Process recommendation method for similar castings

Abstract

The invention relates to a casting process recommendation method for similar castings, and the method specifically comprises the steps: respectively calculating D2 shape features, modulus features, wall thickness features and envelope size features of a to-be-matched casting model and a model in a model library, and respectively calculating Euclidean distance values of the to-be-matched casting model and the features of the model in the model library, weighting fusion similarity according to the Euclidean distance value, wherein the model with the maximum value of the weighting fusion similarity is the model most similar to the casting model to be matched; whether the to-be-matched casting model and the most similar model are the same in material and size or not is judged, and if yes, casting technological parameters of the most similar model are directly extracted; and if the material and / or size of the casting model to be matched is different from the material and / or size of the most similar model, directly calculating the casting process parameters of the casting model to be matched so as to finish the process recommendation of similar castings, thereby solving the problems of low casting development efficiency, low quality and low accuracy.

Description

Technical Field

[0001] The invention belongs to the field of casting product process design, and in particular relates to a casting process recommendation method for similar castings. Background Art

[0002] The manufacturing process of complex castings for major equipment in industries such as aviation, aerospace, automobiles, rail transit, and engineering machinery faces common problems such as "out-of-tolerance critical quality points and large quality fluctuations."

[0003] In the past, foundries primarily adopted a trial-and-error approach during their R&D process. This meant designing new casting processes typically involved "process design - trial pouring - process modification - re-pouring." This process design approach relied heavily on the experience of process personnel and required repeated iterations, which prolonged the product development cycle. Since the 20th century, computer technology has been more widely used in the field of casting process design. Most foundries employ a method that begins with casting process design using 3D CAD modeling software. Then, using computer simulation software, they simulate the casting process to predict potential defects in the castings. Based on this defect information, the casting process was modified and adjusted, and the process was re-simulated. After repeated iterations, the casting process was finally finalized and verified through actual pouring. Compared to traditional casting process design methods, this method eliminates the trial pouring process, shortening casting process design time and reducing costs.

[0004] However, this casting process design method still has certain problems. For example, the initial casting process design and the analysis of simulation results are still highly dependent on the experience of the casting process personnel. This process is time-consuming and labor-intensive, and increases the casting development and production cycle. Most large parts produced by castings have similar structures in some parts, or even in most parts. Therefore, when designing the process for new castings, it is necessary to refer to previously produced castings for process reuse. Manual comparison and searching in a large number of model libraries often greatly reduces work efficiency, and the accuracy of manual search is also a major problem.

[0005] The prior art CN110909697A proposes a method for extracting and measuring the three-dimensional features of castings and measuring their similarity in combination with process parameters. This method realizes the step-by-step screening of similar castings based on process parameter features and D2 shape features, but only provides a method for retrieving similar castings. This method still has some shortcomings: the process features are first used for primary screening, and then the D2 features are used for secondary screening, resulting in a weak coupling between the casting shape features and the process features. After obtaining the similar casting models, the designer still needs to estimate the process dimensions of the casting to be matched based on the process of the similar castings, and finally manually complete the three-dimensional modeling of the casting process to be matched. Summary of the Invention

[0006] In response to the above-mentioned defects or improvement needs of the prior art, the present invention provides a process recommendation method for similar castings, which aims to make full use of 3D CAD modeling technology and 3D model search technology to design a casting process recommendation system and realize process reuse, thereby solving the technical problems of low casting development efficiency and quality, low accuracy, and high production cost in the prior art.

