A building ventilation and purification product order delivery collaborative management method

Abstract

The application belongs to the technical field of order delivery, and discloses a building ventilation and purification product order delivery collaborative management method, a product performance prediction model is constructed based on an enterprise ERP system, a product is decomposed into a basic framework and matched with building performance requirements to determine a primary selected framework; the primary selected framework is combined with replaceable components to generate a candidate product, the candidate product is subjected to pareto screening to determine a feasible product, and the time and capital cost of production and installation are calculated; the optimal transportation mode of the product on the alternative path is determined based on voxel tensor boolean convolution operation, the transportation in-transit time and capital are calculated; all delivery schemes are enumerated and subjected to risk adjustment, and the optimal scheme is determined by taking the customer expected delivery time as a constraint. The precise product configuration is realized through a deep learning model, the optimal transportation mode is automatically determined through voxelized space detection, and the production, transportation and installation full-chain collaborative optimization is realized by introducing a risk adjustment mechanism, so that the scientificity and robustness of order delivery decision are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of order delivery technology, and more specifically to a collaborative management method for order delivery of building ventilation and purification products. Background Technology

[0002] Building ventilation and purification products are key equipment for ensuring indoor air quality and achieving energy conservation and a healthy, comfortable environment in buildings. They are widely used in various building scenarios, including commercial buildings, medical buildings, educational facilities, and residences. These products are typically composed of multiple components such as fan modules, filter modules, heat recovery modules, and control modules, and are characterized by complex product structures, flexible and diverse configurations, and interdependent performance parameters. In actual engineering projects, the selection and delivery of ventilation and purification products must simultaneously meet the building's personalized performance requirements (such as air volume, external static pressure, noise limits, filtration efficiency, and energy consumption limits), and coordinate the time and financial costs of multiple stages, including manufacturing, logistics, and on-site installation.

[0003] However, existing order delivery management methods have the following technical shortcomings: First, in the product selection process, traditional methods mainly rely on engineers' experience for manual matching or simple table lookup for performance comparison. This approach is not only inefficient but also struggles to handle the complex trade-offs between multiple performance dimensions, easily leading to either over-performance (resulting in wasted costs) or under-performance (resulting in failure to pass acceptance tests). Furthermore, the lack of systematic modeling of product configurability prevents the full consideration of the non-linear impact of different component combinations on overall performance.

[0004] Second, in terms of time and cost estimation in the production and installation stages, existing methods usually use fixed working hours or average costs for rough estimation, which fail to fully consider factors such as queuing time in front of the production line, processing dependencies between components, and critical paths in the installation process. This results in large deviations in the forecast of the project schedule and cost, making it difficult to support accurate delivery commitments.

[0005] Third, in the transportation process, building ventilation and purification products are often large in size and irregular in shape, and the transportation route is often subject to spatial constraints such as height restrictions, width restrictions, and turning radii. Traditional methods rely on manual measurement and experience to determine whether disassembly is necessary for transportation, lacking a systematic means of judging spatial passability, which can easily lead to misjudgments (such as being unable to pass without disassembly or increasing costs due to excessive disassembly).

[0006] Fourth, in terms of delivery solution decision-making, existing methods mostly involve independent decisions at each stage, lacking coordinated optimization across the entire chain of production, transportation, and installation. In particular, they fail to effectively quantify the inherent uncertainties at each stage (such as production fluctuations, transportation delays, and differences in installation efficiency), resulting in higher risks in actual implementation of delivery solutions and making it difficult to achieve optimal cost within the customer's expected delivery time.

[0007] Therefore, there is an urgent need for a method that can achieve collaborative management of product configuration, production scheduling, transportation planning and installation organization to improve the efficiency, accuracy and robustness of building ventilation and purification product order delivery. Summary of the Invention

[0008] To address the aforementioned problems, this invention provides a collaborative management method for the order delivery of building ventilation and purification products. This invention is achieved through the following technical solutions.

[0009] A collaborative management method for order delivery of building ventilation and purification products includes the following steps: S1, based on the enterprise ERP system, obtain several historical product data as a training set and construct a product performance prediction model; deconstruct the ventilation and purification products into several basic frameworks, and based on the product performance prediction model, determine the predicted value of the conservative performance vector of each basic framework, match it with the performance demand vector of the building ventilation system, and determine several preliminary frameworks. S2. For each initial selection frame, determine its replaceable component group. For each component in the replaceable component group, determine its available model set. Combine the initial selection frame with the available model set of each replaceable component to generate several candidate products. Screen the candidate products to determine several feasible products. Calculate the production time, production funds, installation time and installation funds required for each feasible product. S3. Based on the location of the transportation origin and destination, determine several alternative routes, determine the optimal transportation mode for each feasible product on each alternative route, and calculate the transportation time and transportation costs for each feasible product on each alternative route based on the optimal transportation mode. S4. Identify all possible delivery options, determine the risk-adjusted total time cost and total capital cost for each possible delivery option, and determine the optimal delivery option with the customer's expected delivery time as a constraint threshold.

[0010] As a further aspect of the present invention, the method for constructing the product performance prediction model in step S1 is as follows: Historical product data includes a set of product component models and corresponding actual test performance vectors. Using the set of product component models as input, an independent embedding layer is set for each component model. The component model is mapped to a low-dimensional embedding vector, concatenated, and then input into the neural network. Three hidden layers are set, and each hidden layer is followed by a Dropout layer with a dropout rate of 0.2. Mean squared error is used as the loss function, and the Adam optimizer is used to update the network parameters, outputting a product performance prediction vector. The conservative performance vector for each basic framework is determined as follows: Each basic framework contains several non-replaceable components. For each basic framework, its component model set includes the model numbers corresponding to the non-replaceable components and the empty model numbers corresponding to the replaceable components. Inputting the component model set corresponding to each basic framework into the product performance prediction model yields the predicted value of the conservative performance vector for each basic framework. , in, For the first Predicted values ​​of conservative performance vectors for each basic framework; For the first Conservative prediction of the rated air supply volume of the basic frame; For the first The basic framework can overcome the conservative prediction of external static pressure; For the first Conservative prediction of operating noise for the basic framework; For the first Conservative predictions of the filtration efficiency of the basic framework filters; For the first A conservative prediction of the upper limit of energy consumption per unit air supply volume in a basic framework.

