A paint spray booth intelligent interconnection control method and system

Abstract

The present invention belongs to the field of intelligent control technology, and discloses an intelligent interconnection control method and system for a paint spraying booth; the method comprises: dividing the interior of the paint spraying booth into n preliminary areas; obtaining a representative environmental parameter value vector of each preliminary area by using a salp group algorithm; constructing a directed graph model based on clustering representative environmental parameter value vectors of each area as nodes; planning the shortest path in the directed graph model, and taking each node on the shortest path as a priority adjustment target; solving the optimal control parameter of the preliminary area corresponding to the priority adjustment target by using an optimization algorithm; and sending the optimal control parameter to an execution device of the corresponding area to realize precise control of each area, which not only improves the paint spraying quality and production efficiency, but also reduces energy consumption and environmental pollution.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and more specifically, to an intelligent interconnection control method and system for a paint spraying and baking room. Background Art

[0002] The patent with application publication number CN110955198A discloses an intelligent interconnected control system for a paint booth, including: a boot page module, a startup page module, a registration module, a login module, a display page module, a camera module, an equipment data display module, a user information module, and a database; the display page module, the camera module, and the equipment data display module are operation modules of the system; the user terminal accesses the operation module of the system through a user account; the user account is marked as multiple levels, and user accounts of different levels access different operation modules in the system; when the user terminal logs in using the user account, the user terminal accesses the corresponding operation module of the system according to the user level, and reads or operates the data of the operation module; the paint booth is made intelligent, so that employees, managers and other workshop staff can remotely control various parameters in the paint booth, improve work efficiency, and solve problems that may be encountered with paint equipment in one stop from control to after-sales.

[0003] However, it is still difficult to achieve refined and intelligent control of the internal environment of the baking room, and it is impossible to meet the strict requirements of high-quality painting. For example, in large-scale automobile painting workshops, due to the vast space and complex structure, it is difficult for existing technologies to accurately obtain complete environmental information inside the entire baking room, resulting in the appearance of monitoring blind spots and control dead corners; secondly, the environment inside the baking room often presents significant spatial non-uniformity, but the existing methods lack effective differentiated control strategies and cannot be accurately adjusted according to the special needs of different areas; this may lead to poor painting quality in some areas, such as uneven paint film thickness or surface defects; in addition, traditional control systems often use simple feedback control or preset programs, lack integrity and coordination, and it is difficult to achieve comprehensive optimization among multiple control targets; this limitation is particularly obvious when dealing with complex multi-target control tasks, which may lead to waste of resources or low production efficiency; and in the face of sudden environmental changes or abnormal situations (such as equipment failure, external interference, etc.), the adaptability and response speed of existing technologies are insufficient, and it is difficult to achieve fast and accurate adjustment, which may cause fluctuations in paint quality or even production interruptions; these problems seriously restrict the improvement of quality and efficiency.

[0004] In view of this, the present invention proposes an intelligent interconnection control method and system for a paint spraying and baking room to solve the above problems. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solutions: A method for intelligent interconnection control of a spray paint baking room, comprising: S1, dividing the interior of the spray paint baking room into n preliminary areas;

[0006] S2, using the salp swarm algorithm to obtain a representative environmental parameter value vector for each preliminary area;

[0007] S3, constructing a directed graph model based on taking the representative environmental parameter value vectors of each region cluster as nodes;

[0008] S4. Plan the shortest path in the directed graph model, and take each node on the shortest path as a priority adjustment target;

[0009] S5. Use an optimization algorithm to solve the optimal control parameters of the preliminary area corresponding to the priority adjustment target; and send the optimal control parameters to the execution device of the corresponding preliminary area.

[0010] Furthermore, the method of dividing the interior of the spray booth into n preliminary areas includes:

[0011] N structured light scanning positions are set inside the spray paint booth, and structured light is used to scan the inside of the spray paint booth at the structured light scanning positions to obtain original point cloud data; the original point cloud data is preprocessed to obtain complete point cloud data;

[0012] Using the complete point cloud data, the geometric shape inside the spray paint booth is modeled to obtain a three-dimensional geometric model of the interior of the spray paint booth;

[0013] A sample point set S1 is obtained inside the spray paint room, where each sample point s' contains its spatial coordinates and corresponding environmental parameters, including temperature, humidity and air flow velocity;

[0014] Based on the sample point set S1, the parameters of the radial basis function are fixed to obtain the interpolation function, and all points in the complete point cloud data in the geometric three-dimensional model are traversed. For each point, the environmental parameter value is calculated using the interpolation function, and the coordinates of each point and the corresponding environmental parameter value are stored as environmental point cloud data; the environmental point cloud data constitutes the environmental parameter field;

[0015] The threshold range of the environmental parameters is preset, and the environmental parameter field of the entire geometric three-dimensional model is divided into initial regions according to the corresponding threshold range; and the initial regions are merged and split to obtain n preliminary regions.

[0016] Furthermore, the initial area is divided into the following ways:

[0017] Traverse each point z' in the environmental point cloud data, assign the point z' to the corresponding threshold range according to the environmental parameter value at the point z', and gather all the points that fall within the same threshold range into an initial area;

[0018] The method of merging and splitting the initial area includes:

[0019] Calculate every two adjacent initial regions and The difference in environmental parameters between ;

[0020] ;

[0021] in, The initial area The mean internal temperature, The initial area The mean internal temperature, is the maximum temperature, is the minimum temperature, The initial area The mean humidity inside The initial area The mean humidity inside is the maximum value of humidity, is the minimum value of humidity, The initial area The mean internal air velocity, The initial area The mean internal air velocity, is the maximum air velocity, is the minimum value of air flow velocity; is the temperature weight coefficient, is the humidity weight coefficient, is the air flow velocity weight coefficient;

[0022] Set the region merging threshold ,like Less than , then and Merge into a new region, repeat until no more merging is possible;

[0023] Calculate the size D' of each initial region; set the size threshold S_lit for splitting the initial region. If D' is greater than S_lit, split the corresponding initial region;

[0024] The ways to split include:

[0025] For each point in the environmental parameter field, calculate the gradient of each environmental parameter at the point respectively, and merge them into a comprehensive gradient; traverse all points in the entire initial area, calculate the comprehensive gradient at each point, preset the gradient threshold A1, mark all points with comprehensive gradients greater than A1 as high gradient points, and aggregate adjacent high gradient points into connected high gradient areas. For each high gradient area H1, fit a best fitting plane PH; calculate the intersection line LH of each best fitting plane and the initial area, sample m4 points along the intersection line LH, and calculate the sum of the comprehensive gradients SH at these points; select the best fitting plane with the largest sum of comprehensive gradients as the optimal segmentation plane; divide the initial area into two along the optimal segmentation plane to generate two sub-areas, and repeat the splitting until the size of all initial areas is less than S_lit; that is, n preliminary areas are obtained.