[0007] To achieve the above object, according to one aspect of the present invention, a process recommendation method for similar castings is provided, comprising the following steps:

[0008] (1) Calculate the D2 shape feature, modulus feature Mod, wall thickness feature Thk and envelope size feature Env of the casting model to be matched and the model in the model library respectively;

[0009] (2) Calculate the Euclidean distance d between the D2 shape features of the casting model to be matched and the model in the model library respectively D2 , the Euclidean distance value d of the modulus feature Mod , the Euclidean distance value d of the wall thickness feature Thk , the Euclidean distance value d of the envelope size feature Env ;

[0010] (3) Calculate the corresponding feature similarity Sim according to each Euclidean distance value i , and then according to each of the feature similarity Sim i Calculating the weighted fusion similarity Sim between the casting model to be matched and the models in the model library;

[0011] (4) sorting the models in the model library according to the value of the weighted fusion similarity Sim, wherein the model with the largest value of the weighted fusion similarity Sim is the model most similar to the casting model to be matched, thereby completing the similarity measurement of the casting;

[0012] (5) Determine whether the material and size of the to-be-matched casting model are the same as those of the most similar model. If they are the same, directly extract the pouring and rigging process parameters of the most similar model, and input the pouring and rigging process parameters into the parametric modeling module to complete the process recommendation. If the material and / or size of the to-be-matched casting model are different from those of the most similar model, directly calculate the pouring and rigging process parameters of the to-be-matched casting model, and input the pouring and rigging process parameters into the parametric modeling module to complete the process recommendation for similar castings.

[0013] Preferably, the feature similarity Sim i satisfy, i is selected from D2, Mod, Thk or Env; the weighted fusion similarity Sim satisfies, Sim=ω1Sim D2 +ω2SimMod +ω3Sim Thk +ω4Sim Env , where w1, w2, w3, and w4 are the preset weights of the feature similarity of the D2 shape feature, the feature similarity of the modulus feature, the feature similarity of the wall thickness feature, and the feature similarity of the envelope size feature, respectively, and ω1+ω2+ω3+ω4=1.

[0014] Preferably, in step (1), the D2 shape feature is calculated using a D2 shape distribution algorithm.

[0015] Preferably, in step (1), the modulus characteristic Mod satisfies: Where V c is the volume of the model, S1 is the heat transfer surface area of the model; the wall thickness characteristic Thk is the average wall thickness of the model; the envelope size characteristic Env is [L / H, W / H], where L, W, H are the length, width, and height of the model.

[0016] Preferably, step (2) is specifically as follows: extracting the feature vectors of the D2 shape feature, modulus feature Mod, wall thickness feature Thk and envelope size feature Env of the casting model to be matched and the model in the model library respectively, and the feature vector of the casting model to be matched is recorded as H i The feature vector of the model in the model library is recorded as I i , where i is selected from D2, Mod, Thk or Env; the characteristic vector of the casting model to be matched and the characteristic vector H of the model in the corresponding model library i and I i The Euclidean distance between Calculate d D2 d Mod d Thk d Env ; where n is the dimension of the feature vector.

[0017] Preferably, the pouring and rigging process parameters include the cross-sectional area of the ingrate, the equivalent diameter of the riser, and the relative coordinates of the ingrate and the riser relative to the centroid of the most similar model.

[0018] Preferably, in step (5), when the size and / or material of the casting model to be matched is inconsistent with that of the most similar model, the cross-sectional area S of the ingrowth of the casting model to be matched is 内 satisfy: Where G L is the total weight of the molten metal flowing through the ingrown section; ρ L is the density of the molten metal; μ is the flow loss coefficient; t is the pouring time; g is the acceleration of gravity; h Pis the pressure height at the ingrate; where t satisfies Where V C is the casting volume.

[0019] Preferably, in step (5), when the size and / or material of the casting model to be matched are inconsistent with those of the most similar model, the equivalent diameter d of the riser of the casting model to be matched satisfies:

[0020]

[0021] Where ε is the shrinkage rate of the metal solidification; M C is the casting modulus; V C is the volume of the casting; where M C satisfy: Among them, S1 is the heat transfer surface area of the casting.

