A scheduling modeling and intelligent scheduling method for multi-wall parallel construction

Abstract

The application provides a multi-wall parallel construction scheduling modeling and intelligent scheduling method, and belongs to the technical field of intelligent scheduling, which considers the heterogeneity of construction period calculation, establishes a scheduling optimization model with the minimum construction period as the target and the number of each type of work as the decision variable, designs an intelligent scheduling algorithm for solving the optimization model based on a particle swarm algorithm, formulates special two-layer coding and decoding rules, uses the difference between the local historical optimum and the global historical optimum in a negative feedback mode to update the particle speed based on nonlinear time-varying inertia parameters, and further realizes position updating, and finally outputs the global optimal particle through multiple iterations to form a personnel scheduling scheme for multi-wall parallel construction. The application has the advantages of strong speciality, strong global search capability, fast search speed and the like, can provide a reasonable multi-wall construction scheduling scheme for construction progress management, maximally shortens the construction period under the condition of ensuring the construction quality, and reduces the construction cost from the perspective of human resource investment.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent scheduling, and particularly relates to a scheduling modeling and intelligent scheduling method for multi-wall parallel construction. BACKGROUND

[0002] As a low-carbon clean energy, nuclear power has the advantages of high density, cleanliness, low carbon, long-term stable operation, etc., and has played a supporting role in power supply as a base load energy. However, in the process of nuclear power construction, especially in the process of civil engineering, the matching mode of construction tasks and "man-machine-material-method-environment-measurement" resources mostly depends on the rough estimation and experience-based decision of relevant personnel, which is limited by the limitations of the human brain in processing speed, complex logic, parallel computing, etc., and the effect of achieving the goal is often not satisfactory. Usually, only partial scale local suboptimal matching can be achieved, and the optimization of the total task duration and quality of nuclear power construction cannot be guaranteed.

[0003] A series of existing technical solutions such as patent application number "202510168349.4" provide related technologies for the construction of nuclear power construction in the civil engineering of multi-wall construction scene, which mainly focus on the solution of general optimization problems. Although such existing technical solutions have some depth in the construction scene, they do not take into account the heterogeneity of construction tasks in different process duration calculations, which leads to the establishment of algorithms that are acceptable in general fields but need to be improved in specialized fields. SUMMARY

[0004] The purpose of the present application is to provide a scheduling modeling and intelligent scheduling method for multi-wall parallel construction, which starts from the actual process of a certain nuclear power project, analyzes the differences in duration calculation in multiple processes in detail, summarizes three types of duration calculation methods, and then expresses them in mathematical models to establish a special mathematical optimization model. Then, according to the differences in duration calculation in different processes, an adaptive particle swarm algorithm is developed, and a nonlinear time-varying inertia parameter is integrated into the particle velocity update to enhance the global optimization ability of the algorithm.

[0005] The application uses the following technical solutions.

[0006] A scheduling modeling and intelligent scheduling method for multi-wall parallel construction, comprising:

[0007] Step 1: Perform multi-wall parallel construction resource scheduling modeling, that is, minimize the duration as the target, integrate construction requirements and multi-type duration calculation characteristics, and establish a multi-wall parallel construction resource scheduling model for nuclear power engineering civil engineering scene;

[0008] Step 2: Perform parameter initialization, that is, determine the related parameters of the particle swarm algorithm, including the particle swarm size, inertia parameter, learning factor, and maximum number of iterations;

[0009] Step 3: Perform encoding decoding scheme design and initialization, that is, design an encoding scheme containing multiple types of durations according to the characteristics of the scheduling model, realize the mapping of the scheduling scheme to the particles; and design an applicable decoding scheme to realize the mapping of the particles to the scheduling scheme, complete the particle fitness calculation, and on this basis, generate a primary particle group according to the particle size;

[0010] Step 4: Perform global historical optimal solution and individual historical optimal position update, that is, decode each particle of the current particle group according to the decoding rule to obtain the completion duration as the fitness, and update the global historical optimal solution and the individual historical optimal position;

[0011] Step 5: Perform particle update, that is, update the speed and position of the particle based on the designed nonlinear time-varying inertia parameter, the global historical optimal solution and the individual historical optimal position, and the position error feedback drives the speed and position update of the particle;

[0012] Step 6: Perform position and speed projection, that is, process the updated particle in position and speed according to the coding requirements in the coding rule;

[0013] Step 7: Perform termination judgment, that is, judge whether the maximum iteration number is reached, if yes, output the current global optimal solution as the optimal scheduling scheme, otherwise, jump to step 4 for the next iteration.