[0026] Furthermore, the method of preprocessing the original point cloud data includes:

[0027] For each point in the original point cloud data , calculate the normal vectors of all points in its neighborhood sphere Np(r); where r is the radius of the neighborhood sphere; for points in Np(r) , calculate its smoothed covariance matrix ;

[0028] ;in, is the total number of points in the neighborhood ball Np(r), is the first points; For point The attribute vector of is the first The attribute vector of each point; is a comprehensive attribute function, is the mean vector of all points in Np(r);

[0029] ;in, is the balance parameter, is the distance attenuation bandwidth; is the attribute attenuation bandwidth;

[0030] The smoothed covariance matrix Perform eigenvalue decomposition to obtain the corresponding eigenvectors and eigenvalues, and find the smallest eigenvalue, which is the normal vector of all points in Np(r) ;calculate Point and its neighboring points Np(r) The normal vector difference angle ;in, Yes The normal vector of

[0031] Preset angle threshold ,like , then point is a noise point, and removes it from Np(r), updates the points in Np(r), and recalculates Normal vector of a point , until no more points are removed within Np(r), and the preliminary point cloud data is obtained;

[0032] Based on the preliminary point cloud data, the hole area is obtained, the surface grid structure of the hole area is constructed, and new point cloud data is generated based on the surface grid structure; the new point cloud data and the preliminary point cloud data are aggregated to obtain complete point cloud data;

[0033] The method of obtaining the hole area based on the preliminary point cloud data includes:

[0034] For each point in the preliminary point cloud data , calculate its Euclidean distance to other points, and record the distance points The nearest k points, which form the point The k nearest neighbor point set N_k(p);

[0035] Point-based and each point in its k-nearest neighbor point set N_k(p) , calculation point Point cloud density ;

[0036] ;in, For point With point The Euclidean distance between is the weight function; is the index parameter;

[0037] Weight function ;in, For the The weight coefficient of the geometric statistical features, Based on point and Point No. Geometric statistical characteristics;

[0038] Preset density threshold ,like , then the marking point For each point marked as a sparse point, find its nearest point that is not marked as a sparse point. If the distance between the two is greater than the preset clustering threshold R, the point Mark them as hole points, and aggregate all adjacent hole points to get the hole area.

[0039] Furthermore, the surface grid structure is constructed in the following manner:

[0040] For each hole region, find the nearest m1 points that are not marked as sparse points, and use these points as the initial neighborhood point set RN of the hole region; calculate the closest distance d1 between each point in RN and the hole region, and if d1 is less than the preset neighborhood threshold R1, add it to RN, and repeat until RN no longer increases; obtain the neighborhood point set of the hole region;

[0041] Preset sphere radius r1 and accuracy threshold angle ; Randomly select a seed point x from the neighborhood point set as the initial sphere center, and construct a sphere B(x, r1) with r1 as the radius;

[0042] Find a set S of all points on the ball B(x, r1), S contains at least 3 points, then perform local surface fitting on the points in S to obtain a local surface patch F, select a new sphere center point x1 from S, and construct a new sphere B(x1, r); and the new sphere center point q satisfies the preset constraints;

[0043] Constraints include overlap constraints and angle constraints;

[0044] The overlap constraint includes: the new ball B(x1, r) has an overlapping area with the ball B(x, r1), and the number of points in the overlapping area is greater than the preset overlap threshold M1;

[0045] The angle constraints include: the angle between the new ball B(x1, r) and the normal vector of the local surface patch F is less than the accuracy threshold angle , and the curvature change rate at the new sphere center point x1 is less than the preset curvature threshold M2;

[0046] Repeatedly construct a new local surface patch on the new ball B(x1, r) and select a new sphere center point until no new sphere center point that meets the constraints can be found. Select a new seed point from the remaining neighborhood point set and repeatedly construct a new local surface patch F'. Topologically connect the new local surface patch F' with the existing local surface patch F to obtain the surface mesh structure G of the hole area.

[0047] The topological connection methods include:

[0048] Calculate the distance field of F and F' in the overlapping area R2 of the two ,in, is a point within the overlapping region R2; and based on the distance field Find the best matching position of the two F and F' in the overlapping area, and at the best matching position, splice F and F' together through interpolation method; that is, obtain the surface mesh structure G of the hole area;

[0049] Distance Field ;in, is a point on F, is a point on F';

[0050] After calculating the distance fields of all points in the overlapping area R2, find the surface composed of points whose distance fields are equal to 0, that is, the best matching position of F and F'.

[0051] Furthermore, the representative environmental parameter value vector is obtained by:

[0052] Density cluster sampling is performed on the environmental point cloud data in each preliminary area to obtain a set of M4 sample points, recorded as sample point set X; the environmental parameters of each sample point in the sample point set X are constructed into a multidimensional vector as the input data vector ;

[0053] A sample point is randomly selected from the sample point set X as the first initial center point c1, and added to the preset initial center point set C1; for each sample point in X , calculate its Euclidean distance to the current closest c1 ; and calculate each sample point The probability of being selected as the next center point ;in, is the density penalty coefficient, For sample points The local density value of

[0054] From X according to the probability Randomly select a sample point and add it as a new center point to C1. Repeat until there are K center points in C1. These K center points are used as the initial center point set of the salp group.

[0055] For each input data vector , calculate its weighted distance to all center points , the input data vector The corresponding sample points are classified into the cluster corresponding to the center point with the smallest weighted distance;

[0056] Weighted distance ;in, The input data vector The weight of the dimension, and is the amplification parameter, is the input data vector No. dimensions, The center point Dimensions;

[0057] For each cluster, recalculate the new centroid of all points in the cluster as the new center point of the corresponding cluster;

[0058] The new centroid is calculated as:

[0059] ;in, To control the distance weight parameter; is the current cluster; is the old centroid of the current cluster, is the new centroid of the current cluster, For sample points and The weighted distance between

[0060] The clustering target is defined as the minimum sum of weighted distances from all sample points in the cluster to the centroid, and the process is repeated until the clustering target is reached or the preset number of iterations is reached; at this time, the center point of each cluster is the representative environmental parameter value vector of each cluster; for each preliminary area, the center point of its corresponding cluster is found, which is the representative environmental parameter value vector of the corresponding preliminary area.