[0022] Preferably, in step (5), when the size of the casting model to be matched is inconsistent with that of the most similar model, the coordinates of the ingode and riser of the casting model to be matched relative to the center of mass of the casting model to be matched are calculated respectively according to the coordinates of the ingode and riser of the most similar model relative to the center of mass of the most similar model; the coordinates of the ingode of the most similar model relative to the center of mass of the most similar model are recorded as P1(P 1x , P 1y , P 1z ), the coordinates of the riser of the most similar model relative to the center of mass of the most similar model are P2(P 2x , P 2y , P 2z ), the coordinates of the center of mass of the casting model to be matched are O(O x , O y , O z ), the length, width and height of the casting model to be matched are X, Y and Z respectively;

[0023] The coordinate P1′ of the ingrow of the casting model to be matched relative to the center of mass of the casting model to be matched is:

[0024]

[0025] The coordinate P2′ of the riser of the casting model to be matched relative to the center of mass of the casting model to be matched is:

[0026]

[0027] Preferably, the preset weight w1 of the Euclidean distance value of the D2 shape feature is 0.4~0.6, the preset weight w2 of the Euclidean distance value of the modulus feature is 0.2~0.3, the preset weight w3 of the Euclidean distance value of the wall thickness feature is 0.1~0.2; the preset weight w4 of the Euclidean distance value of the envelope size feature is 0.1~0.2.

[0028] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0029] 1. The present invention selects D2 shape feature, modulus feature Mod, wall thickness feature Thk and envelope size Env as the shape and process features of the casting model, and uses a weighted fusion similarity retrieval algorithm to achieve the coupling of shape and process features, so that the retrieval accuracy is high and the retrieval results are more meaningful.

[0030] 2. The present invention realizes the adaptation of process parameters of castings of different sizes and materials, can recommend process parameters of castings to be matched based on the processes of similar castings, and can automatically complete the three-dimensional modeling of the process through parametric modeling on the UG platform based on the recommended process parameters, thereby reducing the workload of designers and improving the efficiency of the casting process design process. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a flowchart of the process recommendation algorithm of Example 1 of the present invention.

[0032] Figure 2 This is a schematic flow chart of the process recommendation system of Example 1 of the present invention.

[0033] Figure 3 1 is a schematic diagram of similar casting search results for the example box casting according to embodiment 1 of the present invention.

[0034] Figure 4 It is a schematic diagram of the result of process recommendation based on the most similar casting in Example 1 of the present invention. DETAILED DESCRIPTION

[0035] In order to make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0036] The embodiment of the present invention provides a process recommendation method for similar castings, such as Figure 1 As shown, the following steps are included:

[0037] 1. Calculate the D2 shape features of the casting model to be matched and the models in the model library.

[0038] (1) For each model, read its stl file. The X coordinates of all vertices of the triangle in the stl file are in the list [X1, X2, X3…X n ], the vertex Y coordinate is in the list [Y1,Y2,Y3…Y n ], the Z coordinates of the vertices are in the list [Z1, Z2, Z3…Z n ], thereby obtaining the three side lengths a, b, and c of the triangle; then the area of all triangles is calculated according to the following formula, namely Heron's formula, and stored in the list S[S1, S2, S3…S n ]middle,

[0039] (2) Based on the area list S, calculate the cumulative area list T[T1, T2, T3…T n ], where the i-th item of T refers to the sum of the first i items in table S, T n is the sum of the areas of all triangles.

[0040] (3) Generate a random number M between 0 and Tn, use binary search to find the i-th item in table T, so that T i-1 <M<T i , then the i-th triangle is the random patch found; randomly select a point on the random patch, let E, F, G be the coordinates of the three vertices of the triangle, and point P1 be the randomly sampled point on this triangle, then the coordinate calculation formula of P1 is: Where r1 and r2 are random numbers on [0,1], so that uniformly distributed random points are obtained.

[0041] (4) Repeat (1) to (3) to obtain P2 and calculate the distance between P1 and P2.

[0042] (5) Repeat (1) to (4) n times, where n is preferably 100,000, to obtain n distance values, forming a distance array D[D1, D2, ..., D n ], find the maximum and minimum values D in D max 、D min , get the normalized distance D1[D11,D12,…,D1 n ], in array D1 Both are between [0,1].

