TW201807423A - Circuit board inspection method with inspection route optimization function - Google Patents

Abstract

A Circuit board inspection method with inspection route optimization function, which is used for inspecting a circuit board that including a plurality of inspection nodes, wherein the circuit board inspection method comprises: grouping the plurality of inspection nodes into N groups; encoding the N groups by permutation, so as to obtain an initial inspection route group; and calculating the initial inspection route group by a Particle Swarm Optimization (PSO) algorithm, so as to obtain an optimized inspection route.

Description

Circuit board detection method with detection path optimization function

The invention relates to a circuit board detecting method, in particular to a circuit board detecting method with a detection path optimization function.

In recent years, printed circuit board inspection tools have been equipped with Surface Mounted Technology (SMT), Through Hole (THT), High Density Interconnect (HDI) and electronic packaging technology. Component technology grows rapidly. However, the area of electronic parts is miniaturized, and the board assembly density is also increased. The accompanying problem is to increase the difficulty in detection, which in turn affects the production time course of the product. For the electronics manufacturing industry, if the detection time of the product can be shortened, it means that there is more production in the same time. In terms of shipments of tens of millions of units from major manufacturers, the unit inspection time is reduced, and it is possible to take the lead in the mall or have an advantage in shipping and delivery.

In view of the above problems of the prior art, the object of the present invention is to provide a circuit board detection method with a detection path optimization function, which can obtain an optimal detection path through a discrete particle group optimization algorithm, which is not only automatic. Plan the most efficient inspection path and significantly reduce board inspection time.

An object of the present invention is to provide a circuit board detection method with a detection path optimization function for detecting a circuit board including a plurality of detection points, the circuit board detection method comprising: grouping a plurality of detection points, The N sets of detection points are formed; the N sets of detection points are arranged by string arrangement coding to form an initial detection path group; and the initial detection path group is operated by a Particle Swarm Optimization (PSO) algorithm. Action to get the best detection path.

In addition, the circuit board detecting method with the detection path optimization function of the present invention further includes: defining coordinates of a plurality of detection points.

In addition, the circuit board detecting method with the detection path optimization function of the present invention further comprises: forming N sets of detection points into an initial detection path entity by a neighboring method; and encoding the N sets of detection points in the initial detection path by the string arrangement coding. Arranging to form a plurality of detection path individuals; and randomly selecting a plurality of partial detection path individuals to form an initial detection path group.

In addition, the circuit board detection method with the detection path optimization function of the present invention further comprises: discretizing the particle group optimization algorithm, and performing the operation on the initial detection path group by the discretized particle group optimization algorithm. action.

The discretized particle swarm optimization algorithm further includes: obtaining a position and a moving speed of each of the plurality of detection path individuals in the initial detection path group.

The foregoing discretized particle swarm optimization algorithm further comprises: calculating, by the mobile cost formula, an adaptive function value of each of the plurality of detection path individuals in the initial detection path group.

The foregoing discretized particle swarm optimization algorithm further comprises: updating the individual optimal function value of each of the plurality of detection path individuals according to the adaptive function value.

The foregoing discretized particle swarm optimization algorithm further comprises: updating the group optimal function value of the initial detection path group according to the adaptive function value.

The foregoing discretized particle swarm optimization algorithm further comprises: adjusting the position and moving speed of each of the plurality of detection path individuals in the initial detection path group according to the updated individual optimal function value and the group optimal function value.

The foregoing discretized particle swarm optimization algorithm further includes: outputting an optimal detection path according to the adjusted position and moving speed.

According to the above description, the circuit board detection method with the detection path optimization function of the present invention encodes a plurality of detection points to be detected, and cooperates with the condition limit of the detection probe movement and the Chebyshev distance calculation. The cost of moving converts the path of the probe detection board to the Travelling Salesman Problem (TSP) to plan the permutation code for the detection path. Then, through the discretization addition, multiplication and subtraction elements of the particle swarm optimization algorithm, to plan the best detection path, not only can the detection time of the board be greatly reduced, but also the particle group optimization algorithm in discrete travel. The application of engineering-related engineering practices has also been of considerable help.

For a better understanding of the technical features of the present invention and the technical effects that can be achieved, the preferred embodiments and the detailed description are as follows.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of a circuit board detecting method having a detection path optimization function according to the present invention will be described with reference to the accompanying drawings. For ease of understanding, the same elements in the following embodiments are denoted by the same reference numerals.