[0061] Furthermore, the directed graph model is constructed in the following manner:

[0062] The representative environmental parameter value vector of each preliminary area is taken as a node to form a node set; for any two nodes in the node set, the similarity between them is calculated;

[0063] The similarity calculation formula is:

[0064] ;in, for and The similarity of For the A vector of representative environmental parameter values ​​for the preliminary regions, For the A vector of representative environmental parameter values ​​for the preliminary regions, for and The Euclidean distance between For the The area of ​​the initial region, For the The area of ​​the preliminary region; is an adjustable parameter;

[0065] Preset similarity threshold , traverse all nodes in the node set, for any two nodes and ,like Greater than or equal to , then in and , connect a directed edge between them; add the directed edge to the directed edge set.

[0066] The weight of each directed edge is defined as the similarity between nodes, and then the directed graph model GL=(NL, EL, WL) is obtained, where NL is the node set, EL is the directed edge set, and WL is the set of directed edge weights.

[0067] Furthermore, the shortest path is obtained by:

[0068] Initially, each node in the directed graph model is regarded as an independent community. All nodes in the directed graph model are traversed, and the module gain after adding any node to other communities is calculated.

[0069] Add any node to the community that can obtain the maximum modulus gain, and then rebuild a new network. In the new network, each community is regarded as a new node. Repeat until the modulus gain no longer increases. The final network is the community structure.

[0070] The calculation formula of modulus gain is:

[0071] ;in, To make the node Join the community The modulus gain after is the set representing the remaining nodes; , and To balance the adjustable parameters, is the exponential coefficient, To represent the node With the community The sum of the weights of the edges between all nodes in the To represent the node degree; is the sum of the weights of all edges in the network; For Node With the community The Euclidean distance of

[0072] Define the starting community and regard it as a separate subgraph. The nodes in the subgraph are all the preliminary areas in the community. The edges and weights between the nodes are inherited from the directed graph model. In the subgraph, select any two nodes as the starting point and the end point, and use the shortest path algorithm to calculate the preliminary shortest path from the starting point to the end point; repeatedly traverse all pairs of nodes in the subgraph, calculate their preliminary shortest paths, record all nodes on the preliminary shortest paths, remove duplicate nodes, and obtain the adjusted node set, which is the final shortest path.

[0073] Furthermore, the method for solving the optimal control parameters includes:

[0074] Construct the control objective function, the formula of the control objective function is:

[0075] ; among them, among them, is the control parameter vector; It is the first A function of environmental parameters; It is The weight coefficient of each environmental parameter; For the initial area Standard quantity of environmental parameters; is the control parameter vector The control objective function value of

[0076] For the control parameter vector , define the initial solution ; Set an initial temperature , and a termination temperature ; Define the temperature drop curve as: ,in, is the cooling rate, is the number of iterations, for The current temperature at the iteration;

[0077] At the current temperature, the control parameter vector Perturbations to generate new solutions , calculate the new solution based on the control objective function The corresponding control objective function value ;like , then accept the new solution, that is ;like , then with probability Accept the new solution and repeat until the preset maximum number of iterations is reached. During this process, the current temperature is reduced according to the set temperature drop curve;

[0078] Probability ,in, is a constant parameter; is the maximum number of iterations;

[0079] Until the current temperature drops to the termination temperature Or the control objective function value is terminated when there is no change in P consecutive iterations; at this time That is the optimal control parameter vector; devectorize the optimal control parameter vector to obtain the optimal control parameter.

[0080] A paint spray booth intelligent interconnection control system, which is used to implement a paint spray booth intelligent interconnection control method, comprising: an area division module, which is used to divide the interior of the paint spray booth into n preliminary areas;

[0081] A feature extraction module, for obtaining a representative environmental parameter value vector for each preliminary area using a salp swarm algorithm;

[0082] The model building module constructs a directed graph model based on the representative environmental parameter value vectors of each region cluster as nodes;

[0083] The path planning module is used to plan the shortest path in the directed graph model and take each node on the shortest path as a priority adjustment target;

[0084] The control module is used to use an optimization algorithm to solve the optimal control parameters of the preliminary area corresponding to the priority adjustment target; send the optimal control parameters to the execution device of the corresponding preliminary area; and each module is connected by wired and / or wireless means.

[0085] The technical effects and advantages of the intelligent interconnection control method and system of a spray paint baking room of the present invention are as follows:

[0086] The present invention realizes high-precision perception and comprehensive analysis of the internal environment of the paint spraying booth. Through high-precision three-dimensional reconstruction of the internal environment of the paint spraying booth, it can perform adaptive regional division according to the distribution of actual environmental data, so that the division result is more in line with the actual environmental distribution; and effectively capture the correlation between regions; through community discovery and shortest path planning, the adjustment order and key targets are scientifically determined, resource waste is avoided, and control efficiency is improved. In addition, comprehensive optimization control of multiple environmental parameters is realized, which greatly improves the overall control accuracy. The control strategy takes into account the mutual influence between regions and realizes overall coordinated optimization, making the internal environment of the paint spraying booth more balanced and stable; this refined and intelligent control method not only improves the paint spraying quality and production efficiency, but also reduces energy consumption and reduces environmental pollution. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 A schematic diagram of an intelligent interconnection control method for a paint spraying and baking room of the present invention;

[0088] Figure 2 It is a schematic diagram of an intelligent interconnection control system of a paint spraying and baking room of the present invention. DETAILED DESCRIPTION

[0089] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0090] Example 1

[0091] See also Figure 1 As shown, the intelligent interconnection control method of a paint spray booth described in this embodiment includes:

[0092] S1, dividing the interior of the spray paint room into n preliminary areas;

[0093] S2, using the salp swarm algorithm to obtain a representative environmental parameter value vector for each preliminary area;

[0094] S3, constructing a directed graph model based on taking the representative environmental parameter value vectors of each region cluster as nodes;

[0095] S4. Plan the shortest path in the directed graph model, and take each node on the shortest path as a priority adjustment target;

[0096] S5. Use an optimization algorithm to solve the optimal control parameters of the preliminary area corresponding to the priority adjustment target; send the optimal control parameters to the execution device of the corresponding preliminary area to achieve precise control of each area.

[0097] The methods for dividing the interior of the spray booth into n preliminary areas include:

[0098] N structured light scanning positions are set inside the paint booth, and structured light is used at the structured light scanning positions to scan the inside of the paint booth to obtain original point cloud data (recording a large number of three-dimensional coordinate points on the inner surface of the paint booth).