[0011] As a further aspect of the present invention, the method for determining the initial frame in step S1 is as follows: Obtain the performance requirement vector of the building ventilation system Specifically: , in, To determine the total air supply volume requirement for the building, the functional zones of the building are identified using the building's BIM system, and the air change rate for each functional zone is determined. The air supply volume requirement for each functional zone is then determined by multiplying the air change rate by the volume. Finally, the air supply volume requirements of all functional zones are summed to obtain the total air supply volume requirement for the building. ; The total air supply volume requirement is At that time, the external static pressure that the building ventilation system needs to overcome; The upper limit for the operating noise of the building ventilation system; To meet the filter efficiency requirements; The upper limit of energy consumption per unit air volume; Calculate the difference between the predicted value of each base frame conservative performance vector and the building ventilation system performance requirement vector for each corresponding component: , in, For the first The difference vector between the basic framework and the performance requirement vector of the building ventilation system; , , , , ; For the Each basic framework will be removed if any of the following conditions are met: ,and ; ,and ; ,and ; ,and ; ,and ; in , , , and These are the tolerance thresholds for rated air volume, external static pressure that can be overcome, operating noise, filter efficiency, and energy consumption per unit air volume, respectively. For the remaining basic frameworks after elimination through the tolerance threshold, calculate the overall matching score: , in, For the first The overall matching score of the basic framework; , , , and They are respectively , , , and The corresponding weighted score; It is the absolute value; Based on overall matching score The basic frameworks are sorted from low to high, and the top few basic frameworks with the lowest scores are selected as the initial frameworks.

[0012] As a further aspect of the present invention, the method for determining the feasible product in step S2 is as follows: Each initial frame is combined with all available models of each component in its replaceable component group by Cartesian product to generate several candidate products. For each candidate product, the component models of all its constituent parts are input into the product performance prediction model to obtain the comprehensive performance vector of each candidate product: , in, For the first The comprehensive performance vector of each candidate product; For the first Rated air volume of each candidate product; For the first The external static pressure that each candidate product can overcome; For the first Operating noise of each candidate product; For the first Filter efficiency of each candidate product; For the first Upper limit of energy consumption per unit air volume for each candidate product; Calculate the performance margin vector for each candidate product. : , in For the first The performance margin vector of each candidate product; like If any component in the set is less than zero, it is removed to obtain the first candidate set of feasible products; For each product in the first candidate set of feasible products, based on the Pareto non-dominated sorting method, a dominated counter is set for each product. All products are traversed, and all products with a dominated count of zero are added to the Pareto front. All products on the Pareto front are used as the second candidate set of feasible products. For any two products A and B, if the performance margin vector of A is greater than or equal to the performance margin vector of B in all dimensions, it indicates that A dominates B, that is, B is dominated by A. For each product in the second candidate set of feasible products, calculate the comprehensive matching score: , in, For the second candidate set The overall matching score of each product; Will The products in the second candidate set are sorted from highest to lowest score, and the top few products with the highest scores are selected as feasible products.

[0013] As a further aspect of the present invention, for each feasible product: The required production time is calculated by obtaining the standard production time of each component of the feasible product and the processing dependencies between components from the enterprise ERP system, obtaining the queue of orders to be produced in front of the corresponding production line of each component from the enterprise production scheduling module, and determining the queuing waiting time of each component in front of the corresponding production line. The queuing waiting time is equal to the sum of the cumulative processing time of all orders to be produced before that component. The queuing waiting time of each component is added to its own standard production time to obtain the completion time of that component. According to the processing dependencies between components, the maximum value is taken in parallel for components without dependencies, and the values ​​are accumulated serially for components with dependencies to obtain the required production time of the feasible product. Calculate the funds required for its production, obtain the standard production funds required for each component of the feasible product from the enterprise's ERP system, and sum them up to obtain the funds required for production. Calculate the installation time, obtain the standard working hours of each installation process of the feasible product from the installation history database of the enterprise ERP system, construct a directed acyclic graph of installation processes based on the sequential dependencies between processes, determine the longest path from the start process to the end process as the critical path, and use the sum of the working hours of each process on the critical path as the installation time. Calculate the funds required for its installation. Obtain the number of workers and skill levels, equipment types and usage time required for each installation process of the feasible product from the enterprise ERP system. Calculate the labor cost according to the differentiated labor rate for different skill levels. Calculate the equipment amortization cost according to the equipment type and usage time. Add them together to obtain the funds required for installation.

[0014] As a further aspect of the present invention, in step S3, the method for determining the optimal transportation method for each feasible product on each alternative route is as follows: Obtain spatial constraint point information for each candidate path, and for each spatial constraint point, construct a local voxel tensor based on the spatial constraint parameters. ,in For the first The first of the alternative paths The local voxel tensor of a spatially constrained point, wherein the local voxel tensor is a three-dimensional binary tensor, and a voxel value of 1 indicates that passage is allowed, and a voxel value of 0 indicates that an obstacle exists; The overall dimensions of each feasible product are converted into an overall voxel tensor. , For the first The overall body element tensor of a feasible product; According to the product structure tree, it is decomposed into an L-level set of sub-component voxel tensors: , For the first The first feasible product The set of voxel tensors of sub-components, L is The maximum number of disassembly stages for a feasible product; Based on voxel Boolean convolution, the intersection volume ratio of the entire voxel tensor of the feasible product with the local voxel tensor of each alternative path is calculated. If all ratios are equal to 1, the optimal transportation method is determined to be original container transportation; if any ratio is less than 1, then: Calculate the intersection volume ratio of each sub-component voxel tensor in the voxel tensor set of each sub-component of the feasible product with each local voxel tensor on the alternative path level. If there is a level where all the calculated intersection volume ratios are equal to 1, then the transportation method to disassemble the sub-component at that level is determined to be the optimal transportation method. If, even at the maximum number of disassembly levels, there is still a case where the intersection volume ratio is less than 1, then it is determined that the alternative route cannot be used for transportation.