Please refer to FIG. 1 and FIG. 2. FIG. 1 is a flowchart of a method for detecting a circuit board with a detection path optimization function according to the present invention. 2 is a schematic diagram of a circuit board detecting method for detecting a circuit board using the detection path optimization function of the present invention. The method for detecting a circuit board having the detection path optimization function of the present invention can be used, for example, to detect a circuit board 60 including a plurality of detection points v ₁ to v ₈ by two probes.

In the following embodiments, the present invention uses two probes (left probe P _L and right probe P _R ) as an example, but the present invention is not limited thereto, and the number of probes may be three or more, and detection The number of points is also not limited to the drawing. The left probe P _L can move along the x-axis and/or the y-axis, and the right probe P _R can move along the x-axis and/or the y-axis, so that the left probe P _L and the right probe P _R can be along one The detection path individual detects a plurality of detection points v ₁ to v ₈ of the circuit board 60. Wherein, the x-axis is perpendicular to the y-axis, and the left probe P _L and the right probe P _R can be driven, for example, by two sets of motors, respectively, but the invention is not limited thereto.

First, in step S100, coordinates of a plurality of detection points v ₁ to v ₈ on the circuit board 60 are defined. Next, in step S102, a plurality of detection points v ₁ to v ₈ are grouped to form N sets of detection points. In the embodiment of Fig. 2, two detection points are taken as a group. However, the present invention is not limited thereto. In other embodiments, more than three detection points may be used as a group, and the circuit board 60 may be detected by using more than three probes.

Next, in step S104, the N sets of detection points are formed into an initial detection path individual by the proximity method. Taking the detection points of the eight groups of the circuit board 60 as an example (as shown in Table 1), the detection point group a may be the detection points v ₁ , v ₈ ; the detection point group b may be the detection points v ₁ , v ₂ ; The point group c may be the detection points v ₇ , v ₆ ; the detection point group d may be the detection points v ₂ , v ₄ ; the detection point group e may be the detection points v ₂ , v ₃ ; the detection point group f may be the detection point v ₅ , v ₇ ; the detection point group g may be the detection points v ₈ , v ₇ ; and the detection point group h may be the detection points v ₃ , v ₄ . If the initial detection path is {a, b, c, d, e, f, g, h}, the left probe P _L and the right probe P _R can respectively perform detection points on the individual detection path individuals. Detection. For example, first, the left probe P _L and the right probe P _R can detect the detection points v ₁ , v ₈ respectively, and then the left probe P _L and the right probe P _R can respectively detect the detection points v ₁ , v . ₂ to test, and so on to complete the detection of eight sets of detection points.

Table I

Next, in step S106, the N sets of detection points in the initial detection path individual are arranged by the string arrangement coding to form a plurality of detection path individuals. For example, taking the initial detection path individuals {a, b, c, d, e, f, g, h} after the string arrangement as an example, the plurality of detection path individuals may be, for example, {h, b, d, c, a, f, g, e}, {g, e, c, a, d, f, b, h}, {c, b, a, h, e, f, g, d}... !=40320 individual test path combinations. Then, a plurality of partial detection path individuals are randomly selected to form an initial detection path group.

However, due to the limitation of the mechanism design, the left probe P _L and the right probe P _{R need to be} moved before the initial detection path group is operated by the Particle Swarm Optimization (PSO) algorithm. The conditions are limited to avoid collision of the left probe P _L and the right probe P _R while moving.

Please refer to FIG. 3, FIG. 4(A) and FIG. 4(B) together with FIG. 1 and FIG. 2, FIG. 3 is a moving path of the circuit board detecting method with the detection path optimization function according to condition 3 schematic diagram. 4(A) and 4(B) are schematic diagrams showing the movement path of the circuit board detection method with the detection path optimization function according to condition 4 in the present invention.

It is assumed that the left probe P _L and the right probe P _R are respectively disposed on the left and right sides of the circuit board 60, and the circuit board 60 has a common Group detection point, the detection point group where the left probe P _L and the right probe P _R are currently located The coordinates are and ,among them . Then the left probe P _L and the right probe P _R move to the next detection point group The coordinates are and ,among them , and so on Detection of group detection points.

Coordinates of the x-axis of the left probe P _L due to the limited mechanical design Must be less than or equal to the coordinate of the x-axis of the right probe P _R Meaning To avoid collision between the left probe P _L and the right probe P _R while moving. Therefore, the moving path of the left probe P _L and the right probe P _R must be excluded from the case where it is not feasible to satisfy It must meet the following four conditions:

Condition 1: In the plurality of detection path individuals, if the coordinates of the x-axis of the left probe P _L appear Coordinates larger than the x-axis of the right probe P _R Situation, meaning , the individual detection path is regarded as infeasible, and then the left probe P _L and the right probe P _R are stopped to be driven at the same time (ie, the circuit board 60 is not detected along the detection path to avoid the left probe. P _L and right probe P _R collide when moving).