[0099] The original point cloud data is preprocessed to obtain complete point cloud data; specifically, for each point in the original point cloud data , calculate the normal vectors of all points in its neighborhood sphere Np(r); where r is the radius of the neighborhood sphere; the neighborhood sphere Np(r) is a point is a spherical area with radius r at its center.

[0100] For points within Np(r) , calculate its smoothed covariance matrix ;

[0101] ;

[0102] in, is the total number of points in the neighborhood ball Np(r), is the first points; For point The attribute vector of is the first The attribute vector of a point, and Do not index the same point at the same time; is a comprehensive attribute function, is the mean vector of all points in Np(r).

[0103] ;in, is the balance parameter, is the distance attenuation bandwidth, distance The closer the points are, the greater the weight; The attribute attenuation bandwidth is the same as the attribute weight. The more similar the attributes are, the greater the weight is.

[0104] Among them, point The attribute vector of the point is obtained by It is obtained by concatenating the vectors composed of the reflectivity and curvature at each location.

[0105] The smoothed covariance matrix Perform eigenvalue decomposition (such as Jacobi method) to obtain the corresponding eigenvectors and eigenvalues, and find the smallest eigenvalue, which is the normal vector of all points in Np(r) .

[0106] calculate Point and its neighboring points Np(r) The normal vector difference angle ;

[0107] ;in, Yes The normal vector of The points are calculated in the same way.

[0108] Preset angle threshold (usually between 5 and 30 degrees, depending on the point cloud quality). , then point is a noise point, and removes it from Np(r), updates the points in Np(r), and recalculates Normal vector of a point , until no more points within Np(r) are removed, and the preliminary point cloud data is obtained.

[0109] The hole area is obtained based on the preliminary point cloud data. Specifically, for each point in the preliminary point cloud data , calculate its Euclidean distance to other points, and record the distance points The nearest k points, which form the point The k nearest neighbor point set N_k(p).

[0110] Point-based and each point in its k-nearest neighbor point set N_k(p) , calculation point Point cloud density ;

[0111] ;in, For point With point The Euclidean distance between is the weight function; It is an exponential parameter used to adjust the influence weight of distance.

[0112] Weight function ;in, For the The weight coefficient of the geometric statistical feature is a learnable parameter. Based on point and Point No. The geometric statistical features include distance feature (calculated by distance measurement function, such as Euclidean distance), curvature feature (difference measurement of curvature value), normal vector feature (cosine value of normal vector angle), density feature (difference of density estimation value) and curvature change rate feature (difference measurement of curvature change rate).

[0113] Preset density threshold (Use a histogram to analyze the density distribution of the point cloud and set it to a certain percentage of the average density, such as 20% or 50%). , then the marking point For each point marked as a sparse point, find its nearest point that is not marked as a sparse point. If the distance between the two is greater than the preset clustering threshold R, the point Mark them as hole points, and aggregate all adjacent hole points to get the hole area.

[0114] For each hole area, find the nearest m1 points that are not marked as sparse points, and use these points as the initial neighborhood point set RN of the hole area; calculate the nearest distance d1 between each point in RN and the hole area. If d1 is less than the preset neighborhood threshold R1, add it to RN, and repeat until RN no longer increases; obtain the neighborhood point set of the hole area.

[0115] The neighborhood point set is processed to construct the surface grid structure of the hole area. Specifically, the sphere radius r1 and the accuracy threshold angle are preset. ; The radius of the sphere r1 determines the resolution of the surface grid structure, the accuracy threshold angle Controls the smoothness of the generated surface mesh structure.

[0116] Randomly select a seed point x from the neighborhood point set as the initial sphere center, and construct a sphere B(x, r1) with r1 as the radius;

[0117] Find the set S of all points on the ball B(x, r1), S contains at least 3 points, then perform local surface fitting on the points in S to obtain a local surface patch F, select a new center point x1 from S, and construct a new ball B(x1, r); and the new center point q satisfies the preset constraints.

[0118] Constraints include overlap constraints and angle constraints;

[0119] The overlap constraints include: the new sphere B(x1, r) has an overlapping area with sphere B(x, r1), and the number of points in the overlapping area is greater than the preset overlap threshold M1 (usually set to 10% to 30% of the sphere volume);

[0120] The angle constraints include: the angle between the new ball B(x1, r) and the normal vector of the local surface patch F is less than the accuracy threshold angle , and the curvature change rate at the new sphere center point x1 is less than the preset curvature threshold M2 (set to 2 to 5 times the average curvature) to avoid constructing a new sphere in an area with drastic curvature changes.

[0121] Repeatedly construct a new local surface patch on the new ball B(x1, r) and select a new sphere center point until no new sphere center point that meets the constraints can be found. Select a new seed point from the remaining neighborhood point set and repeatedly construct a new local surface patch F'. Topologically connect the new local surface patch F' with the existing local surface patch F to obtain the surface mesh structure G of the hole area.

[0122] The topological connection methods include:

[0123] Calculate the distance field of F and F' in the overlapping area R2 of the two ,in, is a point within the overlapping region R2; and based on the distance field Find the best matching position of the two F and F' in the overlapping area. At the best matching position, use the interpolation method (using the bicubic B-spline interpolation surface) to smoothly splice F and F' together; that is, the surface mesh structure G of the hole area is obtained.

[0124] Distance Field ;in, is a point on F, is a point on F';

[0125] After calculating the distance fields of all points in the overlapping area R2, find the surface composed of points whose distance fields are equal to 0, that is, the best matching position of F and F'.

[0126] Generate new point cloud data based on the surface mesh structure. Specifically, define the expected point cloud density. , traverse each local surface patch of the surface mesh structure, and for each local surface patch, according to its area and , determine the number of sampling points n3, use the MLS surface equation to perform n3 interpolation sampling inside the local surface patch, generate n3 new points, aggregate all the new points, and obtain new point cloud data.

[0127] The new point cloud data and the preliminary point cloud data are aggregated to obtain complete point cloud data; the geometric shape inside the paint booth is modeled using the complete point cloud data to obtain a geometric three-dimensional model of the interior of the paint booth.

[0128] A sample point set S1 is obtained inside the paint spraying and baking room, wherein each sample point s' contains its spatial coordinates and corresponding environmental parameters, and the environmental parameters include temperature, humidity and air flow velocity.

