WO2021031336A1

WO2021031336A1 - Method for automated construction progress resource optimization employing building information model

Info

Publication number: WO2021031336A1
Application number: PCT/CN2019/113220
Authority: WO
Inventors: 林佳瑞; 王珩玮; 张建平
Original assignee: 清华大学
Priority date: 2019-08-16
Filing date: 2019-10-25
Publication date: 2021-02-25
Also published as: CN110503325B; CN110503325A

Abstract

The invention relates to a method for automated construction progress resource optimization employing a building information model, the method comprising the following steps: preparing a building information model comprising a building element type and a main resource type, and adding necessary information to a work package template database to form multiple required work package templates; performing data integration on the basis of a building component type and material type; using an RCPSP constraint condition and an objective function to generate a progress resource optimization model, and solving to complete construction progress resource optimization. The invention substantially improves data integration and construction optimization efficiency, and reduces the time required for construction optimization from days to hours.

Description

An Automatic Optimization Method of Construction Schedule Resources Based on Building Information Model

Technical field

The invention relates to a resource optimization method, in particular to a construction schedule resource automatic optimization method based on a building information model.

Background technique

Resource-constrained project scheduling problem (RCPSP) is an important mathematical model for resource optimization of construction schedule. RCPSP is an NP problem. Researchers use many heuristic algorithms to solve these problems, including heuristic algorithms based on priority rules such as simplified branch and bound method, local search technology, etc., and meta-heuristic algorithms, such as genetic algorithm, Particle swarm algorithm and tabu search. For construction projects, many studies have established more complex problem models to meet actual needs, such as considering multiple projects and changing resource constraints, as well as resource constraints and time-cost trade-off (TCT) issues. Generally speaking, these issues can be unified into the form of RCPSP. Although the mathematical model of RCPSP can describe the actual engineering requirements and complete the solution within a reasonable time, the existing related research rarely considers the difficulty of data acquisition and the complexity of the required data. Therefore, the RCPSP solution technology is still in practical applications. Facing problems such as low efficiency and difficulty in practical application.

1) Resource-constrained construction project optimization scheduling problem

The most basic RCPSP defines a series of processes that are related to each other. Each process occupies several reusable resources during the process, and the availability of these resources is constrained by a constant upper limit during the entire project period. On this basis, various more complete RCPSP models can be derived. An analysis of 24 recent studies reveals that process duration, pre-post and post-relationships, and resource availability are necessary information for the solution of the basic RCPSP model, and many studies consider constraints such as multi-mode and cost. But only one study considered the reusable types of different resources. Further statistical analysis shows that the problem of schedule resource optimization involves 7 types of information such as process duration, multiple modes, pre- and post-relationships, contracting time and milestones, resource availability, resource reusable types, and costs. However, in the above 24 studies, only consideration is often given. There are no RCPSP models that consider these 7 types of information at the same time for 5 or less factors. In addition, related studies often set the process duration to a fixed value, or express the impact of resource usage or process cost on duration through multiple modes. However, the unit price of resources and the time interval of pre- and post-tasks are not considered, and the consideration of cost is relatively simple. Therefore, the existing optimization models cannot reflect the resource optimization scenarios of the real construction process.

2) Construction schedule resource optimization method based on BIM (Building Information Modeling)

At present, there have been a lot of studies exploring the use of BIM to generate models for scheduling optimization problems. Including the use of BIM models to derive schedules, combining BIM models and discrete event simulations, integrating BIM and particle swarm optimization methods, etc. Relevant studies have also tried to consider conditions such as space constraints. There are also studies that set up a series of simple rules to automatically generate progress and resource plans and optimize them. However, relevant research and methods assume that the BIM model already contains complete schedule, resource, and cost data, and the correlation between the corresponding data is also available. However, this assumption is not correct. The integration and correlation of current progress, resources, cost and other data is still highly dependent on manual work, and there are problems such as low level of automation and long time consumption. At the same time, relevant research has not incorporated the accumulated engineering knowledge such as standard construction techniques into the resource optimization process, and it is difficult to use existing engineering experience and knowledge. Finally, related studies often need to manually extract and convert the data of the BIM model, build a schedule resource optimization model, and solve the resource optimization problem, and finally adjust the BIM model manually according to the optimization results. The entire process requires a lot of manpower and is inefficient. And it is prone to errors and omissions.

In summary, the existing technology has the following problems:

1) The RCPSP model is not complete enough to fully consider the 7 types of information for construction resource optimization.

2) Data integration and correlation processes are highly dependent on manual work, which is inefficient and error-prone.

3) Cannot make full use of existing engineering knowledge or experience such as standard processes.

4) The construction process of schedule resource optimization model requires manual intervention, which is cumbersome and error-prone, and wastes a lot of time.

Summary of the invention

In view of the above problems, the purpose of the present invention is to provide a method for automatically optimizing construction schedule resources based on building information models, which can greatly improve the efficiency of data integration and construction optimization, and shorten the time for optimization of construction plans from several days to several hour.

In order to achieve the above objectives, the present invention adopts the following technical solutions: A method for automatically optimizing construction schedule resources based on building information models, which includes the following steps: 1) Prepare building information models with building element categories and main resource categories, and work Enter or import the required work package templates in the package template database, and use the work package templates to generate work packages. Each work package and the building components are many-to-many associations; 2) Data integration is based on the types of building components and materials 3) Based on the information model formed by data integration, using the constraints and objective function of RCPSP, the schedule resource optimization model is automatically generated and solved automatically to complete the optimization of the construction schedule resource.