[0014] Preferably, step 1 specifically comprises:

[0015] In the multi-wall parallel construction of nuclear power construction, there are face walls, each wall has processes, and represents the process of the wall , , The construction time of the wall , are positive integers;

[0016] There are teams, that is, there are types of work, and each team has personnel, and the wall process requires personnel , are positive integers;

[0017] Let represent end at time, then the end time of the wall , is the start time , denotes the upper bound of the set of multi-wall parallel construction durations, the meaning of which is to search from the first time period until at which point the output is such that the moment; the optimization objective of minimizing the duration can be expressed as minimizing the latest completion time, i.e., as shown in the following equation (1.1):

[0018] ;

[0019] grouping the construction processes according to the duration calculation method, representing a fixed duration process, representing a performance duration process, representing a fuzzy duration process; for , the duration is ; for , let the performance of be ; for , the work limit is , and the duration below or above the limit is and , so that the duration expression shown in the following equation (1.2) is obtained: :

[0020] .

[0021] Preferably, step 1 further comprises:

[0022] Based on equation (1.1) and equation (1.2), other constraints are given:

[0023] Completion time limit: each process has only one completion time, which is described as shown in the following equation (1.3):

[0024] ;

[0025] Immediate constraint: the process constraint and the requirement of continuous construction are represented by using the difference between the completion time of the previous process and the duration of the subsequent task, i.e., as shown in the following equation (1.4):

[0026] ;

[0027] Resource upper limit constraint: define a binary variable as shown in the following equation (1.5):

[0028] ;

[0029] This constraint actually means: for a given , express time To determine if the process is in progress, sum the results for all walls and processes. In all wall finishing processes, by adding You can get The total amount of resources required at any given time.

[0030] Preferably, step 1 further includes:

[0031] The overall optimization problem is obtained as shown in Equation (1.6):

[0032] ;

[0033] Next, we use formula (1.6) The objective function is simplified to the optimization problem shown in Equation 1.7:

[0034] ;

[0035] in Indicates obedience, This indicates that formula (1.2) is being referenced.

[0036] Preferably, step 2 specifically includes:

[0037] The particle swarm is set to have 100 particles, with a maximum inertial parameter of 0.9 and a minimum inertial parameter that is randomly obtained. The individual and swarm learning factors are set to 0.8 and 0.7, respectively, and the maximum number of iterations is 300.

[0038] Preferably, step 3 specifically includes:

[0039] The encoding rule is to use two sets of codes for each particle:

[0040] The first set of codes is priority-based task coding, by... Composed of real numbers, for position value This expresses the construction task. The priority value of a certain process;

[0041] The second set of codes is... indivual The number between the numbers represents the proportion of the number of people assigned to that task process out of the total number of people.

[0042] Preferably, step 3 further comprises:

[0043] The decoding step is: first, two new sets of codes are generated according to the encoding, denoted as the third set of codes and the fourth set of codes, wherein the third set of codes is generated according to the first set of codes to sort the task-process codes according to the priority, and the generation rule is: according to the first set of codes, sort from large to small, generate the third set of codes according to the task number according to the sorting value, that is, the task code with the largest priority is located at the position of the third set of codes, that is, , while in the segment not exceeding the position, that is, the position , the number of times represents the process number of the task code , that is, the position represents the process of the task , and the position of the fourth set of codes is corresponding to the proportion of the number of people; , while the position of the fourth set of codes is corresponding to the proportion of the number of people;

[0044] After obtaining the third set of codes and the fourth set of codes, the decoding operation can be started, that is, the start time and the completion time of each task and each process are calculated;

[0045] Then, according to the particle swarm size, generate the particle swarm based on the coding rule, that is, in addition to using the first set of codes and the second set of codes for position initialization according to the coding rule, the speed of each position also needs to be initialized. For the first set of codes, the speed is randomly generated; for the second set of codes, the speed is randomly generated in the interval , while the speed of the process corresponding to the fixed duration and the fuzzy duration is initialized to 0.

[0046] Preferably, in step 3, the decoding operation comprises the following steps:

[0047] Step 3-1: after the particles are coded, new codes are obtained; initialize the available resource number sequence at each time;

[0048] Step 3-2: obtain the task, process, number of personnel and duration corresponding to each position according to the new code; preliminarily obtain the construction sequence of the wall;

[0049] Step 3-3: calculate the start and end time of each process of the wall whose construction sequence is ;

[0050] Step 3-4: according to the duration expression , calculate the process ​​​start time of the start time and the end time ;

[0051] Step 3-5: if and , the start time of the first process of the wall is taken as the start time, the available resource sequence is searched to find the start time of the resource required by all processes of the wall , if and , , then step 3-4 is executed , if and , , then step 3-4 is executed , if and , , then step 3-4 is executed ;

[0052] Step 3-6: according to the start time , the end time and the required resource number, the available resource sequence at each time is updated;

[0053] Step 3-7: if the start and end time calculation of all processes is not completed, go to step 3-4 to execute; if the start and end time calculation of all processes is completed and the start and end time calculation of all processes of all walls is completed, the decoding operation is ended; if the start and end time calculation of all processes is completed and the start and end time calculation of all processes of all walls is not completed, go to step 3-3 to execute.