[0015] As a further aspect of the present invention, in step S3, the transit time of each feasible product on each alternative route based on the optimal transportation mode includes logistics time and station operation time; the funds required for transportation include logistics transportation funds and station operation funds for the corresponding transportation mode.

[0016] As a further aspect of the present invention, step S4 includes the following: Enumerate all possible delivery plans. Each delivery plan consists of a feasible product, a transportation route, and the optimal transportation method for the current feasible product and transportation route. Specifically: , in, For the first One delivery plan, For the first One feasible product, For the first Transportation routes, For the first The first feasible product in the first The optimal mode of transport on the transport route; for : Its total time cost ,in For the first Production time for each delivery plan For the first The transit time for each delivery solution. For the first Installation time for each delivery plan; Its total cost of capital ,in For the first The funds required for the production of each delivery plan For the first The funds required for transportation of each delivery plan. For the first Funding required for the installation of each delivery plan; Risk adjustments are made to the total time cost and total capital cost, specifically as follows: , , in, and These are the risk-adjusted values ​​for total time cost and total capital cost, respectively. and These are the time risk preference coefficient and the cost risk preference coefficient, respectively. and These are the standard deviations of total time cost and total cost of capital, respectively. , in , and These are the standard deviations of time for the production stage, the transportation stage, and the installation stage, respectively; and , , ; , in , and These are the standard deviations of costs in the production, transportation, and installation stages, respectively; and , , ; , and These are the preset coefficients of variation for the production, transportation, and installation stages, respectively. , , ; Based on the customer's expected delivery time span Select the one that meets the requirements from all delivery options. The feasible solutions constitute a set of feasible solutions; If the set of feasible solutions is not empty, then choose from it. The smallest possible solution is selected as the optimal solution; if the set of feasible solutions is empty, then the solution with the smallest possible solution is selected. The minimum solution is selected as the optimal solution, and a delayed delivery notification is sent to the customer. The calculation is then performed. and The difference is used as the delay time span.

[0017] The beneficial effects of this invention are as follows: First, by constructing a product performance prediction model based on deep learning, and combining a two-stage strategy of conservative performance matching and Pareto selection within a basic framework, precise matching of ventilation and purification product configuration with building performance requirements is achieved, avoiding performance over- or under-performance caused by experience-based selection. Second, in the production and installation stages, by considering queuing, component dependencies, and the critical path method to refine the calculation of time and capital costs, and by innovatively introducing voxel tensor Boolean convolution operations in the transportation stage, the optimal transportation method is automatically determined by calculating the intersection volume ratio of the product with spatial constraint points through step-by-step disassembly, fully considering the station operation costs and time of the entire chain of disassembly, loading, transportation, unloading, and reloading. Finally, by enumerating all delivery options and introducing a risk adjustment mechanism based on the coefficient of variation, the option with the lowest risk-adjusted cost or shortest construction period is selected with the customer's expected delivery time as a constraint, achieving collaborative optimization and risk-controllable decision-making in multiple stages of production, transportation, and installation, significantly improving the scientific nature and robustness of order delivery. Attached Figure Description

[0018] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description of the specific embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 Flowchart of a collaborative management method for order delivery of building ventilation and purification products Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] like Figure 1 As shown, a collaborative management method for order delivery of building ventilation and purification products includes the following steps: S1. Based on the enterprise ERP system, acquire some historical product data as a training set to build a product performance prediction model; deconstruct the ventilation and purification product into several basic frameworks, determine the predicted value of the conservative performance vector of each basic framework based on the product performance prediction model, match it with the performance requirement vector of the building ventilation system, and determine several preliminary frameworks.

[0022] Preferably, in step S1, the method for constructing the product performance prediction model is as follows: Historical product data includes a set of product component models and corresponding actual test performance vectors. Using the set of product component models as input, an independent embedding layer is set for each component model. The component model is mapped to a low-dimensional embedding vector, concatenated, and then input into the neural network. Three hidden layers are set, and each hidden layer is followed by a Dropout layer with a dropout rate of 0.2. Mean squared error is used as the loss function, and the Adam optimizer is used to update the network parameters, outputting a product performance prediction vector. The conservative performance vector for each basic framework is determined as follows: Each basic framework contains several non-replaceable components. For each basic framework, its component model set includes the model numbers corresponding to the non-replaceable components and the empty model numbers corresponding to the replaceable components. Inputting the component model set corresponding to each basic framework into the product performance prediction model yields the predicted value of the conservative performance vector for each basic framework. , in, For the first Predicted values ​​of conservative performance vectors for each basic framework; For the first Conservative prediction of the rated air supply volume of the basic frame; For the first The basic framework can overcome the conservative prediction of external static pressure; For the first Conservative prediction of operating noise for a basic framework; For the first Conservative predictions of the filtration efficiency of the basic framework filters; For the first A conservative prediction of the upper limit of energy consumption per unit air supply volume in a basic framework.

[0023] Specifically, historical product data is first extracted from the enterprise's ERP system. Each historical product data record contains a set of component models included in a ventilation and purification product, as well as the actual test performance vector of the product. The actual test performance vector includes key indicators such as rated air volume, external static pressure, operating noise, filter efficiency, and upper limit of energy consumption per unit air volume.