Further, in the step 2), the data integration method is: 2.1) Based on the classification and coding of building component types, traverse each work package template to establish a coding tree; 2.2) Automatically associate work packages with building components; 2.3) Generate rule-based Work package logic: After the work package is associated with the building component, the sequence logic is generated based on the rules. The definition of the rule is realized based on the attributes of the work package. The attributes of the work package include spatial location, component type and engineering specialty; the basic form of the rule is to have The construction of a work package with a certain attribute or attribute combination should be earlier or later than a work package with another attribute or attribute combination; after these rules are predefined, the relevant work package can be queried through the attributes of the work package, and then according to the pre-defined The defined rules automatically generate the logical sequence of work packages.

Further, in the step 2.1), for each level of the building component type code in each work package template, if the level is not included in the coding tree, the node is added to the coding tree, and the code corresponding to the entire code The tree node is associated with the work package template; the specific process is: a) Traverse all the work package templates to get the template t, set the code of the template t as c; b) set the tree root node of the building component as curnode; c) traverse the c For each layer, get the layer code n; d) Determine whether the child node of the tree root node curnode contains the layer code n, if it does, go to step e); if not, create the child node n for the tree root node curnode, Go to step e); e) assign the tree root node curnode to n in the child node; f) determine whether n exists in the next layer, if it exists, return to step c), otherwise associate the template t with the tree root node curnode, Go to step g); g) repeat the above steps until there is no next template, and complete the establishment of the coding tree.

Further, in the step 2.2), the automatic association of the work package and the building component includes 4 steps: the first matching of the building component and the work package template, the second matching of the building component and the work package template, the instantiation of the work package and the work package Reorganization; the first matching is that each building component is matched step by step from the root node and is associated with the work package template associated with all the matched nodes; traverse the results of the first matching, and eliminate those that do not meet the material coding matching principle Associate, complete the second matching process; the work package instantiation process is the process in which the work package template is divided according to the construction space where the corresponding building components are located; each group of building components after the division corresponds to a work package; work package reorganization It is to traverse all building components and determine whether their attributes meet the usage conditions of each quota of its associated work package; then get the specific quota combination corresponding to each building component. When a certain quota combination completely matches the building component, it will be generated A new work package.

Further, the work package instantiation process is: (1) traverse all work package templates to obtain work package template t; (2) traverse all building components associated with work package template t to obtain building component b; (3) obtain building For the construction area A of component b, judge whether there is a corresponding work package in the construction area A, if yes, the work package corresponding to the facility area A is w, and go to step (4); if there is no work package template for the construction area A t’s work package w, go to step (4); (4) associate the building component b with the work package w, and determine whether there is a next building component, if there is, return to step (2), otherwise go to step (5); 5) Judge whether there is the next work package template, if there is, return to step (1), otherwise end.

Further, the work package reorganization process is: (1) traverse all work packages to obtain work package w; (2) traverse all building components related to work package w to obtain building component b; (3) traverse quotas to obtain work package w; The quota combination q that the component b conforms to, and determine whether the quota combination set s includes the quota combination q; if included, the quota combination q is associated with the construction component b, otherwise, the quota combination q is added to the quota combination set s, and then the quota combination q is associated with the construction component b; (4) Determine whether there is the next building component, if there is, return to step (2), otherwise, create a work package for each quota combination q in the quota combination set s, and add the work The package is associated with all building components related to the quota combination q.

Furthermore, in the step 2.3), the work package logic generation specifically includes the following steps: (1) establish a work package attribute set φ; (2) traverse all work packages to obtain work package w; (3) traverse all work packages w For all attributes, get the attribute t, judge whether the attribute t belongs to the attribute set φ, then go to step (4), otherwise, add the attribute t to the attribute set φ and go to step (4); (4) generate the attribute t and work package The association between w, and judge whether there is still work package w, if there is, return to step (2), otherwise, go to step (5); (5) Traverse all the rules to get the rule r, from the relevant attributes in the rule r , By querying the association between the attribute set φ and the work package set, the previous work package set s1 and the subsequent work package set s2 are obtained; (6) The sequence relationship between the previous work package set s1 and the subsequent work package set s2 is established: Add all the work reports in s1 to the predecessor task set of all work packages in s2.

Further, in the step 3), the constraint conditions include inter-process relationship constraints, milestone constraints, resource availability constraints, process internal constraints and resource mode constraints.

Further, the five types of constraints are: Inter-process relationship constraints: For process i, use T _{i to} represent the start time SS _i or end time SF _i of process i. The difference between the key time attributes of each process has a lower limit and an upper limit, then This relationship is:

minLag≤T _j -T _i ≤maxLag

Among them, T _i represents the SS _i or SF _{i of the} previous step i, T _j represents the SS _j or SF _{j of the} subsequent step j, minLag represents the shortest interval, and maxLag represents the longest interval;

Milestone constraint: single process completion time constraints; single step completion time constraints mainly completion time for step i SF _i programs, which must be earlier than the predetermined milestone M _i, is expressed as:

SF _i ≤M _i ;

Resource availability constraint: at any point in time t, the total demand RD _{kt of} resource k should be less than the total supply RS _kt :

RD _kt ≤RS _kt

For reusable resources, the total demand should be equal to the maximum value of the sum of the demands of the processes in progress at all times before:

Among them, d _ik represents the demand of process i for resource k, DA _t ={i|SS _i ≤t≤SF _i }, which represents the set of processes in progress at time t;