[0129] Based on the sample point set S1, the parameters of the radial basis function are fixed to obtain the interpolation function, and all points in the complete point cloud data in the geometric three-dimensional model are traversed. For each point, the environmental parameter value is calculated using the interpolation function, and the coordinates of each point and the corresponding environmental parameter value are stored as environmental point cloud data; the environmental point cloud data constitutes the environmental parameter field, which describes the continuous distribution of environmental parameters in the entire space.

[0130] In addition, for environmental parameter fields (such as interpolating multiple environmental parameters at the same time), univariate interpolation can be performed separately and then the results can be merged.

[0131] The threshold range of environmental parameters is preset, and the environmental parameter field of the entire geometric three-dimensional model is divided into initial areas according to the corresponding threshold range; and the initial areas are merged and split so that the environmental parameters within each area change less and the differences between areas are larger.

[0132] Specifically, according to the control requirements of environmental parameters, a threshold range is set for each environmental parameter, such as [T_min, T_max] represents the threshold range of temperature, and the threshold ranges of the three environmental parameters are combined to obtain a comprehensive multi-dimensional threshold range.

[0133] Traverse each point z' in the environmental point cloud data, assign the point z' to the corresponding threshold range according to the environmental parameter value at the point z', and gather all points falling within the same threshold range into an initial region. Calculate the initial region for every two adjacent and The difference in environmental parameters between ;

[0134] ;

[0135] in, The initial area The mean internal temperature, The initial area The mean internal temperature, is the maximum temperature (based on the interior of the spray paint booth), is the minimum temperature, The initial area The mean humidity inside The initial area The mean humidity inside is the maximum value of humidity, is the minimum value of humidity, The initial area The mean internal air velocity, The initial area The mean internal air velocity, is the maximum air velocity, is the minimum value of air flow velocity; is the temperature weight coefficient, is the humidity weight coefficient, is the air velocity weight coefficient, and .

[0136] Set the region merging threshold (5% to 15% of the total variation range of environmental parameters), if Less than , then and Merge into a new region, repeat until no more merges are possible.

[0137] Calculate the size D' (such as volume or maximum side length) of each initial region; set the size threshold S_lit for splitting the initial region (usually 1 / 40 to 1 / 20 of the maximum size of the baking room). If D' is greater than S_lit, split the corresponding initial region; that is, get n preliminary regions.

[0138] The ways to split include:

[0139] For each point in the environmental parameter field, the gradient of each environmental parameter at the point is calculated separately and combined into a comprehensive gradient. The gradient is calculated using the finite difference method or other numerical methods.

[0140] Traverse all points in the entire initial area, calculate the comprehensive gradient at each point, preset the gradient threshold A1, mark all points with comprehensive gradients greater than A1 as high gradient points, aggregate adjacent high gradient points into connected high gradient areas, and for each high gradient area H1, use RANSAC or other robust fitting algorithms to fit a best fitting plane PH; calculate the intersection line LH of each best fitting plane and the initial area, sample m4 points along the intersection line LH, and calculate the sum of the comprehensive gradients SH at these points; the larger the SH, the more likely it is that the plane where the intersection line LH is located is the dividing surface with the largest change in environmental parameters in the area.

[0141] Select the best fitting plane with the largest sum of comprehensive gradients as the optimal segmentation plane, divide the initial region into two along the optimal segmentation plane, generate two sub-regions, and repeat the splitting until the size of all initial regions is smaller than S_lit; that is, n preliminary regions are obtained.

[0142] Furthermore, the representative environmental parameter value vector is obtained by:

[0143] Density clustering sampling is performed on the environmental point cloud data in each preliminary area to obtain a set of M4 sample points, which is recorded as the sample point set X; density clustering can retain the original distribution characteristics of the data.

[0144] The environmental parameters of each sample point in the sample point set X are constructed into a multidimensional vector as the input data vector .

[0145] A sample point is randomly selected from the sample point set X as the first initial center point c1, and is added to the preset initial center point set C1 (initially empty).

[0146] For each sample point in X , calculate its Euclidean distance to the current closest c1 ;

[0147] Calculate each sample point The probability of being selected as the next center point .

[0148] ;in, is the density penalty coefficient, is greater than 0, and The larger the value, the greater the penalty for sparse areas. For sample points The local density value of

[0149] From X according to the probability Randomly select a sample point and add it as a new center point to C1. Repeat until there are K center points in C1. These K center points are used as the initial center point set of the salp group.

[0150] For each input data vector , calculate its weighted distance to all center points , the input data vector The corresponding sample points are classified into the cluster corresponding to the center point with the smallest weighted distance;

[0151] Weighted distance ;in, The input data vector The weight of the dimension, and is an amplification parameter used to penalize outliers and amplify the distance from outliers to the center point. is the input data vector No. dimensions, The center point dimensions.

[0152] For each cluster, recalculate the new centroid of all points in the cluster as the new center point of the corresponding cluster;

[0153] The new centroid is calculated as:

[0154] ;in, To control the distance weight parameter, the closer the point is, the greater its contribution to the centroid; is the current cluster, which contains all the sample points assigned to this cluster; is the old centroid of the current cluster, is the new centroid of the current cluster, For sample points and The weighted distance between .

[0155] The clustering target is defined as the minimum sum of weighted distances from all sample points in the cluster to the centroid, and the process is repeated until the clustering target is reached or the preset number of iterations is reached; at this time, the center point of each cluster is the representative environmental parameter value vector of each cluster.

[0156] For each preliminary region, find the center point of its corresponding cluster, which is the representative environmental parameter value vector of the corresponding preliminary region.

[0157] Furthermore, the directed graph model is constructed in the following ways:

[0158] The representative environmental parameter value vector of each preliminary area is taken as a node to form a node set; for any two nodes in the node set, the similarity between them is calculated.

[0159] The similarity calculation formula is:

[0160] ;in, for and The similarity of For the A vector of representative environmental parameter values ​​for a preliminary region (node), For the A vector of representative environmental parameter values ​​for a preliminary region (node), for and The Euclidean distance between For the The area of ​​the initial region, For the The area of ​​the preliminary region; It is an adjustable parameter.

[0161] Preset similarity threshold , traverse all nodes in the node set, for any two nodes and ,like Greater than or equal to , then in and , connect a directed edge between them; add the directed edge to the directed edge set.

[0162] The weight of each directed edge is defined as the similarity between nodes, and then the directed graph model GL=(NL, EL, WL) is obtained, where NL is the node set, EL is the directed edge set, and WL is the set of directed edge weights.

[0163] Furthermore, the shortest path is obtained by:

[0164] Initially, each node in the directed graph model is regarded as an independent community. All nodes in the directed graph model are traversed, and the module gain after adding any node to other communities is calculated.