[0024] The model training process is as follows: A set of component models for a product is taken as input. To avoid information loss caused by discrete model encoding, an independent embedding layer is created for each possible component model. Each embedding layer is a learnable parameter matrix that maps high-dimensional, sparse one-hot encoded model vectors to low-dimensional, dense embedding vectors. For example, a wind turbine model "EC-250-4T" is mapped to a floating-point vector of length 64.

[0025] The embedding vectors of all components are concatenated to form a global feature vector that fully represents the product's component combination. This global feature vector is then input into a deep neural network containing three fully connected hidden layers. The number of neurons in each hidden layer is as follows: 256 in the first layer, 128 in the second layer, and 64 in the third layer. To prevent overfitting and improve generalization ability, a Dropout layer with a dropout rate of 0.2 is connected after each hidden layer. In each training iteration, the Dropout layer randomly sets the output of 20% of the neurons to zero, forcing the network to learn a more robust feature representation.

[0026] The network's output layer is a five-dimensional vector, corresponding to the predicted product performance vector. The model training uses mean squared error as the loss function to measure the difference between the predicted performance vector and the actual detected performance vector. The Adam optimizer is used to adaptively adjust the network's learning rate. The embedding layer parameters and fully connected layer weights in the network are iteratively updated through the backpropagation algorithm until the loss function converges. After training, the model can quickly and accurately output the corresponding product performance prediction vector based on any set of input component models.

[0027] Each basic frame contains a set of non-replaceable components (such as the housing structure, heat exchange core, basic control circuit, etc.), while reserving positions for several replaceable components (such as different fans and filters, etc.). The specific models of these replaceable components in the basic frame are not specified yet, and are represented by "empty model".

[0028] For the nth basic framework, construct its corresponding component model set, which includes: the specific models of all non-replaceable components, and the empty models at all replaceable component positions. Input this set into the product performance prediction model that has been trained above. Since the replaceable component positions are "empty models", the model root will output the basic performance of the basic framework when no specific replaceable components are configured, that is, the predicted value of the conservative performance vector.

[0029] This conservative prediction represents the minimum performance level that the underlying framework itself can contribute. In this way, the product design space can be reduced from a huge portfolio of specific products to a limited number of underlying frameworks, significantly improving the efficiency of the initial selection.

[0030] Preferably, in step S1, the method for determining the initial framework is as follows: Obtain the performance requirement vector of the building ventilation system Specifically: , in, To determine the total air supply volume requirement for the building, the functional zones of the building are identified using the building's BIM system, and the air change rate for each functional zone is determined. The air supply volume requirement for each functional zone is then determined by multiplying the air change rate by the volume. Finally, the air supply volume requirements of all functional zones are summed to obtain the total air supply volume requirement for the building. ; The total air supply volume requirement is At that time, the external static pressure that the building ventilation system needs to overcome; The upper limit for the operating noise of the building ventilation system; To meet the filter efficiency requirements; The upper limit of energy consumption per unit air volume; Calculate the difference between the predicted value of each base frame conservative performance vector and the building ventilation system performance requirement vector for each corresponding component: , in, For the first The difference vector between the basic framework and the performance requirement vector of the building ventilation system; , , , , ; For the Each basic framework will be removed if any of the following conditions are met: ,and ; ,and ; ,and ; ,and ; ,and ; in , , , and These are the tolerance thresholds for rated air volume, external static pressure that can be overcome, operating noise, filter efficiency, and energy consumption per unit air volume, respectively. For the remaining basic frameworks after elimination through the tolerance threshold, calculate the overall matching score: , in, For the first The overall matching score of the basic framework; , , , and They are respectively , , , and The corresponding weighted score; It is the absolute value; Based on overall matching score The basic frameworks are sorted from low to high, and the top few basic frameworks with the lowest scores are selected as the initial frameworks.

[0031] Specifically, key parameters are extracted from the building BIM system to construct a performance requirement vector for the building ventilation system. , for Based on the BIM system, the geometric dimensions of each functional area of ​​the building (such as office area, meeting room, corridor, equipment room, etc.) are obtained, the volume of each functional area (unit: cubic meters) is calculated, and the number of air changes per hour (unit: times / hour) required for each functional area is determined according to relevant national ventilation design specifications or specific project requirements. The volume of each functional area is multiplied by the number of air changes to obtain the air supply volume requirement of that functional area. The air supply volume requirements of all functional areas are added together to obtain the total air supply volume requirement of the building.

[0032] for Based on the duct network model in the BIM system (including the resistance characteristics of components such as duct length, number of bends, tees, reducers, dampers, and terminal air outlets), with a total air supply volume requirement of... Under the operating conditions, the total resistance loss from the outlet of the ventilation and purification product to the most unfavorable terminal air outlet is calculated using duct resistance calculation methods (such as the equal friction method or the static pressure recovery method). This total resistance loss is the external static pressure that the building ventilation system needs to overcome.

[0033] for Based on the usage of each functional area of ​​the building (such as offices, conference rooms, hospital operating rooms, residences, etc.), the upper limit of the operating noise of the building's ventilation system is extracted from the building acoustic design parameters associated with the building BIM system or national sound environment standards (such as GB3096-2008). Usually, the noise limit corresponding to the most stringent area is taken.

[0034] for Based on the air quality level of the area where the building is located and the requirements of the building's function for indoor air cleanliness (such as ordinary comfort air conditioning, hospital clean rooms, museum artifact preservation, etc.), the filtration efficiency requirements for suspended particulate matter (such as PM2.5 and PM10) are obtained from the building BIM system or project technical specifications.

[0035] for Based on the green building design standards or energy-saving targets of the building (such as LEED certification, China Green Building Evaluation Standard), the upper limit of the allowable energy consumption per unit air volume of ventilation and purification products is extracted.