[0054] Preferably, step 4 specifically comprises:

[0055] According to the end time output by the decoding operation, the maximum end time is taken as the completion duration, which is taken as the fitness, and then the global historical optimal solution and the individual historical optimal position are determined according to the optimization target.

[0056] Preferably, step 5 specifically comprises:

[0057] The total number of iterations is set to , the current is the th iteration, the speed and position of the th particle at the th time are and , the global historical optimal solution is , and the individual historical optimal position of each particle is , thus, the particle updating mode is shown in the following formula (1.8):

[0058]

[0059] wherein, are individual and group learning factors, used for describing self-cognition and social cognition ability of the particle, is a random number in the range of [0, 1], are maximum and minimum inertia parameters respectively, is a time-varying inertia parameter at the moment t. is a time-varying inertia parameter at the moment t.

[0060] Preferably, the step 6 specifically comprises:

[0061] After each particle updating, the projection of the speed and the position is carried out, and the rule is:

[0062] (1) if the duration calculation mode of the process is fixed duration or fuzzy duration, then the speed projection corresponding to the 2nd dimension code is 0, and the position is 1;

[0063] (2) if it is a process corresponding to variable duration, then the position projection corresponding to the 2nd dimension is within [0, 1], that is, if it is lower than 0.1, then it is assigned as 0.1, and if it is higher than 0.9, then it is assigned as 0.9;

[0064] Thus, the particle updating is completed.

[0065] The beneficial effects of the present application are that, compared with the prior art, the technical effects of the present application include:

[0066] The method of the present application comprises the following steps: considering the duration calculation heterogeneity, establishing a scheduling optimization model taking the minimum duration as the target and the number of each type of work as the decision variable; based on the particle swarm algorithm, designing an intelligent scheduling algorithm for solving the optimization model, formulating a special two-layer coding and decoding rule, and realizing one-to-one mapping of the scheduling scheme and the particle; based on the nonlinear time-varying inertia parameter, using the difference between the local historical optimum and the global historical optimum to update the particle speed in a negative feedback manner, and further realizing the position updating; finally, through multiple iterations, the global optimal particle is output, and based on the coding mode, the personnel scheduling scheme of multi-wall parallel construction is formed. The present application has the advantages of strong speciality, strong global search ability, fast search speed and the like, can provide a reasonable multi-wall construction scheduling scheme for construction progress management, so as to shorten the construction duration to the maximum extent under the condition of ensuring the construction quality, and reduce the construction cost from the perspective of human resource investment. BRIEF DESCRIPTION OF DRAWINGS

[0067] Figure 1 ​​is the overall flow chart of the scheduling modeling and intelligent scheduling method of the multi-wall parallel construction in the application;

[0068] Figure 2 is the coding example diagram of the scheduling scheme under the multi-wall parallel construction in the application;

[0069] Figure 3 is the coding conversion example diagram in the decoding in the application;

[0070] Figure 4 is the flow chart of the decoding operation in the application. DETAILED DESCRIPTION

[0071] In nuclear power construction, especially in the process of civil engineering, the matching mode of construction tasks and "man-machine-material-method-environment-measurement" resources mostly depends on the rough estimation and experience-based decision of relevant personnel, and is subject to the limitations of the human brain in processing speed, complex logic, parallel computing and the like, and the effect of achieving the target is often not satisfactory. Usually, only partial scale local suboptimal matching can be achieved, and the optimization of the total task duration and quality of nuclear power construction cannot be guaranteed.

[0072] In view of the above problems, the application focuses on a common scene in the civil engineering of nuclear power construction, multi-wall parallel construction, and proposes to establish a parallel construction model with variable duration for minimizing the duration, and to propose an encoding and decoding and intelligent scheduling method based on a particle swarm algorithm. The present application is similar to the prior art, and its main concern is the generality of the model and the updating strategy of the particles in the particle swarm algorithm under the diversity index, while the present application focuses on the multi-wall parallel construction scene, establishes an actual model with variable duration, and exceeds the scope of the similar prior art, and has scene-specific features. In terms of algorithm, the method disclosed in the present application is customized for the special model, and the special model proposed is adapted from the aspects of encoding and decoding.