[0165] Add any node to the community that can obtain the maximum modular gain, and then rebuild a new network. In the new network, each community is regarded as a new node, and repeat until the modular gain no longer increases; the final network obtained is the community structure.

[0166] The calculation formula of modulus gain is:

[0167] ;in, To make the node Join the community The modulus gain after is the set of remaining nodes (except All nodes except , and To balance the adjustable parameters, is an exponential coefficient used to adjust the intensity of the distance penalty. To represent the node With the community The sum of the weights of the edges between all nodes in the To represent the node The degree of a node (the sum of the weights of all edges connected to it); is the sum of the weights of all edges in the network; For Node With the community The Euclidean distance.

[0168] Define the starting community (select the community where the representative environmental parameter value deviates greatly from the expected value as the starting community), and regard the starting community as a separate subgraph (subnetwork). The nodes in the subgraph are all the preliminary areas in the community. The edges and weights between the nodes are inherited from the directed graph model. In the subgraph, select any two nodes as the starting point and the end point, and use the shortest path algorithm (for example, Dijkstra algorithm) to calculate the shortest path from the starting point to the end point. The shortest path consists of a series of nodes, which represents the optimal adjustment route from the starting point to the end point.

[0169] Repeatedly traverse all pairs of nodes in the subgraph, calculate their shortest paths, record all nodes on the shortest paths, remove duplicate nodes, and obtain a set of adjustment nodes. All the obtained sets of adjustment nodes will be used as priority adjustment targets; the preliminary areas corresponding to these nodes will be used as key areas for environmental control; these areas will be accurately adjusted to maximize the improvement of the overall environmental conditions.

[0170] Furthermore, the optimal control parameters are solved by:

[0171] Construct the control objective function, the formula of the control objective function is:

[0172] ; among them, among them, is the control parameter vector, which includes various adjustable parameters affecting the preliminary regional environment, control parameters of the heating and cooling system, control parameters of the ventilation system, etc.; It is the first A function of an environmental parameter (such as temperature) that describes the relationship between this parameter and the control parameter vector The mapping relationship is obtained through data fitting or physical model; It is The weight coefficient of each environmental parameter reflects the importance of the parameter; For the initial area Standard quantity of environmental parameters; is the control parameter vector The control objective function value.

[0173] For the control parameter vector , define the initial solution ; Set a higher initial temperature , and a lower termination temperature ; Define the temperature drop curve as: ,in, is the cooling rate , is the number of iterations, for The current temperature at iteration .

[0174] At the current temperature, the control parameter vector Perturbations to generate new solutions , calculate the new solution based on the control objective function The corresponding control objective function value ;like , then accept the new solution, that is ;like , then with probability Accept the new solution and repeat until the preset maximum number of iterations is reached. During this process, the current temperature is reduced according to the set temperature drop curve.

[0175] Probability ,in, is a constant parameter and is greater than 0; is the maximum number of iterations.

[0176] Until the current temperature drops to the termination temperature Or the control objective function value is terminated when there is no change in P consecutive iterations; at this time That is the optimal control parameter vector; devectorize the optimal control parameter vector to obtain the optimal control parameter.

[0177] This embodiment realizes high-precision perception and comprehensive analysis of the internal environment of the paint booth. Through high-precision three-dimensional reconstruction of the internal environment of the paint booth, it can perform adaptive regional division according to the distribution of actual environmental data, so that the division result is more in line with the actual environmental distribution; and effectively capture the correlation between regions; through community discovery and shortest path planning, the adjustment order and key targets are scientifically determined, resource waste is avoided, and control efficiency is improved. In addition, comprehensive optimization control of multiple environmental parameters is realized, which greatly improves the overall control accuracy. The control strategy takes into account the mutual influence between regions and realizes overall coordinated optimization, making the internal environment of the paint booth more balanced and stable; this refined and intelligent control method not only improves the quality of painting and production efficiency, but also reduces energy consumption and reduces environmental pollution.

[0178] Example 2

[0179] See also Figure 2 As shown, the part not described in detail in this embodiment is described in Example 1, which provides an intelligent interconnection control system for a spray paint baking room, including:

[0180] The area division module is used to divide the interior of the spray paint room into n preliminary areas;

[0181] A feature extraction module, for obtaining a representative environmental parameter value vector for each preliminary area using a salp swarm algorithm;

[0182] The model building module constructs a directed graph model based on the representative environmental parameter value vectors of each region cluster as nodes;

[0183] The path planning module is used to plan the shortest path in the directed graph model and take each node on the shortest path as a priority adjustment target;

[0184] The control module is used to use an optimization algorithm to solve the optimal control parameters of the preliminary area corresponding to the priority adjustment target; send the optimal control parameters to the execution device of the corresponding preliminary area; and connect each module by wired and / or wireless means to realize data transmission between modules.

[0185] Example 3

[0186] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the operation mode of the above-mentioned intelligent interconnection control method for a paint spraying and baking room is implemented.

[0187] Since the electronic device introduced in this embodiment is an electronic device used to implement a method for intelligent interconnection control of a spray paint baking room in the embodiment of this application, based on the method for intelligent interconnection control of a spray paint baking room introduced in the embodiment of this application, a person skilled in the art can understand the specific implementation of the electronic device of this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application is not described in detail here. As long as a person skilled in the art implements the electronic device used in the method for intelligent interconnection control of a spray paint baking room in the embodiment of this application, it belongs to the scope of protection of this application.

[0188] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0189] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for intelligent interconnection control of a paint spraying room, characterized in that: include: S1, dividing the interior of the spray paint room into n preliminary areas; S2, using the salp swarm algorithm to obtain a representative environmental parameter value vector for each preliminary area; S3, constructing a directed graph model based on taking the representative environmental parameter value vectors of each region cluster as nodes; S4. Plan the shortest path in the directed graph model, and take each node on the shortest path as a priority adjustment target; S5. Use an optimization algorithm to solve the optimal control parameters of the preliminary area corresponding to the priority adjustment target; and send the optimal control parameters to the execution device of the corresponding preliminary area.