For non-reusable resources, the total demand should be the sum of the demands of the various started processes:

Where ASA _t = {i|t≥SS _i }, which represents the set of processes that have started at time t;

_{Process internal constraints: the resource quantity qr ik} of the labor resource k required for process i is inversely proportional to the process duration SDi:

qr _ik ＝d _ik SD _i

Among them, _{the unit of qr ik} is person multiplied by time, that is, 1 person needs to spend qr _ik days, or qr _ik individual needs 1 day to complete process i;

Resource mode constraints: Introduce an indicator variable MI _iu that _{indicates whether the resource is selected, diku} represents the demand for resource k in the u-th group of resources in process i; this variable is equal to 0 or 1, and satisfies the following formula:

MI _{iu is} used _{to calculate d ik} to ensure that all resource requirements belong to a certain group:

Further, in the step 3), the RCPSP model takes the total construction period and total cost as the objective function, and the specific calculation method is as follows:

(1) Total construction period

The total construction period TD is calculated using the following equation:

TD=max(SF _i )-SS

Among them, SS is the start time;

(2) Total cost

Including direct cost and indirect cost, direct cost DC is the synthesis of the product of resource quantity and price, namely:

Among them, p _k is the price of resource k;

The indirect cost IC includes loan interest, site lease, design cost, change cost and supervision cost; only the indirect cost related to the construction period is considered, and it is considered to be linearly related to the construction period:

IC=TD·dc

Among them, dc is the daily overhead consumption.

The present invention has the following advantages due to the above technical solutions: 1. The present invention supports the integration of all construction schedule resource optimization related information, can support the data requirements of multiple schedule resource optimization models, and can use BIM and a small number of data sources to automatically generate constraints based on constraints. Plan the schedule resource optimization model and automatically solve the model to obtain an optimized construction plan, which can greatly improve the efficiency of data integration and construction optimization. 2. The present invention establishes a multi-mode RCPSP model that can simultaneously consider the process length, multi-mode, front-to-back relationship, contracting time and milestones, resource availability, resource reusable type, and cost, and makes up for the completeness of the existing RCPSP for engineering construction Sexual issues. 3. The present invention introduces BIM and knowledge data based on the work package database, and provides an automated information integration method to make up for the lack of technical data acquisition links in the construction of RCPSP solution, and solves the problem that existing knowledge and technology cannot be used for actual project schedule resource optimization Therefore, the application efficiency is improved. 4. An automatic construction and solution method for establishing a schedule resource optimization model based on BIM is proposed. By using data sources from the current specifications, it avoids excessive requirements on the data format, and can automatically generate a RCPSP model with good versatility, and automatically Solving the optimization model solves the problem of the low level of automation in the existing technology, can greatly save the optimization model construction and solution time, and reduce manual input.

Description of the drawings

Figure 1 is a schematic diagram of the overall flow of the present invention;

Figure 2 is a structure diagram of the work package template database;

Figure 3 is a flowchart of a coding tree generation method;

Figure 4 is a flowchart of the association between work packages and components;

Figure 5 is a schematic diagram of the instantiation of the work package based on the construction area;

Figure 6 is an example of work package reorganization based on quota;

Figure 7a is a schematic diagram of the association of rules, attributes and work packages;

Figure 7b is a flow chart of a rule-based work package sequence logic generation method;

Figure 8 is the resource model and selection index of the process;

Figure 9 is a flowchart of the CP model generation and solution method;

Figure 10 is the work package and sequence logic corresponding to the 7th layer;

Figure 11 is a schematic diagram of resource constraints in Example 3;

Figure 12 is the total construction period and total cost of the schedule optimization results;

Figure 13 is the process time length of 7 layers in the schedule optimization result;

Figure 14 shows the demand for resource 2 in the results of calculation example 1 and calculation example 2;

Figure 15 shows the demand for resources 38 in the results of calculation examples 1 and 3;

Figure 16a is a semi-automatic establishment of RCPSP and progress optimization in the project using the method provided by the present invention;

Figure 16b is an existing general RCPSP application process.

Best embodiment of the invention

The present invention will be described in detail below with reference to the drawings and embodiments.

As shown in Figure 1, the present invention provides a method for automatically optimizing construction schedule resources based on a building information model. The method is based on a work package to provide an overall structure for data integration and provide support for subsequent optimization of construction schedule resources.

As shown in Figure 2, the building information model is composed of five types of core entities, which are building components, work packages, quotas, quota items, and resources. Among them, building components come from BIM, and the other 4 types of entities come from the work package template database. For each building component, it needs to include basic data, element type and main material properties; correspondingly, the basic data should include the volume, area, length and weight of the building component; element types and main materials are used to search for related work package templates , Can be expressed using a unified classification and coding standard. The work package template database stores a series of work package templates. Each work package template includes the following data: building element types, several quotas, basic units of each quota, usage conditions, quota items, resources and quota amounts for each quota item . The quota data can come from national and local quota standards. Among them, the quota refers to the engineering quantity quota, which is the consumption of various construction resources in the construction process obtained by the state or local through investigation and statistics; the quota item refers to the demand for a certain resource in a construction work.

As shown in Figure 1 and Figure 2, the present invention includes the following steps:

1) Data preparation: Prepare a building information model with building element categories and main resource categories, and enter or import the required work package templates in the work package template database, and then use the work package templates to generate work packages. Each work package is related to There is a many-to-many relationship between building components.

A work package includes multiple quotas, each quota includes multiple quota items, and each quota item corresponds to a resource.