[0073] In order to make the purpose, technical scheme and advantages of the present application clearer, the technical scheme of the present application will be expressed clearly and completely in combination with the drawings in the embodiments of the present application. The embodiments expressed in the present application are only partial embodiments of the present application, not all embodiments. According to the spirit of the present application, other embodiments obtained by those skilled in the art without making creative efforts fall within the protection scope of the present application.

[0074] As shown in Figure 1 , the scheduling modeling and intelligent scheduling method of the multi-wall parallel construction, comprises:

[0075] The present application considers the heterogeneity of construction period calculation, establishes a scheduling optimization model with the minimum construction period as the target and the number of each type of work as the decision variable, and then fuses the customized encoding and decoding scheme, time-varying inertia parameter, and customized particle swarm algorithm to generate a more optimal scheduling scheme faster than artificial experience scheduling.

[0076] Step 1: Multi-wall parallel construction resource scheduling modeling is performed, that is, a multi-wall parallel construction resource scheduling model of a nuclear power engineering civil engineering scene is established by fusing construction requirements and multi-type construction period calculation characteristics with the minimum construction period as the target.

[0077] In the preferred but non-limiting embodiments of the present application, step 1 specifically includes:

[0078] For example, in the multi-wall parallel construction of a certain plant of a certain nuclear power project, there are multiple walls to be constructed on a certain floor. The construction task has the minimum construction period as the target. The construction of each wall has the following procedures: positioning and setting out, steel binding, embedded part installation, formwork erection, concrete pouring, concrete curing, and formwork removal. Among them, the construction period calculation is divided into three types: fixed construction period, work efficiency construction period, and fuzzy construction period. The fixed construction period means that the personnel is fixed and the construction period is fixed. In the work efficiency construction period, the construction period is related to the work efficiency and the number of workers, and the approximate relationship is construction period = engineering quantity / (work efficiency x number of workers). The fuzzy construction period has a work quantity limit, and whether it is exceeded corresponds to two types of personnel configuration and construction period. The construction period, work quantity, and work efficiency calculation of the six procedures are shown in Table 1:

[0079] Table 1

[0080]

[0081] In addition to the above construction period calculation, the entire multi-wall parallel construction scene has the following constraints:

[0082] 1. Each procedure needs to arrange personnel, and the corresponding personnel can only perform one construction at the same time;

[0083] 2. Each procedure will only be completed at one time;

[0084] 3. Immediate constraint: procedure constraint, the next procedure can only be performed after the previous procedure is completed; and the interval time between procedures is 0, that is, the next procedure needs to be performed immediately after the completion of the previous procedure;

[0085] 4. Resource upper limit: the personnel of each team is fixed, and the arranged personnel per day cannot exceed the upper limit of the team personnel.

[0086] Here, the mathematical modeling form of the above problem is given:

[0087] In the multi-wall parallel construction of nuclear power construction wall, each wall has In one embodiment, the method comprises the following steps: The wall construction process , The construction time is defined as , are positive integers;

[0088] There are classes, that is, there are types of work, and each class has people, and the wall construction process requires class personnel , are positive integers;

[0089] Let represent the end time at , then The end time is , The start time is , represents the upper bound of the set wall parallel construction period, The meaning is to search from the first time period until , at which time the time will be output; thus, the optimization goal of minimizing the construction period can be expressed as minimizing the latest completion time, that is, as shown in the following formula (1.1):

[0090] ;

[0091] According to the construction process, the construction process is grouped according to the construction period calculation method, representing the fixed construction period process, representing the efficiency construction period process, representing the fuzzy construction period process; thus, for , The construction period is ; for , let The efficiency is , and the total workload is ; for , the workload limit is , and the construction period below or above the limit is and , thus, the construction period expression shown in the following formula (1.2) can be obtained: :

[0092] ​ .

[0093] In the preferred but non-limiting embodiment of the present application, step 1 specifically further comprises:

[0094] Based on equation (1.1) and equation (1.2), other constraints can be given as follows:

[0095] Completion time limit: each process has only one completion time, which can be described as follows:

[0096] ;

[0097] Immediate constraint: since the completion time of is represented here, the difference between the completion time of the continuous process is used to represent the process constraint (immediate constraint) and the requirement of flow construction, that is, as shown in the following equation (1.4):

[0098] ;

[0099] Resource upper limit constraint: the occupation of the corresponding resource is limited within the execution time of any task. Define the binary variable as shown in the following equation (1.5):

[0100] ;

[0101] The actual meaning of this constraint is: for a given , represents whether is executing at , and then summing all wall surfaces and processes, the total number of wall surface processes working at can be obtained. By adding , the total number of resources required at can be obtained.