[0036] Each component in the performance requirement vector of a building ventilation system can be calculated using information obtained from the building BIM system and relevant regulations.

[0037] The predicted value of the conservative performance vector of each basic frame is calculated, and the difference between the predicted value and the performance requirement vector of the building ventilation system at each corresponding component is calculated. A tolerance threshold mechanism is introduced to eliminate basic frames with excessively large gaps in key performance dimensions that cannot be compensated by subsequent replacement components. The specific meaning is as follows: This represents the difference between the total air supply volume requirement and the conservative air supply volume of the basic frame. This indicates that the air supply to the basic frame is insufficient. If the air supply is insufficient and the gap exceeds the tolerance limit, it will be rejected.

[0038] This represents the difference between the required external static pressure and the conservative static pressure of the foundation frame. This indicates a performance gap in the external static pressure of the basic frame. If the static pressure outside the machine is insufficient and the gap exceeds the tolerance, it should be rejected.

[0039] The basic framework's conservative noise level exceeds the noise ceiling. This indicates that the noise level of the basic framework itself has exceeded the limit. If the noise level exceeds the upper limit and the excess exceeds the tolerance, the device will be removed.

[0040] This represents the difference between the required filtration efficiency and the conservative filtration efficiency of the underlying framework. This indicates a gap in the filtering performance of the underlying framework. If the filter is not efficient enough and the gap exceeds the tolerance, it should be discarded.

[0041] This represents the amount by which the basic framework's conservative unit energy consumption exceeds the energy consumption ceiling. This indicates that the energy efficiency of the basic framework no longer meets energy-saving requirements. If the energy consumption per unit exceeds the upper limit and the excess is greater than the tolerance level, it will be excluded.

[0042] Specifically, the tolerance threshold is determined based on experience. for 10%, , , =5%, .

[0043] For the remaining basic frameworks after elimination through the tolerance threshold, calculate the comprehensive matching score. , , , and These are the weighting coefficients for the five performance components, satisfying... The weights can be adjusted according to the actual focus of the construction project. For example, for green building projects, the weights can be increased. (Energy weight); for projects located in hospitals or precision laboratories, the weight can be increased. (Filtering efficiency weight); For residential projects, the weight can be increased. Noise weighting), in the absence of special preferences, can be used with equal weights (0.2 each).

[0044] The significance of taking the absolute value is that both insufficient performance (positive gap) and excessive performance (negative gap) will result in cost or efficiency diseconomies, and therefore both are considered as deviations in the degree of matching.

[0045] S2. For each initial selection frame, determine its replaceable component group. For each component in the replaceable component group, determine its available model set. Combine the initial selection frame with the available model set of each replaceable component to generate several candidate products. Screen the candidate products to determine several feasible products, and calculate the production time, production funds, installation time and installation funds required for each feasible product.

[0046] Preferably, in step S2, the method for determining the feasible product is as follows: Each initial frame is combined with all available models of each component in its replaceable component group by Cartesian product to generate several candidate products. For each candidate product, the component models of all its constituent parts are input into the product performance prediction model to obtain the comprehensive performance vector of each candidate product: , in, For the first The comprehensive performance vector of each candidate product; For the first Rated air volume of each candidate product; For the first The external static pressure that each candidate product can overcome; For the first Operating noise of each candidate product; For the first Filter efficiency of each candidate product; For the first Upper limit of energy consumption per unit air volume for each candidate product; Calculate the performance margin vector for each candidate product. : , in For the first The performance margin vector of each candidate product; like If any component in the set is less than zero, it is removed to obtain the first candidate set of feasible products; For each product in the first candidate set of feasible products, based on the Pareto non-dominated sorting method, a dominated counter is set for each product. All products are traversed, and all products with a dominated count of zero are added to the Pareto front. All products on the Pareto front are used as the second candidate set of feasible products. For any two products A and B, if the performance margin vector of A is greater than or equal to the performance margin vector of B in all dimensions, it indicates that A dominates B, that is, B is dominated by A. For each product in the second candidate set of feasible products, calculate the comprehensive matching score: , in, For the second candidate set The overall matching score of each product; Will The products in the second candidate set are sorted from highest to lowest score, and the top few products with the highest scores are selected as feasible products.

[0047] Specifically, each candidate product is uniquely determined by "initial selection framework + a specific model in each replaceable component group". For each generated candidate product, the model set of all its constituent components (including the fixed model of the non-replaceable component in the initial selection framework and the specific model selected in each replaceable component group) is input into the product performance prediction model constructed in step S1, and the comprehensive performance vector of the candidate product is output.

[0048] To assess how well each candidate product meets the actual needs of the building, its performance margin vector is calculated.

[0049] Each component of the performance margin vector represents the extent to which the candidate product exceeds (positive) or falls short (negative) the building's requirements in that dimension. For air volume and external static pressure, a positive value indicates a margin; for noise, a positive value indicates noise is below the upper limit (i.e., a margin); for filtration efficiency, a positive value indicates efficiency is above the requirement; and for energy consumption, a positive value indicates energy consumption is below the upper limit (i.e., a margin).

[0050] like If any component in the formula is less than zero, it indicates that the candidate product cannot meet the basic performance requirements of the building in that dimension, and it is directly eliminated. After traversing all candidate products, the remaining products after filtering by this rule constitute the first candidate set of feasible products.

[0051] In the first candidate set of feasible products, there may be multiple products that meet all hard constraints, but different products have their own advantages and disadvantages in different performance dimensions (for example, the first product has a larger air volume but higher noise, while the second product has lower noise but smaller air volume). In order to select products that are competitive in multiple objectives, Pareto non-dominated ranking is introduced.

[0052] The specific steps are as follows: 1. Set a "dominated counter" for each product in the first candidate set, with an initial value of 0.