Condition 2: In the individual of the plurality of detection paths, if the coordinates of the x-axis of the left probe P _L Less than the coordinate of the x-axis of the right probe P _R Meaning At the same time, the left probe P _L and the right probe P _R are driven to move. In other words, in each detection point group, the detection points with smaller coordinates of the x-axis are detected by the left probe _PL , and the detection points with larger coordinates of the x-axis are detected by the right probe P _R , thereby conforming to the mechanism. space limitations, not only to reduce the movement time left probe P _L and P _R of the right probe, left probe also prevent collision P _L and P _R occurs at the right probe is moved.

Condition 3: In the individual of the plurality of detection paths, if the coordinates of the x-axis of the left probe P _L Equal to the coordinate of the x-axis of the right probe P _R Meaning , it will be possible to create a situation where the movement path is ambiguous. Therefore, you need to define a set of mobile states. Where 0 is the coordinate of the y-axis of the left probe P _L Less than or equal to the coordinates of the y-axis of the right probe P _R Meaning ; and 1 represents the coordinates of the y-axis of the left probe P _L Coordinates larger than the y-axis of the right probe P _R Meaning . In addition, the parameters , and Respectively indicating the left probe P _L and the right probe P _R before, after, and at the current detection point group With the next checkpoint group The coordinate state parallel to the y-axis.

Therefore, as shown in Figure 3, when At the same time, the left probe P _L and the right probe P _R share six movement paths. Then, these moving paths can be calculated according to the moving cost formula to select one of the moving paths according to the calculated moving cost and feasibility.

Condition 4: Among the plurality of detection path individuals, if the above conditions 1 to 3 are not met and the and , the left probe P _L and the right probe P _R will generate a moving path interleaving, and this moving state is Express; otherwise, It means no interlacing.

Therefore, as shown in FIG. 4(A) and FIG. 4(B), when the movement path interleaving occurs, one of the left probe P _L and the right probe P _R can be driven to move first, and then the left probe P is driven. _{The other of L} and the right probe P _R moves to prevent the left probe P _L and the right probe P _R from colliding when the moving paths are staggered.

Taking the left probe P _L and the right probe P _R as the examples through the two sets of motor driving, respectively, assuming that the moving speed of the motor driving left probe P _L and right probe P _R is constant and equal, from the current detection point group Move to the next detection point group , taking the larger moving distance of the left probe P _L and the right probe P _R is the moving cost And the mobile cost formula is as follows:

(1)

Based on a single probe controlled by two sets of motors, a single probe moves from the current position to any point on the board. The time required for the movement should be the maximum value of the movement time of the two sets of motors, so use The moving cost formula (1) of Euclidean's straight line calculation does not correspond to the actual situation.

In view of the above problems, the present invention uses a Chebyshev distance calculation, also known as measure. This distance is calculated in the plane geometry, and the distance between two points is defined as the maximum value of the difference between the coordinates of each coordinate. Therefore, the moving cost of the left probe P _L and the right probe P _R will be the maximum amount of movement in the four groups of motors, so the moving cost formula can be corrected as follows:

(2)

If the moving state of the left probe P _L and the right probe P _R meets the above condition 4, the moving cost formula (2) can be modified to the following moving cost formula (3):

(3)

Therefore, the objective function for the detection problem of the left probe P _L and the right probe P _R can be expressed by the following formula:

(4)

It must comply with:

;

;as well as

.

among them For a binary variable. when When the left probe P _L and the right probe P _{R are} before moving After moving And parallel to the y-axis Under the condition, select from the detection point group Move to the detection point group ;when When the left probe P _L and the right probe P _{R are} before moving After moving And parallel to the y-axis Condition, do not select from the detection point group Move to the detection point group Detection path. In addition, the condition of the formula (4) is that the restriction left probe P _L and the right probe P _R can detect only one set of detection points at a time, and In the group detection points, each group of detection points can only be detected once.

Therefore, with the limitations of the above conditions 1 to 4, the present invention can effectively avoid collision when driving the left probe P _L and the right probe P _{R to} move.

