Method for performing online pick-and-place path optimization simultaneously during the production of a surface mount machine

A method of performing online optimization on a pick-and-place path during production of a surface mount machine, comprising following steps:

• at step 1, based on a mount optimization solver of the surface mount machine, obtaining an initial pick-and-place process optimization result, and transmitting the result into an industrial control computer of the surface mount machine to start a surface mount production;
• at step 2, obtaining surface mount machine parameters, a circuit board production data and an initial pick-and-place path optimization result from the pick-and-place process optimization result;
• at step 3, based on an imported initial pick-and-place (PAP) path optimization result, determining candidate positions for placement points of various types of components, along with an initial sequence array of placement points for the various types of components, comprising following steps:
• at step 31, a PAP cycle count k=1 is initialized;
• corresponding to each type of components c∈[1, . . . , C] with a total number of component types being C, an array PL{c}∈[PL{1}, . . . , PL{C}] with a row number n{c} and a column number 2 with all components being zero is initialized to store the candidate positions of the placement points contained by each type of components, wherein PL{c} is referred to as a candidate position array of the components of a c-th type;
• corresponding to each type of components c∈[1, . . . , C], an array ps0{c}∈[ps0{1}, . . . , ps0{C}] with a row number n{c} and a column number 1 with all components being zero is initialized to store the placement point assignment sequence of each type of components, wherein ps0{c} is referred to as a assignment sequence array of the components of the c-th type;
• corresponding to each type of components c∈[1, . . . , C], a count variable i{c}=1 is initialized as an index of the placement points of each type of components;
• at step 32, if k>K, wherein K is direct to a total number of the PAP cycles, it indicates that one traversal has been completed for all PAP cycles, and step 4 is performed, and otherwise, a sucking rod count variable h=1 is initialized and step 33 is performed;
• at step 33, if h>H, wherein H is directed to a total number of sucking rods, it indicates that one traversal has been completed for all sucking rods, cyclic variable k=k+1 is updated and the step 32 is repeated, and otherwise, step 34 is performed;
• at step 34, a component type picked and mounted by a h-th sucking rod in a k-th PAP cycle is c=CG(k,h), and it is determined whether c=0;
• if yes, it indicates that the sucking rod in this cycle does not perform any mounting, a cyclic variable h=h+1 is updated, and step 33 is repeated, and otherwise, step 35 is performed;
• at step 35, PL{c}(i{c},1)=k and PL{c}(i{c},2)=h are assigned and ps0{c}(i{c})=PA0(k,h) is also assigned;
• a count variable i{c}=i{c}+1 is updated, and a cyclic variable h=h+1 is updated, and step 33 is repeated;
• and
• at step 4, performing initialization on a tabu search, comprising following steps:
• at step 41, an upper limit of the number of times of local search is initialized and assigned to IT=┌(P)0.5┐, wherein a symbol ┌ ┐ represents performing up-rounding for a value inside;
• a component type count c=1 is initialized;
• at step 42, if c>C, it indicates that one traversal has been completed for all component types, and step 48 is performed and otherwise, step 43 is performed;
• at step 43, the number of times that the components of a c-th type perform swap operation is SWN{c}=0.5*n{c}*(n{c}−1);
• an array SWL{c} with a row number SWN{c} and column number 2 with all components being zero is initialized to store swap operations the components of the c-th type perform;
• a swap operation count swn=0 is initialized;
• and a prior placement point count ni=1 is initialized;
• at step 44, if ni>n{c}−1, it indicates that one traversal has been completed for all prior placement points, and step 47 is performed;
• and otherwise, a posterior placement point count nj=ni+1 is initialized and step 45 is performed;
• at step 45, if nj>n{c}, it indicates that one traversal has been completed for all posterior placement points, and a count variable ni=ni+1 is updated, and step 44 is repeated, and otherwise, step 46 is performed;
• at step 46, a count variables swn=swn+1, SWL{c}(swn,:)=[ni,nj] are updated, and a count variable nj=nj+1 is updated and step 45 is repeated;
• at step 47, an array TC{c} with a row number SWN{c} and a column number 1 with all components being zero is initialized as a tabu table for the components of the c-th type, wherein each component in this array indicates whether a corresponding swap operation is disabled or not and a number of times of disabling, and an upper limit of the number of times of disabling is TL{c}=┌(n{c})0.5┐;
• a count variable c=c+1 is updated and step 42 is repeated;
• and
• at step 48, an array swb with a row number C and a column number 1 with all components being zero are initialized to store an optimal swap operation of each type of component;
• the array swt with the row number C and the column number 1 with all components being zero are initialized to store a non-disabled optimal swap operation of each type of components;
• at step 5, employing the tabu search to compare and select optimal result for pick-and-place path optimization, transmitting a newly found optimal result to an industrial control computer of the surface mount machine, and based on this result, commencing subsequent surface mount production starting from a next circuit board, comprising following steps:
• at step 51, a historical optimal pick-and-place path length DZ, a currently-solved pick-and-place path length DZb, a currently-solved placement point distribution result PAb, a currently-solved placement sequence result PSb are initialized, and DZ=DZ0, DZb=DZ0, PAb=PA0, PSb=PS0 are assigned;
• corresponding to each type of components c∈[1, . . . , C], ps{c}=ps0{c} and psb{c}=ps{c} are initialized and assigned to store an optimal assignment sequence, and pst{c}=ps{c} is initialized and assigned to store a non-disabled optimal assignment sequence;
• and a non-improved search count variable IM=0 is initialized and assigned;
• at step 52, if IM>IT, it indicates that the number of times of the non-improved search reaches an upper limit, and step 51 is performed to start a next round of search, and otherwise, a non-disabled shortest path length DZt is initialized to infinity, a local search count variable it=1 is initialized and step 53 is performed;
• at step 53, if it>IT, it indicates that the number of times of the local search reaches the upper limit and step 54 is performed;
• and otherwise, corresponding to each type of components c1∈[1, . . . , C], psn{c1}=ps{c1} is initialized and assigned;
• and the component type count c=1 is initialized, and step 6 is performed;
• at step 54, it is determined whether DZb<DZ;
• if yes, it indicates that a better pick-and-place path is obtained, and step 55 is performed and otherwise, step 56 is performed;
• at step 55, the optimal result is updated and then transmitted into the industrial control computer of the surface mount machine and by starting from a next circuit board, subsequent surface mount production is performed based on the result;
• and
• at step 56, a non-disabled optimal result is updated;
• and
• at step 6, performing neighborhood search operations on arrays of assignment sequences for various types of components, decoding manipulated assignment sequence arrays to obtain new placement point distribution results and correspondingly, obtaining optimal placement sequencing results based on dynamic programming;
• if new optimization results correspond to the better pick-and-place path;
• and storing the new optimization results as the optimal result.