2) Data integration: data integration based on building component types and material types;

The basis of data integration is two unified classification and coding systems, which respectively indicate the type of building components and the type of materials. Building component types can use OmniClass table 21 or Uniclass table Ef, and material types can use Omniclass table 23 or Uniclass table Pr. Before data integration, the building components in the default BIM database are attached with their component types and corresponding codes of key materials. The work package templates in the default work package template database all have codes corresponding to the types of building components, and the types of material resources in the resource list all have corresponding material codes.

The data integration method is:

2.1) Based on the classification and coding of building component types, traverse each work package template to establish a coding tree;

As shown in Figure 3, for each level of building component type coding in each work package template, if the level is not included in the coding tree, the node is added to the coding tree. Associate the code tree node corresponding to the entire code with the work package template. The specific process is:

a) Traverse all work package templates to get template t, and set the code of template t to c;

b) Let the tree root node of the building component be curnode;

c) Traverse each layer of c to get the layer code n;

d) Judge whether the child node of the tree root node curnode contains the layer code n, if it does, go to step e); if it does not, create a child node n for the tree root node curnode and go to step e);

e) Assign the value of n in the child node to the tree root node curnode;

f) Judge whether n has the next layer, if so, return to step c), otherwise associate the template t with the tree root node curnode, and go to step g).

g) Repeat the above steps until there is no next template, and complete the establishment of the coding tree.

2.2) Automatically associate work packages with building components;

As shown in Figure 4, the automatic association of work packages and building components includes the following four steps: first matching of building components and work package templates, second matching of building components and work package templates, work package instantiation and work package reorganization.

Among them, the first matching is that each building component is matched step by step from the root node and is associated with the work package template associated with all the matched nodes. Traverse the results of the first matching, and eliminate the associations that do not meet the material coding matching principle, and then the second matching process can be completed.

After the two matches are completed, the corresponding relationship between the work package template and the building component has been established. This relationship means that the building component can be constructed using the work package template. Normally, the result is many-to-many. Different building components will use the same work package template for construction, and a building component may also have multiple work package templates for selection.

The work package instantiation process is a process in which the work package template is divided according to the construction space where the corresponding building components are located. Each group of building components after division corresponds to a work package, as shown in Figure 5. The specific process is:

(1) Traverse all the work package templates to get the work package template t;

(2) Traverse all building components associated with the work package template t to obtain building component b;

(3) Obtain the construction area A of the building component b, and determine whether the construction area A has a corresponding work package. If yes, the work package corresponding to the facility area A is w, and go to step (4); if there is no work package, it is the construction area A creates the work package w of the work package template t and proceeds to step (4);

(4) Associate the building component b with the work package w, and judge whether there is the next building component, if there is, return to step (2), otherwise go to step (5);

(5) Judge whether there is the next work package template, if there is, return to step (1), otherwise end.

Work package reorganization is to traverse all building components and determine in turn whether their attributes meet the usage conditions of each quota of its associated work package; then the specific quota combination corresponding to each building component can be obtained. For a work package containing n quotas, the quota combination may be 2 ⁿ -1. When a certain quota combination completely matches the building component, a new work package is generated. As shown in Figure 6, the specific process is:

(1) Traverse all work packages to get work package w;

(2) Traverse all building components related to work package w to obtain building component b;

(3) Traverse the quota, get the quota combination q that matches the building component b, and determine whether the quota combination set s includes the quota combination q; if it does, associate the quota combination q with the construction component b, otherwise, add the quota combination q to the quota In the combination set s, then the quota combination q is associated with the construction component b;

(4) Determine whether there is the next building component. If there is, return to step (2). Otherwise, create a work package for each quota combination q in the quota combination set s, and associate the work package with the quota combination q All building components are connected.

2.3) Generate rule-based work package logic;

After the work package is associated with the building component, the sequence logic is generated based on the rules. The definition of this rule is mainly realized based on the attributes of the work package. The attributes of the work package involved include information such as spatial location, component type and engineering specialty. The basic form of the rule is that the construction of a work package with a certain attribute (or attribute combination) should be performed before or after the work package with another attribute (or attribute combination) (as shown in Figure 7a). After pre-defining these rules, related work packages can be queried through the attributes of the work packages, and then the logical sequence of the work packages is automatically generated according to the predefined rules.

As shown in Figure 7b, the work package logic generation specifically includes the following steps:

(1) Establish a work package attribute set φ;

(2) Traverse all work packages to get work package w;

(3) Traverse all the attributes of all work packages w to get the attribute t, judge whether the attribute t belongs to the attribute set φ, then go to step (4), otherwise, add the attribute t to the attribute set φ and go to step (4);

(4) Generate the association between the attribute t and the work package w, and judge whether there is still work package w, if there is, go back to step (2), otherwise go to step (5);

(5) Traverse all the rules to get the rule r. From the relevant attributes in the rule r, by querying the association between the attribute set φ and the work package set, the pre-order work package set s1 and the subsequent work package set s2 are obtained;

(6) Establish the sequence relationship between the pre-order work package set s1 and the subsequent work package set s2: that is, add all the work reports in s1 to the pre-task set of all work packages in s2.