Then the basic definition of Travelling Salesman Problem (TSP) is: Given a network , where N is the set of nodes, and A is the set of links; on this road, one is obtained at a minimum cost, starting from one point and passing through all the nodes in N, and then back to the beginning. Point the route. If the node is regarded as a group of detection points, formula (4) conforms to the model of the traveling salesman problem. however, and Restrictions increase the number of infeasible solutions in space, and increasing the complexity of time also makes the search space more like a singularity arrangement, which is more difficult than the traveling salesman problem. So a more efficient search algorithm is the key to solving such problems.

The traveling salesman problem is often used to test the effectiveness of the algorithm to solve the problem. From the experience of the heuristic algorithm used in the past, the biogeographic algorithm has a lower computational time than the genetic algorithm. High advantage. However, both of them have an evolutionary mechanism of selection in the architecture flow. It is difficult to reduce the calculation time by taking the time of competition, roulette, stability, and sorting. Does not meet the practical benefits. Based on the experience of the research, the present invention uses the particle swarm optimization algorithm as the main algorithm of the present invention, and the basic idea is to avoid selecting operators to increase the computing efficiency. The traditional particle swarm optimization algorithm is only suitable for solving numerical problems. The problem of arranging combination types is not mentioned, so it is the biggest challenge to solve the board detection problem by using particle swarm optimization algorithm.

In view of the above problem, in the step S106, the present invention arranges the N sets of detection points in the initial detection path by the string arrangement coding so that the board detection problem conforms to the evolutionary calculation.

The Particle Swarm Optimization (PSO) algorithm is an optimization method based on ethnic dynamics. The basic concept is derived from the simulation of social behavior. In a socialized group, each individual's behavior is not only affected by its past experience and cognition, but also by the overall social behavior. The particle swarm optimization algorithm has similar operational characteristics to the mutual coordination of the biological groups in the real environment and the consistency of the group behavior. Each individual has its own best self-optimal experience, and the whole best solution relative to the individual. The memory of the global optimal parameter solution of the group is called the social nature of the biological group, so that the experiences between the biological individuals can be exchanged and passed on each other. In each iteration process, all individuals in the group have their position and movement speed in the search space, and the probabilistic search strategy is adjusted according to the best experience of the past and the best behavior of the group. The learning formula is as follows:

(5)

among them, and Particles at time and Location, Indicates the current status, The state of the previous time; For the speed of movement of the particles; The number of the particle; Is the dimension of the particle; To move the inertia constant; versus The learning rate constants of the individual particles and the particle population; versus They are the individual optimal function values of the particles and the group optimal function values.

When the velocity of the particle is determined, the particle solution of the next time state Can be corrected to:

(6)

However, according to the above-mentioned continuous particle group optimization algorithm, the circuit board detection problem of the array coding method cannot be solved. Therefore, the present invention discretizes the particle swarm optimization algorithm and applies it to the traveling salesman problem to perform an arithmetic operation on the initial detection path group through the discretized particle swarm optimization algorithm. First, the meanings of the general numeric type operators "+", "-", and "‧" in equation (5) must be redefined in the meaning of the permutation combination. The explanation is as follows:

Definition 1: Suppose a solution sequence is: , . Define a swap operator Exchange solution Medium element with ,then For solution Operator element The new solution after the operation gives a new meaning to the symbol "+" in the discretized particle swarm optimization algorithm.

As shown in Table 1, the traveling salesman problem of eight nodes (that is, eight sets of detection points) is taken as an example, and its solution is If the exchange operator is ,then .

Definition 2: One or more exchange operands can be merged into one set ,among them It is an independent exchange of operands, and the order between them is meaningful. The set of exchange operands represents all of the exchange operands in the exchange sequence acting sequentially on the solution (ie, a plurality of detection path individuals), ie:

Definition 3: Different exchange orders may produce the same new solution on the same solution, and the set of exchange orders of all the same effects is called the equivalent set of exchange orders.

Definition 4: Several exchange orders can be combined into one new exchange order, defined as the merge operation elements of the two exchange orders." "." Assume two sets of exchange orders with Acting on the solution according to its order Get a new solution . Suppose there is another exchange order set Act on the same solution Can get the same solution , can be defined , with Belong to the same set of equivalences.

Definition 5: In an equivalent set of exchange order sets, the set of exchange orders with the least number of exchange operation elements is called the basic exchange order set of the equivalence set, and a basic exchange order set can be constructed as follows:

Assume given two solutions with , which needs to construct a basic exchange order set Make ,among them , . by The first element starts searching in sequence, and the process is as follows:

, , . Therefore, the first exchange operand is , ,get .