[0053] 2. Traverse all products in the first candidate set. For any two different products A and B, compare their performance margin vectors. If the performance margin vector of A is greater than or equal to the performance margin vector of B in all dimensions, it indicates that A dominates B, that is, B is dominated by A. Increment the dominance counter of B by 1.

[0054] 3. After completing all pairwise comparisons, extract all products with a dominated counter of zero. These products constitute the first Pareto front, and no other product is inferior to them in all dimensions.

[0055] 4. All products on the Pareto front are used as the second candidate set of feasible products.

[0056] Pareto screening eliminated inferior candidate products that were completely outperformed by another product in all performance dimensions, and retained "non-dominated" products that achieved a balance among multiple performance objectives.

[0057] For each product in the second candidate set of feasible products, calculate its comprehensive matching score. This score is used to quantify the total experience margin of the product in each performance dimension. The higher the score, the more abundant the comprehensive performance. Select the top few products with the highest scores as feasible products.

[0058] Preferably, for each feasible product: Calculate the required production time by obtaining the standard production time of each component of the feasible product and the processing dependencies between components from the enterprise ERP system, obtaining the queue of pending orders in front of the corresponding production line of each component from the enterprise production scheduling module, determining the queuing waiting time of each component in front of the corresponding production line, which is equal to the sum of the cumulative processing time of all pending orders in front of that component; add the queuing waiting time of each component to its own standard production time to obtain the completion time of that component; according to the processing dependencies between components, take the maximum value in parallel for components without dependencies, and accumulate the values ​​serially for components with dependencies to obtain the required production time of the feasible product.

[0059] Calculate the funds required for its production, obtain the standard production funds required for each component of the feasible product from the enterprise's ERP system, and sum them up to obtain the funds required for production.

[0060] Calculate the installation time by obtaining the standard working hours of each installation process of the feasible product from the installation history database of the enterprise ERP system. Construct a directed acyclic graph of the installation process based on the sequential dependencies between the processes, determine the longest path from the start process to the end process as the critical path, and use the sum of the working hours of each process on the critical path as the installation time.

[0061] Calculate the funds required for its installation. Obtain the number of workers and skill levels, equipment types and usage time required for each installation process of the feasible product from the enterprise ERP system. Calculate the labor cost according to the differentiated labor rate for different skill levels. Calculate the equipment amortization cost according to the equipment type and usage time. Add them together to obtain the funds required for installation.

[0062] S3. Based on the location of the transportation origin and destination, determine several alternative routes, determine the optimal transportation mode for each feasible product on each alternative route, and calculate the transportation time and transportation costs for each feasible product on each alternative route based on the optimal transportation mode.

[0063] Preferably, in step S3, the method for determining the optimal transportation method for each feasible product on each alternative route is as follows: Obtain spatial constraint point information for each candidate path, and for each spatial constraint point, construct a local voxel tensor based on the spatial constraint parameters. ,in For the first The first of the alternative paths The local voxel tensor of each spatial constraint point is a three-dimensional binary tensor. A voxel value of 1 indicates that passage is allowed, and a voxel value of 0 indicates that an obstacle exists. The overall dimensions of each feasible product are converted into an overall voxel tensor. , For the first The overall body element tensor of a feasible product; According to the product structure tree, it is decomposed into an L-level set of sub-component voxel tensors: , For the first The first feasible product The set of voxel tensors of sub-components, L is The maximum number of disassembly stages for a feasible product; Based on voxel Boolean convolution, the intersection volume ratio of the entire voxel tensor of the feasible product with the local voxel tensor of each alternative path is calculated. If all ratios are equal to 1, the optimal transportation method is determined to be original container transportation; if any ratio is less than 1, then: Calculate the intersection volume ratio of each sub-component voxel tensor in the voxel tensor set of each sub-component of the feasible product with each local voxel tensor on the alternative path level. If there is a level where all the calculated intersection volume ratios are equal to 1, then the transportation method to disassemble the sub-component at that level is determined to be the optimal transportation method. If, even at the maximum number of disassembly levels, there is still a case where the intersection volume ratio is less than 1, then it is determined that the alternative route cannot be used for transportation.

[0064] Specifically, firstly, based on the geographical locations of the transportation origin (the location of the production plant) and the destination (the location of the construction project), and combined with electronic maps and actual road network information, several alternative transportation routes are determined. Each alternative route consists of a series of continuous spatial nodes such as highways, bridges, tunnels, culverts, turns, height restriction poles, and toll stations.

[0065] For each candidate path, identify all spatial constraint points on that path (i.e., key locations that may restrict the size of transported objects, such as height restriction barriers, tunnel entrances, narrow bridges, right-angle turns, etc.). For each spatial constraint point, construct its local voxel tensor based on its spatial constraint parameters (including clearance height, clearance width, turning radius, and channel shape, etc.). .

[0066] The local voxel tensor is a three-dimensional binary array (i.e., a voxel mesh). Its construction method is as follows: A local three-dimensional coordinate system is established with the geometric center of the spatial constraint point as the origin. The passage space around the constraint point is discretized into several cubic voxels of the same size (e.g., each voxel represents a 5cm × 5cm × 5cm actual space). For each voxel, if an object is allowed to pass through at that location (i.e., within the clearance range), the voxel value is set to 1; if an obstacle exists at that location (such as a tunnel wall, height restriction beam, or road shoulder), the voxel value is set to 0. Therefore, The spatial shape through which the spatial constraint point passes is fully characterized in the form of a three-dimensional binary tensor.

[0067] For each feasible product, first obtain its overall dimensions (length, width, and height, in centimeters or meters). Discretize these overall dimensions at the same voxel resolution (e.g., 5 cm) as the local voxel tensor to construct the overall voxel tensor.