3) Based on the information model formed by data integration, using the constraints and objective function of RCPSP, the schedule resource optimization model is automatically generated, and automatically solved to complete the optimization of construction schedule resources;

Constraints include the following five types of constraints: inter-process relationship constraints, milestone constraints, resource availability constraints, process internal constraints and resource model constraints. Among these constraints, the first three are constraints that are generally used in CP solving RCPSP, and the latter two define a new multi-mode RCPSP. In the traditional problem model, one model corresponds to the duration and cost of a process. In the present invention, the duration and cost of each model in the problem model are _{calculated by r ik} and q _i , which can be compared with the actual project quota and each process. The amount of work is directly linked. Among them, r _ik represents the quantity ratio of resource k in process i, and represents the quantity of resource k that needs to be consumed per unit basic quantity. q _i represents the basic quantity of the final result of process i, such as volume, area, weight, etc.

The five constraints are specifically:

3.1) Inter-process relationship constraints: For process i, use T _{i to} represent the start time SS _i or end time SF _i of process i, then the main manifestation of the inter-process relationship is the difference between the key time attributes of each process. This difference can have a lower limit and an upper limit, then this relationship can be expressed as:

minLag≤T _j -T _i ≤maxLag (1)

Among them, T _i represents the SS _i or SF _{i of the} previous step i, T _j represents the SS _j or SF _{j of the} subsequent step j, minLag represents the shortest interval, and maxLag represents the longest interval. In this embodiment, the generated inter-process relationship basically does not include maxLag, and minLag=0 in most cases.

3.2) Milestone constraints: time constraints for the completion of a single process

In schedule management, it is a general and effective method to set deadlines to control schedules for key processes. Single process completion time constraints mainly completion time for step i SF _i programs, which must be earlier than the predetermined milestone M _i, it is expressed as:

SF _i ≤M _i (2)

3.3) Resource availability constraints

At any point in time t, the total demand RD _{kt of} resource k should be less than the total supply RS _kt , namely:

RD _kt ≤RS _kt (3)

The calculation method of total demand is related to whether resources can be reused. For reusable resources such as labor and machinery, the total demand should be equal to the maximum value of the sum of the demands of the processes in progress at all times before:

Among them, d _ik represents the demand of process i for resource k, which is a quantity that does not change with the progress of the optimization process, which is the setting of most RCPSPs. DA _t ={i|SS _i ≤t≤SF _i }, representing the set of processes in progress at time t.

For non-reusable resources, such as most materials, the total demand should be the sum of the demands of the various started processes:

Where ASA _t = {i|t≥SS _i }, which represents the set of processes that have started at time t. In this embodiment, it is assumed that _dik is a quantity that does not change with the progress of the optimization process.

3.4) Process internal constraints

_{The resource quantity qr ik} of the labor resource k required by the process i is inversely proportional to the process duration SDi, namely

qr _ik = d _ik SD _i (6)

Among them, _{the unit of qr ik} is person times time, that is, it takes qr _ik days for _{one person, or one day for qr ik} individuals to complete process i. The duration of each process has a clear correlation with the amount of allocated labor resources _dik. Equation (6) is just a typical duration function, it can also be in other forms.

3.5) Resource model constraints

In this embodiment, the resource cost of each process is related to the selection of resources. In the process of schedule optimization, it is necessary to perform a single selection among the groups of resources owned by a work package. Different choices affect the cost and duration, which will directly affect the results of the schedule. To this end, it is necessary to introduce an indicator variable MI _iu that marks whether the resource is selected, as shown in Figure 8, where _diku represents the demand for resource k in the u-th group of resources in process i. This variable is equal to 0 or 1, and satisfies the following formula:

MI _{iu is} used _{in the calculation of d ik} to ensure that all resource requirements belong to a certain group:

In this embodiment, the RCPSP model simply considers the total construction period and total cost as the objective function (different objective functions can be used in the specific implementation process, and the overall method of the present invention does not need to be changed), and the specific calculation method is as follows:

(1) Total construction period

The total construction period TD is calculated using the following equation:

TD=max(SF _i )-SS (9)

Among them, SS is the start time.

(2) Total cost

Including direct costs and indirect costs. The direct cost DC is the synthesis of the product of resource quantity and price, namely:

Among them, p _k is the price of resource k.

The indirect cost IC includes loan interest, site lease, design cost, change cost, supervision cost, etc. Because this part of the cost is very complicated and has a small correlation with the schedule arrangement, when the schedule is optimized, generally only the overhead related to the construction period is considered, and it is considered to be linearly related to the construction period:

IC = TD dc (11)

Among them, dc is the daily overhead consumption.

Based on the above definition, as shown in Figure 9, the CP solution method is:

(1) Create a CP object.

(2) Put all the created variables and calculation expressions of the objective function into the CP object.

(3) First determine the solution parameters (such as the solution time limit, set the configuration when the CP object is solved).

(4) Obtain data from the data model and generate a series of objects of key data classes and other related data classes required for the input of resource optimization model. These data classes and objects are related to the variables and objective functions contained in the CP object created above. The calculation expression parameters correspond.

(5) Use formula (1) to formula (11) to link and correlate constraint variables, objective functions, and generated data objects to build the correlation between variables and parameters, and finally call the Solve() method of the CP object to implement the model Automatic solution.

Examples:

1) Data preparation

Using data preparation methods, related work package templates were created. Taking into account the difference in construction resources, the cast-in-place concrete construction work is divided into three work package templates, namely formwork engineering, steel reinforcement engineering and concrete engineering. At the same time, a work package template was established for the precast concrete wall and the precast concrete slab. Therefore, there are a total of 8 work package templates for test verification. Each work package template is composed of four parts, namely basic information, classification, process flow and resources. Among them, the classification and resources are related to the information integration process. At present, there is only one architectural element code in the classification, which is used for the first matching in the core model integration process. The resource part includes several quotas, and each quota includes several quota items. Each quota item corresponds to a resource in the resource database. The resources in a quota that can reflect the characteristics of the process are designated as the main material, and their codes can be used to complete the second matching process. The data of these quotas and resources are collected from the consumption quota of prefabricated construction projects (TY 01-01(01)-2016).