, , . Therefore, the second exchange operand is , ,get .

, , . Therefore, the third exchange operand is ,get .

, , . Therefore, the fourth exchange operand is ,get .

, , . Therefore, the fifth exchange operand is ,get .

Calculate to this step So that you get a basic exchange order: .

Since the numeric "-" operator does not have a proper symbolic representation in the discretized particle swarm optimization algorithm, the "-" operator is used.

Definition 6: Assumption Is a real number between (0,1) and ,Have Exchange order set of exchange operands ,then" The operands can be defined as follows:

(7)

among them For a random number function, a real number between [0, 1] can be generated. The exchange operation set is empty to cancel the exchange operation. The purpose of the above formula (7) is to define " The operand determines whether to keep the exchange order set in random numbers. Elements in which The greater the retention, the higher the chance.

According to the above six definitions, the relevant operands of the learning formula (5) are updated as follows:

(8)

When the velocity of the particle is determined, the particle solution of the next time state Can be corrected to:

(9)

among them, and Individual time for each of the multiple detection paths and Location, Indicates the current status, The state of the previous time; The moving speed of each of the plurality of detection paths; The number of each individual detection path; Dimensions for each of the plurality of detection paths; To move the inertia constant; versus The learning rate constants of each of the plurality of detection path individuals and the initial detection path group; versus They are the individual best function values and the group optimal function values.

The above formulas (8) and (9) are the learning formulas of the Discrete Particle Swarm Optimization (DPSO) algorithm. It can be found from equations (8) and (9) that the discretized particle swarm optimization algorithm retains the key concept of the moving speed reference group optimal and individual best of the continuous particle swarm optimization algorithm. Among them, the operand of "-" describes the difference between the two sets of coded strings; The function is to combine different exchange operands, let multiple sets of different exchange orders be combined, and use this operand to merge; and The concept is combined with a random number function to express the kinetic energy that randomly changes the movement of particles.

Then, the initial detection path group can be operated by the discretized particle group optimization algorithm. In step S108, the position of each of the plurality of detection path individuals in the initial detection path group is obtained. And moving speed Thereafter, the fitness function value (ie, the moving cost) of each of the plurality of detection path individuals in the initial detection path group is calculated by the moving cost formula (3).

Next, in step S110, the individual optimal function value of each of the plurality of detection path individuals is updated according to the fitness function value calculated by the movement cost formula (3). . If the calculated fitness function value is better than the individual optimal function value memorized by the individual in the detection path , replacing the individual optimal function value memorized by the individual in the detection path with the calculated fitness function value And adapt to the position of the function And moving speed It replaces the optimal position and optimal moving speed of the individual detected by the detection path; conversely, if the calculated adaptive function value is not better than the individual optimal function value memorized by the individual detecting path , to maintain the original state.

Then, the group optimal function value of the initial detection path group is updated according to the fitness function value calculated by the moving cost formula (3) . If the calculated fitness function value is better than the group optimal function value memorized by the initial detection path group , replacing the optimal group function value remembered by the initial detection path group with the calculated fitness function value And adapt to the position of the function And moving speed Replaces the optimal position and optimal moving speed remembered by the initial detection path group; conversely, if the calculated fitness function value is not better than the group optimal function value remembered by the initial detection path group , to maintain the original state.

Next, in step S112, according to the updated individual optimal function value Group optimal function value , adjusting the position of each of the plurality of detection paths in the initial detection path group And moving speed And further determine whether the position of all the individual detection paths has been adjusted And moving speed . If the position of all individual detection paths has not been adjusted And moving speed Then, the process returns to step S108 to calculate the adaptive function value of the remaining detection path individuals.

If the position of all individual detection paths has been adjusted And moving speed , proceeding to step S114 to determine the position of each of the plurality of detection paths according to the adjustment And moving speed , output the best detection path.

Please refer to FIG. 5(A), FIG. 5(B), FIG. 5(C) and FIG. 5(D) together with FIG. 1 and FIG. 2, FIG. 5(A), FIG. 5(B), FIG. 5(C). And FIG. 5(D) is a schematic diagram of the calculation process of the circuit board detecting method with the detection path optimization function of the present invention.

Take the eight sets of detection points shown in Figure 2 and Table 1 as an example to illustrate the process of particle evolution:

Suppose there is a particle individual (ie, the detection path individual) at the time The location is The detection path is shown in Fig. 5(A); it is assumed that the individual optimal solution (ie, the individual optimal function value) memorized by the individual particles is The detection path is shown in Figure 5(B); suppose the population optimal solution of the current particle population (ie, the population optimal function value of the initial detection path group) is , whose detection path is as shown in Figure 5(C); and assuming that the individual particles are in time The moving speed is And make an exchange order set .