[0102] In the preferred but non-limiting embodiment of the present application, step 1 specifically further comprises:

[0103] Therefore, the overall optimization problem (objective function) shown in the following equation (1.6) can be obtained:

[0104] ;

[0105] Then, by using to simplify the objective function, the optimization problem shown in the following equation (1.7) can be obtained:

[0106] ;

[0107] wherein denotes subject to, denotes reference to formula (1.2).

[0108] It is very obvious that the optimization problem is difficult to solve due to the existence of (1.2), that is, the different calculation mode of the construction period, and the high degree of nonlinearity.

[0109] Step 2: parameter initialization, that is, determining the related parameters of the particle swarm algorithm, including the particle swarm size, inertia parameter, learning factor and maximum iteration number, etc.

[0110] Consider that 12 walls of a floor of a nuclear power plant need to be constructed, due to space limitations, the total construction amount of each wall process and the responsible team are given in Table 2. In this case, the measuring and placing team has 8 people, the steel reinforcement team has 70 people, the embedded part team has 27 people, the carpentry team has 60 people, and the comprehensive team has 16 people.

[0111] Table 2

[0112]

[0113] Therefore, the particle swarm algorithm is introduced to solve it by adaptation and modification. The related steps are described as follows:

[0114] In the preferred but non-limiting embodiments of the application, step 2 specifically includes:

[0115] Set 100 particles in the particle swarm, the maximum inertia parameter is 0.9, the minimum inertia parameter is randomly obtained, the individual and group learning factors are 0.8 and 0.7 respectively, and the maximum iteration number is 300 times.

[0116] Step 3: design and initialize the encoding and decoding scheme, that is, design an encoding scheme containing multiple types of construction period according to the characteristics of the scheduling model, realize the mapping of the scheduling scheme to the particle, and design a suitable decoding scheme to realize the mapping of the particle to the scheduling scheme, complete the particle fitness calculation, and on this basis, generate the initial particle swarm according to the particle size;

[0117] In the preferred but non-limiting embodiments of the application, step 3 specifically includes:

[0118] Encoding refers to the conversion of a specific scheduling plan into a specific format of expression form so that the algorithm can recognize and process it. Decoding is to restore the encoded information after algorithm operation to the actual scheduling plan, so as to obtain a directly applicable solution result.

[0119] The encoding and decoding mode for the multi-wall parallel scheduling problem is given here. The encoding rule is to use two groups of encoding for each particle:

[0120] The first group of codes is priority-based task coding, which is generated by The second group of codes is composed of The value of represents the priority value of the construction task of a certain process;

[0121] The second group of codes is composed of The second group of codes is composed of The second group of codes is composed of Figure 2 Note that due to different construction period calculation modes, the number of people is given after the total work amount is given, so the codes corresponding to the process positioning and layout (process number 1), concrete pouring (process number 5), and template removal (process number 6) are 1. Taking the construction of a 2-face wall as an example, the code display is shown in .

[0122] In a preferred but non-limiting embodiment of the present application, step 3 further comprises:

[0123] Next, the decoding rules are given: the main steps of decoding are: first, generate two new groups of codes according to the codes, denoted as the third group of codes and the fourth group of codes, wherein the third group of codes generates task-process codes sorted by priority according to the first group of codes, and the generation rule is: according to the first group of codes, sort them from large to small, and generate the third group of codes according to the task number sorted by the sorted value, that is, the The largest priority corresponds to the position of the third group of codes, that is , and in the segment not exceeding this position, that is, in the position , the number of times appears represents the process number of the task code , that is, the position represents the process of the task , that is , and the position of the fourth group of codes is corresponding to the number of people, that is , as shown in Figure 3 .

[0124] After obtaining the third group of codes and the fourth group of codes, the decoding operation can be started, that is, the start time and completion time of each task and each process are calculated; the overall calculation process is relatively complex, and the calculation process is shown in Figure 4 The core of decoding is to calculate the start time according to whether the construction sequence is the first and whether it is the first process, and then directly obtain the end time of each process according to the construction period.

[0125] In a preferred but non-limiting embodiment of the application, in step 3, the decoding operation includes the following steps:

[0126] Step 3-1: Obtain the new encoding after encoding the particles; initialize the sequence of available resources (personnel) at each time point;

[0127] Step 3-2: Obtain the tasks, processes, number of personnel, and duration corresponding to each position according to the new encoding; preliminarily obtain the construction sequence of the wall;

[0128] Step 3-3: Calculate the start and end times of each process of the wall with the construction sequence .