2) Data integration

In order to establish the sequence logic between work packages, six construction sequence rules as shown in the table were established. Among them, the first and second rules define the spatial sequence, and the rest define the inter-process sequence. These rules are restricted to the construction area (floor), so each floor defines a rule. The first category defines 22 rules in total, and the remaining 5 categories define 23 rules, totaling 137 rules.

Table 1 Construction sequence rules

Through a two-step matching process, a connection is established between the architectural elements in BIM and the work package template. The number of architectural elements that may be associated with the work package template in the whole process is divided by element type, which is listed in Table 2. In theory, the first matching process can complete the screening of components by category, so after the first screening, except for the walls and boards related to the process template library, other components are not related. In the second matching process, some walls that did not conform to the construction materials in the process template were excluded from the model through the material type. It is worth noting that in the first time, part of the panels and walls were not associated with the process template, because the process type was not added to these components during the coding process. In addition, Table 3 counts the number of components associated with each process template in the two-step matching process. After completing the first time, the number of components associated with the process template associated with the same component type is the same, and equal to the sum of the number of corresponding component types in Table 2. This result is consistent with the theory and verifies the correctness of the first matching algorithm. After the second matching is completed, compared to the result of the previous step, the components are further divided according to the material properties. Among the 4600 related wall components, there are 1831 precast concrete walls, 1946 cast-in-place concrete walls and 382 other types of walls (such as masonry walls). Although the three work package templates related to the in-situ concrete casting process are not bound, since the same type code and the material code corresponding to each work package template are added to each related building element, they can all be connected with each other. Formwork engineering, steel reinforcement engineering and concrete engineering are related, and there is nothing missing.

Table 2 The number of various architectural elements in the modeling process

Table 3 Number of building elements associated with the work package template

After completing the association between the architectural elements and the work package template, it is first necessary to complete the instantiation and reorganization of the work package according to the construction area. We choose each floor of the building as a construction area, so if there are related architectural elements on this floor, each work package template will generate a work package on this floor. After this step, we got a total of 167 work packages (-1 to 22 layers, consistent with Table 3, 7 for each layer, plus 6 for the top layer). After that, the regrouping process was completed by judging that each component can meet the requirements of the quota. The precast shear wall work package of each layer was continuously broken down into two, one of which corresponds to the 0th quota in the work package, and the other One corresponds to the second and fourth quotas. -1 to 21 floors of prefabricated shear wall work packages were decomposed, each layer has 1 work package, so there are 22 work packages, so this verification finally got 189 work packages.

For these work packages, the relationship between the work packages generated based on the six types of rules in Table 1 is shown in Figure 10. Although there is a redundant relationship, there is no conflict. In this way, the relationship between work packages and other information related to work packages can be directly transformed into the core model based on IFC.

3) Progress optimization

Four different schedule optimization calculation examples are designed, and their resource constraints or objective functions are different, as shown in Table 4. Example 1 is the control group. Case 2 has less constraints on resources 2 than Case 1, Case 3 tidies up the time-varying resource constraints, and Case 4 chooses to minimize the total cost as the optimization goal.

Table 4 Progress optimization settings of different calculation examples

The result of schedule optimization is shown in Figure 12, which is consistent with the theory. As the case 2 and case 3 have stricter resource constraints than case 1, the total construction period and total cost are higher. Compared with case 1, because case 4 aims to optimize the total cost, its total cost is lower, but the total construction period is longer.

In the result of schedule optimization, the comparison result of process duration is shown in Figure 13. Compared with case 1, case 2 changes the constraint of resource 2. Resource 2 is a template mechanic and is used by both process 27 and process 31. The results show that the duration of process 27 and process 31 increased after the resource 2 constraint was reduced from 30 to 20, while the duration of other processes did not change. In addition, compared with the calculation example 1, in the calculation example 3, the resource 38 is set with time-varying constraints, and the resource is the precast concrete external wall panel required by the process 24, so the duration of the process 24 has also changed. In calculation example 4, due to the change of the optimization target, the duration of multiple processes is affected.

Figure 14 shows the demand for resource 2 in case 1 and case 2. The reduction in the supply of reusable resources not only caused a reduction in daily usage, but also extended the total construction period. This is the same as when labor and mechanical equipment are constrained in actual engineering.

Figure 15 shows the comparison of the demand for resources 38 between calculation example 1 and calculation example 3. When the resource 38 is restricted, the resource usage decreases and the total construction period increases. This can happen when a material is entered in batches at different time periods.

4) Practical efficiency estimation

The above application process should be carried out as shown in Figure 16a in the engineering project. The whole process includes 9 tasks, of which 4 need to be processed manually, and the other 5 tasks are automatically completed by the computer. Because there are tasks to be completed manually and the actual situation of the project will affect the time required for the process, some assumptions need to be made to estimate the time required for the application process. These assumptions include:

Use the above data as a data basis.

The application process is performed 10 times, and the average duration of each time is calculated.

First estimate the time-consuming of each piece of data, and then summarize it into the time-consuming manual task.

Tasks that are automatically completed do not count as time-consuming.

Perform 5 progress optimizations.