[0043] (6) Divide the interval [0,1] into K bins, preferably 128. Divide each item in the D1 array into each bin according to its value. Count the number of values in each bin to obtain a frequency distribution histogram, where K is its dimension.

[0044] (7) Divide the value in each box by n to obtain a normalized frequency distribution histogram, and form the dimension values of the frequency histogram into a K-dimensional vector, denoted by SD[SD1,SD2,…,SD K ], which is the D2 shape feature.

[0045] 2. Calculate the modulus feature Mod, wall thickness feature Thk and envelope size feature Env of the casting model to be matched and the model in the model library;

[0046] (1) Calculation of modulus characteristics: The definition of modulus is volume divided by surface area. The result of dividing the volume of each model by the surface area is used as the modulus characteristic, which is a one-dimensional vector; that is, Among them, V c is the volume of the casting, S1 is the heat transfer surface area of the casting; the modulus characteristics of the casting model to be matched and the modulus characteristics of the model in the model library are calculated according to the above formula.

[0047] (2) Calculation of wall thickness characteristics: The average wall thickness of each model can be used as the wall thickness characteristic, which is a one-dimensional vector. The wall thickness characteristics of the casting model to be matched and the wall thickness characteristics of the models in the model library are obtained respectively.

[0048] (3) Calculation of envelope size: Assuming that the length, width and height of the model envelope are L, W and H, [L / H, W / H] is used as the envelope size feature of the model, which is a two-dimensional vector.

[0049] 3. Calculate the Euclidean distance d between the D2 shape features of the casting model to be matched and the model in the model library respectively D2 , the Euclidean distance value d of the modulus feature Mod , the Euclidean distance value d of the wall thickness feature Thk , the Euclidean distance value d of the envelope size feature Env ;

[0050] Extract D2 shape feature, modulus feature Mod, wall thickness feature Thk and envelope size feature Env to obtain four vectors of different dimensions. The feature vector of the casting model to be matched and the feature vector of the model in the corresponding model library are two n-dimensional vectors H. i and I i The Euclidean distance between According to this formula, we can calculate d D2 d Mod d Thk d Env .

[0051] 4. Calculate the corresponding similarity Sim based on the Euclidean distance value i ,in i is selected from D2, Mod, Thk or Env; then the weighted fusion similarity Sim between the casting model to be matched and the model in the model library is calculated,

[0052] Sim=ω1Sim D2 +ω2Sim Mod +ω3Sim Thk +ω4Sim Env ;

[0053] Among them, ω1+ω2+ω3+ω4=1; take ω1=0.6, ω2=0.2, ω3=0.1, and ω4=0.1.

[0054] The models in the model library are sorted according to the value of the weighted fusion similarity Sim, wherein the model with the largest value of the weighted fusion similarity Sim is the model most similar to the casting model to be matched, thereby completing the similarity measurement of the casting. The top three models with the highest similarity are used as the search results for reference.

[0055] (5) Extracting the pouring and rigging process parameters of the most similar model, wherein the pouring and rigging process parameters of the most similar model include the cross-sectional area of the ingode, the equivalent diameter of the riser, and the relative coordinates of the ingode and the riser relative to the center of mass of the casting model; calculating the process parameters of the casting model to be matched based on the pouring and rigging process parameters of the most similar model, thereby completing the process recommendation of similar castings.

[0056] (1) Calculation of process parameters of the model to be matched:

[0057] The cross-sectional area S of the inner runner of the casting model to be matched 内 satisfy: Where G L is the total weight of the molten metal flowing through the ingrown section; ρ L is the density of the molten metal; μ is the flow loss coefficient; t is the pouring time; g is the acceleration of gravity; h P is the pressure height at the ingrate; where t satisfies Where V C is the casting volume.

[0058] (2) Calculate the equivalent diameter d of the riser of the casting model to be matched, Calculate the equivalent diameter of the riser of the model to be matched, where Where V R is the volume of the riser; ε is the shrinkage rate of the metal solidification; M C is the casting modulus; A R is the surface area of the fed casting; V C is the volume of the casting; where M C satisfy: Vc is the volume of the casting, and S1 is the heat transfer surface area of the casting.