Individual best solution and individual difference calculation:

Suppose the two real numbers generated by the random number function versus , then keep and And make an exchange order set .

Group optimal solution and individual difference calculation:

;

Suppose the four real numbers generated by the random number function , , and ,then with Will be abandoned, versus Will be retained and make one .

Calculated according to formula (8), Can be expressed as the following equation:

;

;

;

After calculation, you can get the updated particles. The detection path is as shown in Fig. 5(D).

Therefore, the addition, multiplication and subtraction elements of the discretized particle group optimization algorithm can not only plan the optimal detection path, but also greatly reduce the detection time of the board, and have industrial utilization value.

The various embodiments disclosed above are illustrative only and not limiting. Equivalent modifications or variations of the embodiments of the present invention are intended to be included within the scope of the appended claims.

60‧‧‧ boards
S100, S102, S104, S106, S108, S110, S112, S114‧‧ steps
P _L ‧‧‧Left probe
P _R ‧‧‧Right probe
v ₁ , v ₂ , v ₃ , v ₄ , v ₅ , v ₆ , v ₇ , v ₈ ‧‧‧ detection points

1 is a flow chart of a method for detecting a circuit board with a detection path optimization function according to the present invention.

2 is a schematic diagram of a circuit board detecting method for detecting a circuit board using the detection path optimization function of the present invention.

FIG. 3 is a schematic diagram of a moving path when the circuit board detecting method with the detection path optimization function according to the present invention meets the condition 3.

4(A) and 4(B) are schematic diagrams showing the movement path of the circuit board detection method with the detection path optimization function according to condition 4 in the present invention.

5(A), 5(B), 5(C) and 5(D) are schematic diagrams showing the calculation process of the circuit board detecting method with the detection path optimization function of the present invention.

Claims (10)

A circuit board detecting method for detecting a path optimization function for detecting a circuit board including a plurality of detecting points, wherein the circuit board detecting method comprises: grouping the plurality of detecting points to form N sets of detecting Pointing; arranging the N sets of detection points by a string arrangement code to form an initial detection path group; and performing a preliminary detection path group by a Particle Swarm Optimization (PSO) algorithm The operation is performed to obtain an optimal detection path. The method for detecting a circuit board with a detection path optimization function according to the first aspect of the patent application includes: defining a coordinate of the plurality of detection points. The method for detecting a circuit board having the detection path optimization function according to the first aspect of the patent application includes: forming, by a neighboring method, the N sets of detection points into an initial detection path; and arranging the coding pairs through the string The N sets of detection points in the initial detection path individual are arranged to form a plurality of detection path individuals; and the plurality of detection path individuals are randomly selected to form the initial detection path group. The method for detecting a circuit board having the detection path optimization function described in claim 3, further comprising: discretizing the particle swarm optimization algorithm to optimize the particle swarm optimization through the discretization The method performs the operation on the initial detection path group. The method for detecting a circuit board with a detection path optimization function according to claim 4, wherein the discretization of the particle group optimization algorithm further comprises: obtaining each of the plurality of initial detection path groups Detecting the location of one of the individual paths and a moving speed. The method for detecting a circuit board with the detection path optimization function described in claim 5, wherein the discretization of the particle swarm optimization algorithm further comprises: calculating the initial detection path group by using a mobile cost formula One of the plurality of detection path individuals adapts to the function value. The method for detecting a circuit board with the detection path optimization function described in claim 6 wherein the discretization of the particle swarm optimization algorithm further comprises: updating each of the plurality of detection paths according to the adaptation function value. One of the individual's best function values. The method for detecting a circuit board with a detection path optimization function according to the seventh aspect of the patent application, wherein the discretization of the particle swarm optimization algorithm further comprises: updating the initial detection path group according to the adaptation function value; A group of best function values. The method for detecting a circuit board with the detection path optimization function described in claim 8 wherein the discretization of the particle swarm optimization algorithm further comprises: according to the updated optimal function value of the individual and the The group optimal function value adjusts the position of the plurality of detection path individuals in the initial detection path group and the moving speed. The method for detecting a circuit board having a detection path optimization function according to claim 9 is characterized in that the discretization of the particle group optimization algorithm further comprises: outputting according to the adjusted position and the moving speed, The best detection path.