Table 5 calculates the average time consumption of the application process in Figure 16a. Among them, 8 work package templates need to be established in task 1. During the creation of each work package template, adding basic information takes up to 5 minutes, and it takes 40 minutes in total. After that, the quota needs to be added, and the time is mainly consumed in adding resources and filling in the quota value. Assuming that it takes an average of 15 seconds to complete a resource, a total of 204 resources will take 51 minutes. After calculating the basic unit of each quota and the setting of applicable conditions, it takes 2 minutes, and it takes 38 minutes for a total of 19 quotas. Then the total time spent in task 1 is 129 minutes.

Task 2 can be considered as a two-step cycle. First, filter the components by category, and then add corresponding codes to all the components in the filtering result. During the verification process, 26 filterings were performed. Assuming that each time takes 1 minute, task 2 will total 26 minutes.

Task 5 includes 6 types of rules. Assuming that each type of rule takes 5 minutes, the total is 30 minutes.

The time required for task 7 is negligible and is conservatively set to 5 minutes.

In summary, considering that Task 1 and Task 5 only need to be completed once out of 100 similar projects, the average time required to complete one application process is estimated to be (129+30)÷10+26+5*5≈67 minutes

Table 5 Estimated time-consuming application of the patented method in actual projects

The control group used the general RCPSP-based problem modeling and schedule optimization process, as shown in Figure 16b. In the time-consuming estimation of the entire process, all calculations are considered to be automatically completed by the computer, and only the time for data entry is considered, so the total time-consuming obtained is relatively small compared with the real application scenario.

Task 1 is to establish a WBS with a process as a leaf node, and establish a contextual relationship. In this task, you can first establish a layer of WBS, and then establish a complete WBS and the pre- and post-relationship by copying. Since the relationship between WBS and the front and rear in this study is relatively simple, it will take about 5 minutes to consider.

Task 2 is to determine the total engineering quantity of related components in each process through manual screening. In this task, each process needs to be processed once, and it takes about 2 minutes to process one time, and 189 processes take a total of 398 minutes.

Task 3 is to add patterns for each process. Each mode in each quota in each process includes a unique duration, and then the corresponding cost can be obtained by combining the quota, total engineering quantity and resource unit price. Assuming that the data entry of a quota item takes 5s, it takes 7 minutes for 84 quota items. Since only the time for data entry is considered, it takes about 10s to complete the calculation of a mode. A quota needs to be set with 4 modes to ensure the accuracy of the result as much as possible, so a total of 9 quotas for each layer needs to consume 10*4*9=360s=6 minutes. Then task 3 consumes about 13 minutes.

Task 4 is the same as task 7 in Figure 15a, assuming it takes 5 minutes. In summary, the average completion time for the problem modeling and schedule optimization process of the control group is estimated to be 5+398+7+6+5*5=441 minutes.

Table 6 Time-consuming estimation of RCPSP general application process

Through comparison, it can be seen that the time consumed by the application process proposed by the present invention is conservatively estimated, which is less than 1/7 of the time consumed by the general RCPSP-based problem modeling and schedule optimization process. It is also worth noting that the model established by the latter is not as complicated as the former, for example, a process has only one set of resource requirements.

The foregoing embodiments are only used to illustrate the present invention, and each step can be changed. On the basis of the technical solution of the present invention, any improvement and equivalent transformation of individual steps based on the principles of the present invention should not be excluded Outside the protection scope of the present invention.

Claims

A method for automatically optimizing construction schedule resources based on a building information model is characterized by including the following steps:

1) Prepare building information models with building element categories and main resource categories, and enter or import required work package templates in the work package template database, and use the work package templates to generate work packages. There is a gap between each work package and the building components. Many-to-many association;

2) Data integration based on building component types and material types;

3) Based on the information model formed by data integration, using the constraints and objective function of RCPSP, the schedule resource optimization model is automatically generated and solved automatically to complete the optimization of the construction schedule resource.
5. The optimization method of claim 1, wherein in step 2), the data integration method is:

2.1) Based on the classification and coding of building component types, traverse each work package template to establish a coding tree;

2.2) Automatically associate work packages with building components;

2.3) Generating rule-based work package logic: After the work package is associated with the building component, the sequence logic is generated based on the rules. The definition of the rule is realized based on the attributes of the work package. The work package attributes include spatial location, component type and engineering specialty; The basic form of the rule is that the construction of a work package with a certain attribute or combination of attributes should be carried out before or later than the work package with another attribute or combination of attributes; after these rules are predefined, the relevant information can be found through the attributes of the work package. Work packages, and then automatically generate a logical sequence of work packages according to predefined rules.
2. The optimization method according to claim 2, characterized in that: in step 2.1), for each level of building component type coding in each work package template, if the level is not included in the coding tree, the node Add to the code tree, and associate the code tree node corresponding to the entire code with the work package template; the specific process is:

a) Traverse all work package templates to get template t, and set the code of template t to c;

b) Let the tree root node of the building component be curnode;

c) Traverse each layer of c to get the layer code n;

d) Judge whether the child node of the tree root node curnode contains the layer code n, if it does, go to step e); if it does not, create a child node n for the tree root node curnode and go to step e);

e) Assign the value of n in the child node to the tree root node curnode;

f) Judge whether n has the next layer, if so, return to step c), otherwise associate the template t with the tree root node curnode, and go to step g);

g) Repeat the above steps until there is no next template, and complete the establishment of the coding tree.
The optimization method according to claim 2, characterized in that: in the step 2.2), the automatic association of the work package and the building component includes 4 steps: the first matching of the building component and the work package template, the first time the building component and the work package template are matched Secondary matching, work package instantiation and work package reorganization;