[0129] Step 3-4: Calculate the start time and end time of the process according to the duration expression .

[0130] Step 3-5: If and , search for the start time of the wall that meets the resource requirements for the construction of all processes of the wall in the sequence of available resources starting from the start time of the first process of the wall , if and , , then execute , if and , , then execute , if and , , then execute .

[0131] Step 3-6: Update the sequence of available resources at each time point according to the start time , end time , and required resource number.

[0132] Step 3-7: If the start and end times of all processes are not calculated, go to step 3-4 to execute; if the start and end times of all processes are calculated and the start and end times of all processes of all walls are calculated, end the decoding operation; if the start and end times of all processes are calculated and the start and end times of all processes of all walls are not calculated, go to step 3-3 to execute.

[0133] Then, according to the particle swarm size, the particle swarm is generated based on the coding rule, that is, in addition to the position initialization using the first group of codes and the second group of codes according to the aforementioned coding rule, the speed of each position also needs to be initialized, and for the first group of codes, the speed is randomly generated, and for the second group of codes, the speed is randomly generated in the interval [0, 1]. The fixed duration and the fuzzy duration correspond to the process, and the speed is initialized to 0.

[0134] Step 4: Perform global historical optimal solution and individual historical optimal position update, that is, decode each particle in the current particle swarm according to the decoding rule to obtain the completion duration as the fitness, and update the global historical optimal solution and the individual historical optimal position;

[0135] In the preferred but non-limiting embodiments of the application, step 4 specifically includes:

[0136] In the particle update, the global historical optimal solution and the individual historical optimal position are required. The execution of this step depends on the decoding step, mainly according to the maximum end time obtained from the end time output by the decoding operation as the completion duration, and taking this as the fitness, and then determining the global historical optimal solution and the individual historical optimal position according to the optimization target.

[0137] Step 5: Perform particle update, that is, update the speed and position of the particle based on the designed nonlinear time-varying inertia parameter, the global historical optimal solution and the individual historical optimal position, and the position error feedback drives the speed and position update of the particle;

[0138] In the preferred but non-limiting embodiments of the application, step 5 specifically includes:

[0139] The particle update involves the update of the speed and the position, and the nonlinear time-varying inertia parameter is introduced here to match the nonlinearity of the optimization problem, so as to realize large-scale search in the early stage of operation and small-step adjustment in the later stage. The total number of iterations is set to , the current is the th iteration, the speed and position of the particle at the moment are and , the global historical optimal solution (the position of the optimal particle) is , and the individual historical optimal position of each particle is , so the particle update method is shown in the following formula (1.8):

[0140]

[0141] wherein are the individual and group learning factors, respectively, used to describe the self-cognition and social cognition ability of the particle, is a random number in the inner, respectively maximum minimum inertia parameter, is time-varying inertia parameter at the moment, used to reduce inertia to increase search space in the early search stage, and increase inertia to keep the historical exploration direction in the late search stage.

[0142] Step 6: Perform position and velocity projection, that is, according to the encoding requirements in the encoding rule, the updated particles are processed in position and velocity to prevent exceeding the actual limit, such as the number of assignments being greater than the total number of people;

[0143] In the preferred but non-limiting embodiment of the application, step 6 specifically includes:

[0144] Under the update rule shown in step 5, some out-of-range problems will occur, such as the position of the second group of codes exceeding To better solve the original problem, this step will project the velocity and position after each particle update is completed, and the rule is:

[0145] (1) If the process duration calculation method is fixed duration or fuzzy duration, the velocity projection corresponding to the second dimension code is 0 and the position is 1;

[0146] (2) If it is a process corresponding to variable duration, the second dimension corresponding position is projected into , that is, if it is lower than 0.1, it is assigned as 0.1, and if it is higher than 0.9, it is assigned as 0.9;

[0147] In this way, the particle update is completed.

[0148] Step 7: Perform termination judgment, that is, judge whether the maximum number of iterations is reached, if yes, output the current global optimal solution as the optimal scheduling scheme, otherwise, jump to step 4 for the next iteration. That is, if the current iteration has reached the maximum number of iterations, decode the current global optimal solution as the optimal scheduling scheme output; otherwise, go to step 4 to continue the next round of iteration.

[0149] The modeling method designed in the application describes a practical multi-wall parallel construction resource scheduling problem with fewer constraints. In the algorithm design, a set of encoding and decoding schemes and nonlinear time-varying inertia parameters are specially customized for the actual problem, so that the algorithm obtains a better solution. In the given case, compared with manual experience scheduling, the method of the application has obtained more than 10% improvement in duration, that is, from 45 days of manual scheduling to 40 days.