[0068] Furthermore, utilizing the product structure tree, this feasible product is disassembled level by level. The product structure tree, according to the assembly hierarchy, is broken down from the complete machine (level 0) into sub-components, sub-assemblies, and finally the smallest non-disassembleable part. Let the maximum number of disassembleable levels for this product be... (For example, L=3 indicates that it can be disassembled down to the 3rd level of sub-components). Obtain the dimensions of all sub-components at each level, construct the voxel tensor of each sub-component, and form the voxel tensor set of sub-components at that level. For example, the 1st level sub-component may include "fan module", "filter module", "control box", etc.; the 2nd level sub-component may further disassemble the "fan module" into "impeller", "motor", "volute", etc.

[0069] Perform a 3D Boolean convolution between the product voxel tensor and the local voxel tensors of the path space constraint points: at each possible position offset, calculate the number of primes in the intersection of the product voxel tensor and the path voxel tensor, and divide it by the total number of primes in the product voxel tensor to obtain the intersection volume ratio. If it is equal to 1, it means that all voxels of the product are located within the path's allowed passage space at all possible position offsets (i.e., no interference at all); if it is not equal to 1, it means that there is interference.

[0070] The determination process is as follows: 1. Try shipping in the original packaging: For the first Whole voxel tensor of a feasible product , in turn with the Local voxel tensors at each spatial constraint point on the alternative paths ( Perform voxel Boolean convolution operations to calculate the intersection volume ratio at each spatial constraint point. If the intersection volume ratio of all constraint points is equal to 1, then the feasible product is determined to be at the [missing value]. The optimal transportation method on the alternative routes is original container transportation (i.e., no disassembly required, shipment as a whole).

[0071] 2. If interference exists, attempt disassembly and transportation step by step: If, during the above process, the intersection volume ratio of any constraint point is less than 1, it indicates that the entire machine cannot pass through that spatial constraint point. In this case, starting from the first-level sub-component, proceed with... For each sub-component voxel tensor in the set, repeat the Boolean convolution operation with all spatial constraint points. If the intersection volume ratio of all sub-components at all constraint points is equal to 1, then the transportation method of disassembling the product into sub-components at that level is determined to be the optimal transportation method. That is, after disassembling the feasible product into several sub-components at level l, they can be transported separately and can pass through the entire path without interference.

[0072] 3. Progress through the disassembly levels until the maximum number of levels is reached. If interference still exists at level l (i.e., the volume ratio of the intersection of a sub-component at a certain spatial constraint point is less than 1), then continue to try the next level, gradually increasing the disassembly precision.

[0073] 4. Determine if the route is infeasible: If, after traversing to the maximum disassembly level L (i.e., the product has been completely disassembled into its smallest parts), there is still a sub-component whose intersection volume ratio at a certain constraint point is less than 1, then the feasible product cannot be transported via the t-th alternative route. In this case, the (product, route) combination should be marked as infeasible.

[0074] Preferably, in step S3, the transit time of each feasible product on each alternative route based on the optimal transportation mode includes logistics time and station operation time; the transportation funds include logistics transportation funds and station operation funds for the corresponding transportation mode.

[0075] Specifically, if the original packaging is used for transportation, the station operation time includes the loading and unloading time of the entire feasible product. If the transportation method of disassembling to a certain sub-component is used, the station operation time includes the disassembly time of disassembling to that sub-component, the reassembly time after disassembly to that sub-component, and the loading and unloading time of each unit after disassembly.

[0076] If the original container is used for transportation, the station operation funds include the loading and unloading funds for the entire feasible product; if the transportation method of disassembling to a certain sub-component is used, the station operation funds include the disassembly operation funds for disassembling to that sub-component, the loading operation funds for each unit after disassembly, the unloading operation funds for each unit after disassembly, and the reassembly operation funds after disassembling to that sub-component.

[0077] S4. Identify all possible delivery options, determine the risk-adjusted total time cost and total capital cost for each possible delivery option, and determine the optimal delivery option with the customer's expected delivery time as a constraint threshold.

[0078] Preferably, step S4 includes the following: Enumerate all possible delivery plans. Each delivery plan consists of a feasible product, a transportation route, and the optimal transportation method for the current feasible product and transportation route. Specifically: , in, For the first One delivery plan, For the first One feasible product, For the first Transportation routes, For the first The first feasible product in the first The optimal mode of transport on the transport route; for : Its total time cost ,in For the first Production time for each delivery plan For the first The transit time for each delivery solution. For the first Installation time for each delivery plan; Its total cost of capital ,in For the first The funds required for the production of each delivery plan For the first The funds required for transportation of each delivery plan. For the first Funding required for the installation of each delivery plan; Risk adjustments are made to the total time cost and total capital cost, specifically as follows: , , in, and These are the risk-adjusted values ​​for total time cost and total capital cost, respectively. and These are the time risk preference coefficient and the cost risk preference coefficient, respectively. and These are the standard deviations of total time cost and total cost of capital, respectively. , in , and These are the standard deviations of time for the production stage, the transportation stage, and the installation stage, respectively; and , , ; , in , and These are the standard deviations of costs in the production, transportation, and installation stages, respectively; and , , ; , and These are the preset coefficients of variation for the production, transportation, and installation stages, respectively. , , ; Based on the customer's expected delivery time span Select the one that meets the requirements from all delivery options. The feasible solutions constitute a set of feasible solutions; If the set of feasible solutions is not empty, then choose from it. The smallest possible solution is selected as the optimal solution; if the set of feasible solutions is empty, then the solution with the smallest possible solution is selected. The minimum solution is selected as the optimal solution, and a delayed delivery notification is sent to the customer. The calculation is then performed. and The difference is used as the delay time span.