2. The method for intelligent interconnection control of a paint spraying room according to claim 1 is characterized in that: The method of dividing the interior of the spray paint booth into n preliminary areas includes: N structured light scanning positions are set inside the spray paint booth, and structured light is used to scan the inside of the spray paint booth at the structured light scanning positions to obtain original point cloud data; the original point cloud data is preprocessed to obtain complete point cloud data; Using the complete point cloud data, the geometric shape inside the spray paint booth is modeled to obtain a three-dimensional geometric model of the interior of the spray paint booth; A sample point set S1 is obtained inside the spray paint room, where each sample point s' contains its spatial coordinates and corresponding environmental parameters, including temperature, humidity and air flow velocity; Based on the sample point set S1, the parameters of the radial basis function are fixed to obtain the interpolation function, and all points in the complete point cloud data in the geometric three-dimensional model are traversed. For each point, the environmental parameter value is calculated using the interpolation function, and the coordinates of each point and the corresponding environmental parameter value are stored as environmental point cloud data; the environmental point cloud data constitutes the environmental parameter field; The threshold range of the environmental parameters is preset, and the environmental parameter field of the entire geometric three-dimensional model is divided into initial regions according to the corresponding threshold range; and the initial regions are merged and split to obtain n preliminary regions.

3. The intelligent interconnection control method for a paint spraying room according to claim 2 is characterized in that: The division method of the initial area includes: Traverse each point z' in the environmental point cloud data, assign the point z' to the corresponding threshold range according to the environmental parameter value at the point z', and gather all the points that fall within the same threshold range into an initial area; The method of merging and splitting the initial area includes: Calculate every two adjacent initial regions and The difference in environmental parameters between ; ； in, The initial area The mean internal temperature, The initial area The mean internal temperature, is the maximum temperature, is the minimum temperature, The initial area The mean humidity inside The initial area The mean humidity inside is the maximum value of humidity, is the minimum value of humidity, The initial area The mean internal air velocity, The initial area The mean internal air velocity, is the maximum air velocity, is the minimum value of air flow velocity; is the temperature weight coefficient, is the humidity weight coefficient, is the air flow velocity weight coefficient; Set the region merging threshold ,like Less than , then and Merge into a new region, repeat until no more merging is possible; Calculate the size D' of each initial region; set the size threshold S_lit for splitting the initial region. If D' is greater than S_lit, split the corresponding initial region; The ways to split include: For each point in the environmental parameter field, calculate the gradient of each environmental parameter at the point respectively, and merge them into a comprehensive gradient; traverse all points in the entire initial area, calculate the comprehensive gradient at each point, preset the gradient threshold A1, mark all points with comprehensive gradients greater than A1 as high gradient points, and aggregate adjacent high gradient points into connected high gradient areas. For each high gradient area H1, fit a best fitting plane PH; calculate the intersection line LH of each best fitting plane and the initial area, sample m4 points along the intersection line LH, and calculate the sum of the comprehensive gradients SH at these points; select the best fitting plane with the largest sum of comprehensive gradients as the optimal segmentation plane; divide the initial area into two along the optimal segmentation plane to generate two sub-areas, and repeat the splitting until the size of all initial areas is less than S_lit; that is, n preliminary areas are obtained.

4. The method for intelligent interconnection control of a paint spraying and baking room according to claim 3 is characterized in that: The method of preprocessing the original point cloud data includes: For each point in the original point cloud data , calculate the normal vectors of all points in its neighborhood sphere Np(r); where r is the radius of the neighborhood sphere; for points in Np(r) , calculate its smoothed covariance matrix ; ;in, is the total number of points in the neighborhood ball Np(r), is the first points; For point The attribute vector of is the first The attribute vector of each point; is a comprehensive attribute function, is the mean vector of all points in Np(r); ;in, is the balance parameter, is the distance attenuation bandwidth; is the attribute attenuation bandwidth; The smoothed covariance matrix Perform eigenvalue decomposition to obtain the corresponding eigenvectors and eigenvalues, and find the smallest eigenvalue, which is the normal vector of all points in Np(r) ;calculate Point and its neighboring points Np(r) The normal vector difference angle ;in, Yes The normal vector of Preset angle threshold ,like , then point is a noise point, and removes it from Np(r), updates the points in Np(r), and recalculates Normal vector of a point , until no more points are removed within Np(r), and the preliminary point cloud data is obtained; Based on the preliminary point cloud data, the hole area is obtained, the surface grid structure of the hole area is constructed, and new point cloud data is generated based on the surface grid structure; the new point cloud data and the preliminary point cloud data are aggregated to obtain complete point cloud data; The method of obtaining the hole area based on the preliminary point cloud data includes: For each point in the preliminary point cloud data , calculate its Euclidean distance to other points, and record the distance points The nearest k points, which form the point The k nearest neighbor point set N_k(p); Point-based and each point in its k-nearest neighbor point set N_k(p) , calculation point Point cloud density ; ;in, For point With point The Euclidean distance between is the weight function; is the index parameter; Weight function ;in, For the The weight coefficient of the geometric statistical features, Based on point and Point No. Geometric statistical characteristics; Preset density threshold ,like , then the marking point For each point marked as a sparse point, find its nearest point that is not marked as a sparse point. If the distance between the two is greater than the preset clustering threshold R, the point Mark them as hole points, and aggregate all adjacent hole points to get the hole area.

5. The intelligent interconnection control method for a paint spraying and baking room according to claim 4 is characterized in that: The surface grid structure is constructed in the following manner: For each hole region, find the nearest m1 points that are not marked as sparse points, and use these points as the initial neighborhood point set RN of the hole region; calculate the closest distance d1 between each point in RN and the hole region, and if d1 is less than the preset neighborhood threshold R1, add it to RN, and repeat until RN no longer increases; obtain the neighborhood point set of the hole region; Preset sphere radius r1 and accuracy threshold angle ; Randomly select a seed point x from the neighborhood point set as the initial sphere center, and construct a sphere B(x, r1) with r1 as the radius; Find a set S of all points on the ball B(x, r1), S contains at least 3 points, then perform local surface fitting on the points in S to obtain a local surface patch F, select a new sphere center point x1 from S, and construct a new sphere B(x1, r); and the new sphere center point q satisfies the preset constraints; Constraints include overlap constraints and angle constraints; The overlap constraint includes: the new ball B(x1, r) has an overlapping area with the ball B(x, r1), and the number of points in the overlapping area is greater than the preset overlap threshold M1; The angle constraints include: the angle between the new ball B(x1, r) and the normal vector of the local surface patch F is less than the accuracy threshold angle , and the curvature change rate at the new sphere center point x1 is less than the preset curvature threshold M2; Repeatedly construct a new local surface patch on the new ball B(x1, r) and select a new sphere center point until no new sphere center point that meets the constraints can be found. Select a new seed point from the remaining neighborhood point set and repeatedly construct a new local surface patch F'. Topologically connect the new local surface patch F' with the existing local surface patch F to obtain the surface mesh structure G of the hole area. The topological connection methods include: Calculate the distance field of F and F' in the overlapping area R2 of the two ,in, is a point within the overlapping region R2; and based on the distance field Find the best matching position of the two F and F' in the overlapping area, and at the best matching position, splice F and F' together through interpolation method; that is, obtain the surface mesh structure G of the hole area; Distance Field ;in, is a point on F, is a point on F'; After calculating the distance fields of all points in the overlapping area R2, find the surface composed of points whose distance fields are equal to 0, that is, the best matching position of F and F'.