[0059] According to the relative coordinates of the runner and riser of the most similar model relative to the mass center of the most similar model and the length, width and height of the casting model to be matched, the relative coordinates of the runner and riser of the casting model to be matched relative to the mass center of the casting model to be matched are calculated; the coordinates of the runner of the most similar model are recorded as P1(P 1x , P 1y , P 1z ), the coordinates of the riser of the most similar model are recorded as P2(P 2x , P 2y , P 2z ); The coordinates of the center of mass of the casting model to be matched are O(O x , O y , O z ), the length, width and height of the casting model to be matched are X, Y and Z respectively;

[0060] The relative coordinates of the runner of the to-be-matched casting model relative to the center of mass of the to-be-matched casting model are:

[0061]

[0062] The relative coordinates of the riser of the to-be-matched casting model relative to the center of mass of the to-be-matched casting model are:

[0063]

[0064] Using this relative coordinate combined with the outline dimensions of the new casting, the coordinates of the connection point between the new ingrown and the new casting can be calculated, and the same applies to the placement of the riser.

[0065] (6) After obtaining the size and position of the pouring and rising system of the new casting through the above calculation, the cross-sectional area of the runner and the size information of the equivalent diameter of the riser of the model to be matched, as well as the coordinate information of the runner and riser relative to the center of mass, are input into the parametric modeling module of the NX platform. Combined with the secondary development technology based on UG, parametric modeling is realized on the UG three-dimensional modeling software platform, and the process conversion algorithm is embedded in the parametric modeling to realize the intelligent design process of "inputting a new casting model that requires process design - retrieving similar castings - recommending process automated three-dimensional modeling".

[0066] like Figure 3 As shown in the figure, the similar casting search results of the example box casting are shown, and the first three most similar castings are displayed. These three search results can be used as reference. Then the casting model with the highest similarity of 0.9679 is selected as the basis for process recommendation, such as Figure 4The figure shows the process recommendation based on the most similar casting, which shows that this method can effectively retrieve similar casting models, make process recommendations based on their processes, and automatically generate process three-dimensional modeling.

[0067] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A process recommendation method for similar castings, characterized in that: The following steps are involved: (1) Calculate the D2 shape feature, modulus feature Mod, wall thickness feature Thk and envelope size feature Env of the casting model to be matched and the model in the model library respectively; (2) Calculate the Euclidean distance d between the D2 shape features of the casting model to be matched and the model in the model library respectively D2 , the Euclidean distance value d of the modulus feature Mod , the Euclidean distance value d of the wall thickness feature Thk , the Euclidean distance value d of the envelope size feature Env ; (3) Calculate the corresponding feature similarity Sim according to each Euclidean distance value i , and then according to each of the feature similarity Sim i Calculating the weighted fusion similarity Sim between the casting model to be matched and the models in the model library; (4) sorting the models in the model library according to the value of the weighted fusion similarity Sim, wherein the model with the largest value of the weighted fusion similarity Sim is the model most similar to the casting model to be matched, thereby completing the similarity measurement of the casting; (5) Determine whether the material and size of the to-be-matched casting model are the same as those of the most similar model. If they are the same, directly extract the pouring and rigging process parameters of the most similar model, and input the pouring and rigging process parameters into the parametric modeling module to complete the process recommendation. If the material and / or size of the to-be-matched casting model are different from those of the most similar model, directly calculate the pouring and rigging process parameters of the to-be-matched casting model, and input the pouring and rigging process parameters into the parametric modeling module to complete the process recommendation for similar castings.

2. A process recommendation method for similar castings according to claim 1, characterized in that: The feature similarity Sim i satisfy, i is selected from D2, Mod, Thk or Env; the weighted fusion similarity Sim satisfies, Sim=ω1Sim D2 +ω2Sim Mod +ω3Sim Thk +ω4Sim Env , where w1, w2, w3, and w4 are the preset weights of the feature similarity of the D2 shape feature, the feature similarity of the modulus feature, the feature similarity of the wall thickness feature, and the feature similarity of the envelope size feature, respectively, and ω1+ω2+ω3+ω4=1.