[0150] The beneficial effects of the application are that, compared with the prior art, the technical effects of the application include:

[0151] The method of the present application comprises the following steps: considering the heterogeneity of construction period, establishing a scheduling optimization model with the minimum construction period as the target and the number of each type of work as the decision variable; based on the particle swarm algorithm, designing an intelligent scheduling algorithm to solve the optimization model, formulating special two-layer coding and decoding rules to realize one-to-one mapping between the scheduling scheme and the particle; based on the nonlinear time-varying inertia parameter, using the difference between the local historical optimum and the global historical optimum to update the particle speed in a negative feedback manner to further realize position updating; finally, through multiple iterations, the global optimal particle is output, and based on the coding mode, the personnel scheduling scheme of multi-wall parallel construction is formed. The present application has the advantages of strong speciality, strong global search ability, fast search speed and the like, can provide a reasonable multi-wall construction scheduling scheme for construction progress management, so as to shorten the construction period to the maximum extent under the condition of ensuring the construction quality, and reduce the construction cost from the perspective of human resource investment.

[0152] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and not to limit them, although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent replacements can still be made to the specific embodiments of the present application without departing from the spirit and scope of the present application, and any modifications or equivalent replacements should be covered within the protection scope of the claims of the present application.

Claims

1. A scheduling modeling and intelligent scheduling method for parallel construction of multiple walls, characterized in that, include: Step 1: Perform multi-wall parallel construction resource scheduling modeling, that is, with the goal of minimizing the construction period, integrate construction requirements and multi-type construction period calculation characteristics to establish a multi-wall parallel construction resource scheduling model for nuclear power engineering civil engineering scenarios; Step 2: Perform parameter initialization, that is, determine the relevant parameters of the particle swarm algorithm, including particle swarm size, inertia parameter, learning factor and maximum number of iterations; Step 3: Design and initialize the encoding and decoding scheme. Based on the characteristics of the scheduling model, design an encoding scheme that can accommodate multiple types of work periods to realize the mapping from the scheduling scheme to the particles; and design a suitable decoding scheme to realize the mapping from the particles to the scheduling scheme, complete the particle fitness calculation, and on this basis, generate the first generation of particle swarm according to the particle size. Step 3 specifically includes: The encoding rule is to use two sets of codes for each particle: Group 1 coding is priority-based task coding, by Composed of real numbers, for position value It expresses the wall The priority value of a certain process; Group 2 coding is indivual The number between the two represents the proportion of the number of people assigned to that task process to the total number of people; Step 3 also includes: The decoding steps are as follows: First, generate two new sets of codes according to the encoding rules, denoted as the third set of codes and the fourth set of codes. The third set of codes generates task-process codes sorted by priority based on the first set of codes. The generation rule is: sort the priority values ​​in the first set of codes from largest to smallest, and generate the third set of codes according to the sorting values. The higher priority level corresponds to the task code located in the 3rd group of codes. place, The wall corresponding to this priority. And in segments not exceeding that position, i.e., position middle, The number of times it appears represents the task code. Process number That is, location Representative task process ,Right now The position of the fourth group of codes for The corresponding proportion of people; After obtaining the third and fourth sets of codes, the decoding operation can begin, which involves calculating the start and completion times of each task and each step. Step 4: Execute global historical best solution and individual historical best position update, that is, decode each particle in the current particle swarm according to the decoding rules to obtain the completion time as fitness, and update the global historical best solution and individual historical best position accordingly. Step 5: Perform particle update, which is to update the particle velocity and position based on the designed nonlinear time-varying inertial parameters, the global historical optimal solution and the individual historical optimal position, using position error feedback to drive the particle velocity and position update. The nonlinear time-varying inertial parameters are calculated based on the inertial parameters in Step 2. Step 6: Perform position and velocity projection, that is, process the updated particles in terms of position and velocity according to the coding requirements in the coding rules; Step 7: Execute the termination judgment, that is, determine whether the maximum number of iterations has been reached. If so, output the current global optimal solution as the optimal scheduling scheme; otherwise, jump to step 4 and proceed to the next iteration.