6. The method for controlling a paint spray booth in intelligent interconnection according to claim 5, characterized in that: The representative environmental parameter value vector is obtained by: Density cluster sampling is performed on the environmental point cloud data in each preliminary area to obtain a set of M4 sample points, recorded as sample point set X; the environmental parameters of each sample point in the sample point set X are constructed into a multidimensional vector as the input data vector ; A sample point is randomly selected from the sample point set X as the first initial center point c1, and added to the preset initial center point set C1; for each sample point in X , calculate its Euclidean distance to the current closest c1 ; and calculate each sample point The probability of being selected as the next center point ;in, is the density penalty coefficient, For sample points The local density value of From X according to the probability Randomly select a sample point and add it as a new center point to C1. Repeat until there are K center points in C1. These K center points are used as the initial center point set of the salp group. For each input data vector , calculate its weighted distance to all center points , the input data vector The corresponding sample points are classified into the cluster corresponding to the center point with the smallest weighted distance; Weighted distance ;in, The input data vector The weight of the dimension, and is the amplification parameter, is the input data vector No. dimensions, The center point Dimensions; For each cluster, recalculate the new centroid of all points in the cluster as the new center point of the corresponding cluster; The new centroid is calculated as: ;in, To control the distance weight parameter; is the current cluster; is the old centroid of the current cluster, is the new centroid of the current cluster, For sample points and The weighted distance between The clustering target is defined as the minimum sum of weighted distances from all sample points in the cluster to the centroid, and the process is repeated until the clustering target is reached or the preset number of iterations is reached; at this time, the center point of each cluster is the representative environmental parameter value vector of each cluster; for each preliminary area, the center point of its corresponding cluster is found, which is the representative environmental parameter value vector of the corresponding preliminary area.

7. The method for controlling a paint spray booth in an intelligent interconnection manner according to claim 6, characterized in that: The construction method of the directed graph model includes: The representative environmental parameter value vector of each preliminary area is taken as a node to form a node set; for any two nodes in the node set, the similarity between them is calculated; The similarity calculation formula is: ;in, for and The similarity of For the A vector of representative environmental parameter values ​​for the preliminary regions, For the A vector of representative environmental parameter values ​​for the preliminary regions, for and The Euclidean distance between For the The area of ​​the initial region, For the The area of ​​the preliminary region; is an adjustable parameter; Preset similarity threshold , traverse all nodes in the node set, for any two nodes and ,like Greater than or equal to , then in and , connect a directed edge between them; add the directed edge to the directed edge set; The weight of each directed edge is defined as the similarity between nodes, and then the directed graph model GL=(NL, EL, WL) is obtained, where NL is the node set, EL is the directed edge set, and WL is the set of directed edge weights.

8. The method for intelligent interconnection control of a paint spraying and baking room according to claim 7 is characterized in that: The shortest path acquisition method includes: Initially, each node in the directed graph model is regarded as an independent community. All nodes in the directed graph model are traversed, and the module gain after adding any node to other communities is calculated. Add any node to the community that can obtain the maximum modulus gain, and then rebuild a new network. In the new network, each community is regarded as a new node. Repeat until the modulus gain no longer increases. The final network is the community structure. The calculation formula of modulus gain is: ;in, To make the node Join the community The modulus gain after is the set representing the remaining nodes; , and To balance the adjustable parameters, is the exponential coefficient, To represent the node With the community The sum of the weights of the edges between all nodes in the To represent the node degree; is the sum of the weights of all edges in the network; For Node With the community The Euclidean distance of Define the starting community and regard it as a separate subgraph. The nodes in the subgraph are all the preliminary areas in the community. The edges and weights between the nodes are inherited from the directed graph model. In the subgraph, select any two nodes as the starting point and the end point, and use the shortest path algorithm to calculate the preliminary shortest path from the starting point to the end point; repeatedly traverse all pairs of nodes in the subgraph, calculate their preliminary shortest paths, record all nodes on the preliminary shortest paths, remove duplicate nodes, and obtain the adjusted node set, which is the final shortest path.

9. The method for controlling a paint spray booth in intelligent interconnection according to claim 8, characterized in that: The method for solving the optimal control parameters includes: Construct the control objective function, the formula of the control objective function is: ;in, is the control parameter vector; It is the first A function of environmental parameters; It is The weight coefficient of each environmental parameter; For the initial area Standard quantity of environmental parameters; is the control parameter vector The control objective function value of For the control parameter vector , define the initial solution ; Set an initial temperature , and a termination temperature ; Define the temperature drop curve as: ,in, is the cooling rate, is the number of iterations, for The current temperature at the iteration; At the current temperature, the control parameter vector Perturbations to generate new solutions , calculate the new solution based on the control objective function The corresponding control objective function value ;like , then accept the new solution, that is ;like , then with probability Accept the new solution and repeat until the preset maximum number of iterations is reached. During this process, the current temperature is reduced according to the set temperature drop curve; Probability ,in, is a constant parameter; is the maximum number of iterations; Until the current temperature drops to the termination temperature Or the control objective function value is terminated when there is no change in P consecutive iterations; That is the optimal control parameter vector; devectorize the optimal control parameter vector to obtain the optimal control parameter.

10. A paint spraying room intelligent interconnection control system, which is used to implement the paint spraying room intelligent interconnection control method according to any one of claims 1 to 9, characterized in that: include: The area division module is used to divide the interior of the spray paint room into n preliminary areas; A feature extraction module, for obtaining a representative environmental parameter value vector for each preliminary area using a salp swarm algorithm; The model building module constructs a directed graph model based on the representative environmental parameter value vectors of each region cluster as nodes; The path planning module is used to plan the shortest path in the directed graph model and take each node on the shortest path as a priority adjustment target; The control module is used to use an optimization algorithm to solve the optimal control parameters of the preliminary area corresponding to the priority adjustment target; send the optimal control parameters to the execution device of the corresponding preliminary area; and each module is connected by wired and / or wireless means.

Images