3. A process recommendation method for similar castings according to claim 1, characterized in that: In step (1), the D2 shape feature is calculated using a D2 shape distribution algorithm.

4. A process recommendation method for similar castings according to claim 1, characterized in that: In step (1), the modulus characteristic Mod satisfies: Where V c is the volume of the model, S1 is the heat transfer surface area of the model; the wall thickness characteristic Thk is the average wall thickness of the model; the envelope size characteristic Env is [L / H, W / H], where L, W, H are the length, width, and height of the model.

5. A process recommendation method for similar castings according to claim 1, characterized in that: Step (2) is specifically as follows: The feature vectors of the D2 shape feature, modulus feature Mod, wall thickness feature Thk and envelope size feature Env of the casting model to be matched and the model in the model library are extracted respectively. The feature vector of the casting model to be matched is recorded as H i The feature vector of the model in the model library is recorded as I i , where i is selected from D2, Mod, Thk or Env; the characteristic vector of the casting model to be matched and the characteristic vector H of the model in the corresponding model library i and I i The Euclidean distance between Calculate d D2 d Mod d Thk d Env ; where n is the dimension of the feature vector.

6. A process recommendation method for similar castings according to claim 1, characterized in that: The pouring and rigging process parameters include the cross-sectional area of the ingrate, the equivalent diameter of the riser, and the relative coordinates of the ingrate and the riser relative to the centroid of the most similar model.

7. A process recommendation method for similar castings according to claim 6, characterized in that: In step (5), when the size and / or material of the casting model to be matched is inconsistent with that of the most similar model, the cross-sectional area S of the ingrowth of the casting model to be matched is 内 satisfy: Where G L is the total weight of the molten metal flowing through the ingrown section; ρ L is the density of the molten metal; μ is the flow loss coefficient; t is the pouring time; g is the acceleration of gravity; h P is the pressure height at the ingrate; where t satisfies Where V C is the casting volume.

8. A process recommendation method for similar castings according to claim 6, characterized in that: In step (5), when the size and / or material of the casting model to be matched are inconsistent with the most similar model, the equivalent diameter d of the riser of the casting model to be matched satisfies: Where ε is the shrinkage rate of the metal solidification; M C is the casting modulus; V C is the volume of the casting; where M C satisfy: Among them, S1 is the heat transfer surface area of the casting.

9. A process recommendation method for similar castings according to claim 6, characterized in that: In step (5), when the size of the casting model to be matched is inconsistent with that of the most similar model, the coordinates of the ingode and riser of the casting model to be matched relative to the center of mass of the casting model to be matched are calculated respectively according to the coordinates of the ingode and riser of the most similar model relative to the center of mass of the most similar model; the coordinates of the ingode of the most similar model relative to the center of mass of the most similar model are recorded as P1(P 1x , P 1y , P 1z ), the coordinates of the riser of the most similar model relative to the center of mass of the most similar model are P2(P 2x , P 2y , P 2z ), the coordinates of the center of mass of the casting model to be matched are O(O x , O y , O z ), the length, width and height of the casting model to be matched are X, Y and Z respectively; The coordinate P1′ of the ingrow of the casting model to be matched relative to the center of mass of the casting model to be matched is: The coordinate P2′ of the riser of the casting model to be matched relative to the center of mass of the casting model to be matched is:

10. A process recommendation method for similar castings according to claim 1, characterized in that: The preset weight w1 of the Euclidean distance value of the D2 shape feature is 0.4~0.6, the preset weight w2 of the Euclidean distance value of the modulus feature is 0.2~0.3, the preset weight w3 of the Euclidean distance value of the wall thickness feature is 0.1~0.2; the preset weight w4 of the Euclidean distance value of the envelope size feature is 0.1~0.2.

Images