The first matching is that each building component is matched step by step from the root node and is associated with the work package template associated in all the matched nodes; traverse the results of the first matching, and eliminate the associations that do not meet the material coding matching principle. Complete the second matching process;

The work package instantiation process is the process in which the work package template is divided according to the construction space where the corresponding building components are located; each group of building components after the division corresponds to a work package;

Work package reorganization is to traverse all building components and determine in turn whether their attributes meet the usage conditions of each quota of its associated work package; then get the specific quota combination corresponding to each building component, when a certain quota combination completely matches the building component , A new work package is generated.
5. The optimization method of claim 4, wherein the process of instantiating the work package is:

(1) Traverse all the work package templates to get the work package template t;

(2) Traverse all building components associated with the work package template t to obtain building component b;

(3) Obtain the construction area A of the building component b, and determine whether the construction area A has a corresponding work package. If yes, the work package corresponding to the facility area A is w, and go to step (4); if there is no work package, it is the construction area A creates the work package w of the work package template t and proceeds to step (4);

(4) Associate the building component b with the work package w, and judge whether there is the next building component, if there is, return to step (2), otherwise go to step (5);

(5) Judge whether there is the next work package template, if there is, return to step (1), otherwise end.
5. The optimization method of claim 4, wherein the work package reorganization process is:

(1) Traverse all work packages to get work package w;

(2) Traverse all building components related to work package w to obtain building component b;

(3) Traverse the quota, obtain the quota combination q that matches the building component b, and determine whether the quota combination set s includes the quota combination q; if it does, associate the quota combination q with the construction component b, otherwise, add the quota combination q to the quota In the combination set s, then the quota combination q is associated with the construction component b;

(4) Determine whether there is the next building component. If there is, return to step (2). Otherwise, create a work package for each quota combination q in the quota combination set s, and associate the work package with the quota combination q All building components are connected.
3. The optimization method according to claim 2, wherein in step 2.3), the logic generation of the work package specifically includes the following steps:

(1) Establish a work package attribute set φ;

(2) Traverse all work packages to get work package w;

(3) Traverse all the attributes of all work packages w to get the attribute t, judge whether the attribute t belongs to the attribute set φ, then go to step (4), otherwise, add the attribute t to the attribute set φ and go to step (4);

(4) Generate the association between the attribute t and the work package w, and judge whether there is still work package w, if there is, go back to step (2), otherwise go to step (5);

(5) Traverse all the rules to get the rule r. From the relevant attributes in the rule r, by querying the association between the attribute set φ and the work package set, the pre-order work package set s1 and the subsequent work package set s2 are obtained;

(6) Establish the sequence relationship between the pre-order work package set s1 and the subsequent work package set s2: add all the work reports in s1 to the pre-task set of all work packages in s2.
The optimization method according to claim 1, wherein in step 3), the constraint conditions include inter-process relationship constraints, milestone constraints, resource availability constraints, process internal constraints and resource mode constraints.
8. The optimization method of claim 8, wherein the five types of constraints are:

Relational constraints between processes: For process i, use T i to represent the start time SS i or end time SF i of process i. The difference between the key time attributes of each process has a lower limit and an upper limit, then the relationship is:

minLag≤T j -T i ≤maxLag

Among them, T i represents the SS i or SF i of the previous step i, T j represents the SS j or SF j of the subsequent step j, minLag represents the shortest interval, and maxLag represents the longest interval;

Milestone constraint: single process completion time constraints; single step completion time constraints mainly completion time for step i SF i programs, which must be earlier than the predetermined milestone M i, is expressed as:

SF i ≤M i ;

Resource availability constraint: at any point in time t, the total demand RD kt of resource k should be less than the total supply RS kt :

RD kt ≤RS kt

For reusable resources, the total demand should be equal to the maximum value of the sum of the demands of the processes in progress at all times before:

Among them, d ik represents the demand of process i for resource k, DA t ={i|SS i ≤t≤SF i }, which represents the set of processes in progress at time t;

For non-reusable resources, the total demand should be the sum of the demands of the various started processes:

Where ASA t = {i|t≥SS i }, which represents the set of processes that have started at time t;

Process internal constraints: the resource quantity qr ik of the labor resource k required for process i is inversely proportional to the process duration SDi:

qr ik ＝d ik SD i

Among them, the unit of qr ik is person multiplied by time, that is, 1 person needs to spend qr ik days, or qr ik individual needs 1 day to complete process i;

Resource mode constraints: Introduce an indicator variable MI iu that indicates whether the resource is selected, diku represents the demand for resource k in the u-th group of resources in process i; this variable is equal to 0 or 1, and satisfies the following formula:

MI iu is used to calculate d ik to ensure that all resource requirements belong to a certain group:
The optimization method according to claim 8 or 9, characterized in that: in the step 3), the RCPSP model takes the total construction period and the total cost as the objective function, and the specific calculation method is as follows:

(1) Total construction period

The total construction period TD is calculated using the following equation:

TD=max(SF i )-SS

Among them, SS is the start time;

(2) Total cost

Including direct cost and indirect cost, direct cost DC is the synthesis of the product of resource quantity and price, namely:

Among them, p k is the price of resource k;

The indirect cost IC includes loan interest, site lease, design cost, change cost and supervision cost; only the indirect cost related to the construction period is considered, and it is considered to be linearly related to the construction period:

IC=TD·dc

Among them, dc is the daily overhead consumption.