2. The scheduling modeling and intelligent scheduling method for parallel construction of multiple walls according to claim 1, characterized in that, Step 1 specifically includes: In the parallel construction of multiple walls in nuclear power plant construction Each wall has Each process, with Represents wall process , Construction time is defined as , All are positive integers; exist Each work group, that is, there exists Types of work, each shift has 100 people. One, wall process need Team members Bit, It is a positive integer; make express exist When the moment ends, then, The end time is , Start time is , This indicates the upper limit of the construction period for multiple walls constructed in parallel. This means that the search starts from the first time period and continues until... At this point, it will output the result that... of The optimization objective of minimizing the project duration is expressed as minimizing the latest completion time, as shown in formula (1.1) below: ； The construction process is grouped according to the method of calculating the construction period. Work processes representing fixed durations, The process representing efficiency and time. The process representing a vague timeframe; for , The construction period is ;for ,make The work efficiency is The total workload is ;for The workload limit is Construction periods that are below or above the limit are and Therefore, the project duration expression is obtained as shown in formula (1.2). : 。 3. The scheduling modeling and intelligent scheduling method for parallel construction of multiple walls according to claim 2, characterized in that, Step 1 also includes: Based on formulas (1.1) and (1.2), other constraints are given: Completion time limit: Each process has only one completion time, as described in formula (1.3) below. ； Preceding constraints: The process constraints and flow construction requirements are expressed by using the time difference between the completion of successive processes equal to the duration of subsequent tasks, as shown in the following formula (1.4): ； Resource upper limit constraint: Define a binary variable as shown in formula (1.5): ； This constraint actually means: for a given , express time To determine if the process is in progress, sum the results for all walls and processes. In all wall finishing processes, by adding You can get The total amount of resources required at any given time.

4. The scheduling modeling and intelligent scheduling method for parallel construction of multiple walls according to claim 3, characterized in that, Step 1 also includes: The overall optimization problem is obtained as shown in Equation (1.6): ； Next, we use formula (1.6) The optimization problem, simplified by the objective function, is shown in Equation (1.7): ； in Indicates obedience, This indicates that formula (1.2) is being referenced.

5. The scheduling modeling and intelligent scheduling method for parallel construction of multiple walls according to claim 4, characterized in that, Step 2 specifically includes: The particle swarm is set to have 100 particles, with a maximum inertial parameter of 0.9 and a minimum inertial parameter that is randomly obtained. The individual and swarm learning factors are set to 0.8 and 0.7, respectively, and the maximum number of iterations is 300.

6. The scheduling modeling and intelligent scheduling method for parallel construction of multiple walls according to claim 1, characterized in that, Step 3 also includes: Based on the particle swarm size, a particle swarm is generated according to encoding rules. This means that in addition to initializing positions using the first and second sets of codes according to the encoding rules, the velocities at each position also need to be initialized. For the first set of codes, the velocities are randomly generated; for the second set of codes, the velocities are... The process is randomly generated within the interval, while the speed of the processes corresponding to fixed and fuzzy durations is initialized to 0.

7. The scheduling modeling and intelligent scheduling method for parallel construction of multiple walls according to claim 6, characterized in that, In step 3, the decoding operation specifically includes the following steps: Step 3-1: After encoding the particles, a new encoding is obtained; initialize the sequence of available resource counts at each time step; Step 3-2: Based on the new coding, obtain the task, process, number of personnel, and construction period corresponding to each location; preliminarily obtain the construction sequence of the wall; Step 3-3: Calculate the construction sequence as follows The wall Start and end times of each process; Step 3-3-1: Based on construction time Calculation process start time With end time ; Specifically: If and Based on the construction sequence The first process of building the wall begins at the start time, searching the available resource sequence for resources that satisfy the wall's requirements. Start time of all processes and flow construction required for the resources to begin Then execute ;if and , Then execute ,if and , Then execute ,if and , Then execute ; Steps 3-4: Based on start time End time And update the sequence of available resources at each time point to the required resource count; Steps 3-5: If the start and end times of all processes have not been calculated, Then proceed to step 3-3-1; if the start and end times of all processes have been calculated and the start and end times of all processes for all walls have been calculated, the decoding operation ends; if the start and end times of all processes have been calculated but the start and end times of all processes for all walls have not been calculated, Then proceed to step 3-3 to execute.

8. The scheduling modeling and intelligent scheduling method for parallel construction of multiple walls according to claim 7, characterized in that, Step 4 specifically includes: Based on the end time output by the decoding operation, the maximum end time is obtained as the completion period and used as the fitness. Then, the global historical best solution and the individual historical best position are determined according to the optimization objective. Step 5 specifically includes: Set the total number of iterations to Currently the number In the next iteration, the particle of The velocity and position at time t are respectively and The global historical optimal solution is The individual historical optimal position of each particle is Therefore, the particle update method is as shown in formula (1.8): ； in, These are individual and group learning factors, used to describe the particle's self-awareness and social cognitive abilities. yes Random numbers within, These are the maximum and minimum inertial parameters, respectively. yes Nonlinear time-varying inertial parameters at time t.

Images