-chain and helical teeth processes switching between driving-chain.After change gear plate wearing and tearing, will directly affect the engagement between differential hanging gear, the helical gear machining precision of final impact.Fig. 2 (a) is for dismantle from certain waste and old hobbing machine the change gear plate obtained, and noticeable wear appears in the deep-slotted chip breaker region of change gear plate, and other positions in order, now manufacture reparation to its enforcement again.

After the disassembly, cleaning of change gear plate parts, at its surface spraying white developer to strengthen scanning effect, utilize ATOS optical three-dimensional scanning instrument collection surface cloud data, gatherer process as shown in Figure 2.

Because the original cad model of change gear plate is lost, need according to its original cad model of scanning Surface Reconstruction from Data Cloud, restructuring procedure is shown in Fig. 5 and Fig. 6.Learn according to registration result analysis, change gear plate damage field is concentrated and wearing and tearing are darker, be greater than 5% of overall dimensions, therefore select to add formula recovery scenario, add formula repair process as shown in Fig. 7 (lowermost part), original for change gear plate cad model and surface point cloud data model are carried out registration by the method for registering described in embodiment 3, then boolean operation is implemented to two models after registration, obtain the model of defect.Finally, defect model is cut into slices, generate defect model cross section profile data, simulated laser deposition path.

Embodiment 5

The present embodiment applies the method described in embodiment 1,2 or 3, carries out repair process to the hammering mould of wearing and tearing.

In use, mold cavity bears huge shock load to hammering mould, and die cavity local produces plastic yield, and meanwhile, blank metal and mold cavity surface produce severe friction, and mold cavity surface there will be wearing and tearing and even peels off.Again reparation is manufactured to inefficacy hammering mould, can die life be extended, reduce production cost.Figure 11 (a) is depicted as the turbine blade hammering mould scrapped, and existing development it manufactures reparation again.

The gatherer process of die surface cloud data as shown in figure 11, because mould upper surface is groundwork face, therefore only carries out data acquisition to mould upper surface, improves remediation efficiency.

Because mould damage field is large, wear and tear more shallow, be less than 5% of overall dimensions, and process redundancy is sufficient, therefore select to subtract formula recovery scenario, by mould upper surface according to original design shape again milling shape.Subtract formula repair process as shown in figure 12, the method for registering that original for mould cad model and surface point cloud data model describe by embodiment 3 is carried out registration, Figure 13 is the Comparative result of traditionally ICP algorithm registration and Revised ICP algorithm registration, by same point on Measurement and analysis point cloud model, wear extent respectively under these two kinds of method for registering, as shown in table 1.

The measurement of table 1 failure site wear extent

The result of table 1 shows, when using traditional IC P algorithm registration, the wear extent of plane domain is at 0.49 ~ 0.82mm, when using Revised ICP algorithm registration, the wear extent of plane domain is about 0.02mm, and this mold plane region exists wearing and tearing hardly, therefore it is more accurate to improve ICP registration result.Measure the distance between two model corresponding point after registration, show that between corresponding point, ultimate range is 4.53mm, namely the greatest wear depth H of mould is 4.53mm, then subtract formula Repair gene surplus and get 4.53mm.Finally, by original cad model and subtract formula Repair gene margin value, generate mechanical process tool route.

Claims (1)

1., based on a spare parts remanufacture method for reverse-engineering, it is characterized in that, comprise the following steps:

1) the surface point cloud data model of waste and old parts is obtained;

2) the original cad model of described waste and old parts is obtained;

3) by step 1) surface point cloud data model and step 2) original cad model registration;

4) according to step 3) the result of registration, obtain the maximum lesion depths of described waste and old parts;

5) if step 4) the maximum lesion depths that obtains lower than threshold value, with described maximum lesion depths for the amount of feeding, formula reparation is subtracted to described waste and old parts;

If step 4) the maximum lesion depths that obtains higher than threshold value, formula reparation is added to described waste and old parts.