[0079] Specifically, by enumerating feasible products, transportation routes, and optimal transportation methods to construct delivery plans, and introducing a risk adjustment mechanism based on the coefficient of variation, the total time cost and total capital cost after risk adjustment are calculated in combination with the risk preference coefficient. Thus, under the constraint of meeting the customer's expected delivery time, the plan with the lowest risk-adjusted cost is selected first, or the plan with the shortest time is selected when the project schedule cannot be met and the delay span is quantified. This achieves intelligent decision-making for order delivery that takes into account project schedule constraints, optimal cost, and controllable risk.

[0080] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A collaborative management method for order delivery of building ventilation and purification products, characterized in that, Includes the following steps: S1. Based on the enterprise ERP system, obtain some historical product data as a training set to build a product performance prediction model; deconstruct the ventilation and purification product into several basic frameworks, and based on the product performance prediction model, determine the predicted value of the conservative performance vector of each basic framework, match it with the performance requirement vector of the building ventilation system, and determine several preliminary frameworks. S2. For each initial selection frame, determine its replaceable component group. For each component in the replaceable component group, determine its available model set. Combine the initial selection frame with the available model set of each replaceable component to generate several candidate products. Screen the candidate products to determine several feasible products. Calculate the production time, production funds, installation time and installation funds required for each feasible product. S3. Based on the location of the transportation origin and destination, determine several alternative routes, determine the optimal transportation mode for each feasible product on each alternative route, and calculate the transportation time and transportation costs for each feasible product on each alternative route based on the optimal transportation mode. S4. Identify all possible delivery options, determine the risk-adjusted total time cost and total capital cost for each possible delivery option, and determine the optimal delivery option with the customer's expected delivery time as a constraint threshold.

2. The collaborative management method for order delivery of building ventilation and purification products according to claim 1, characterized in that, In step S1, the method for constructing the product performance prediction model is as follows: Historical product data includes a set of product component models and corresponding actual test performance vectors. Using the set of product component models as input, an independent embedding layer is set for each component model. The component model is mapped to a low-dimensional embedding vector, concatenated, and then input into the neural network. Three hidden layers are set, and each hidden layer is followed by a Dropout layer with a dropout rate of 0.

2. Mean squared error is used as the loss function, and the Adam optimizer is used to update the network parameters, outputting a product performance prediction vector. The conservative performance vector for each basic framework is determined as follows: Each basic framework contains several non-replaceable components. For each basic framework, its component model set includes the model numbers corresponding to the non-replaceable components and the empty model numbers corresponding to the replaceable components. Inputting the component model set corresponding to each basic framework into the product performance prediction model yields the predicted value of the conservative performance vector for each basic framework. ， in, For the first Predicted values ​​of conservative performance vectors for each basic framework; For the first Conservative prediction of the rated air supply volume of the basic frame; For the first The basic framework can overcome the conservative prediction of external static pressure; For the first Conservative prediction of operating noise for the basic framework; For the first Conservative predictions of the filtration efficiency of the basic framework filters; For the first A conservative prediction of the upper limit of energy consumption per unit air supply volume in a basic framework.

3. The collaborative management method for order delivery of building ventilation and purification products according to claim 2, characterized in that, In step S1, the method for determining the initial frame is as follows: Obtain the performance requirement vector of the building ventilation system Specifically: ， in, To determine the total air supply volume requirement for the building, the functional zones of the building are identified using the building's BIM system, and the air change rate for each functional zone is determined. The air supply volume requirement for each functional zone is then determined by multiplying the air change rate by the volume. Finally, the air supply volume requirements of all functional zones are summed to obtain the total air supply volume requirement for the building. ; The total air supply volume requirement is At that time, the external static pressure that the building ventilation system needs to overcome; The upper limit for the operating noise of the building ventilation system; To meet the filter efficiency requirements; The upper limit of energy consumption per unit air volume; Calculate the difference between the predicted value of each base frame conservative performance vector and the building ventilation system performance requirement vector for each corresponding component: ， in, For the first The difference vector between the basic framework and the performance requirement vector of the building ventilation system; , , , , ; For the Each basic framework will be removed if any of the following conditions are met: ,and ; ,and ; ,and ; ,and ; ,and ; in , , , and These are the tolerance thresholds for rated air volume, external static pressure that can be overcome, operating noise, filter efficiency, and energy consumption per unit air volume, respectively. For the remaining basic frameworks after elimination through the tolerance threshold, calculate the overall matching score: ， in, For the first The overall matching score of the basic framework; , , , and They are respectively , , , and The corresponding weighted score; It is the absolute value; Based on overall matching score The basic frameworks are sorted from low to high, and the top few basic frameworks with the lowest scores are selected as the initial frameworks.

4. The collaborative management method for order delivery of building ventilation and purification products according to claim 3, characterized in that, In step S2, the method for determining the feasible product is as follows: Each initial frame is combined with all available models of each component in its replaceable component group by Cartesian product to generate several candidate products. For each candidate product, the component models of all its constituent parts are input into the product performance prediction model to obtain the comprehensive performance vector of each candidate product: ， in, For the first The comprehensive performance vector of each candidate product; For the first Rated air volume of each candidate product; For the first The external static pressure that each candidate product can overcome; For the first Operating noise of each candidate product; For the first Filter efficiency of each candidate product; For the first Upper limit of energy consumption per unit air volume for each candidate product; Calculate the performance margin vector for each candidate product. : ， in For the first The performance margin vector of each candidate product; like If any component in the set is less than zero, it is removed to obtain the first candidate set of feasible products; For each product in the first candidate set of feasible products, based on the Pareto non-dominated sorting method, a dominated counter is set for each product. All products are traversed, and all products with a dominated count of zero are added to the Pareto front. All products on the Pareto front are used as the second candidate set of feasible products. For any two products A and B, if the performance margin vector of A is greater than or equal to the performance margin vector of B in all dimensions, it indicates that A dominates B, that is, B is dominated by A. For each product in the second candidate set of feasible products, calculate the comprehensive matching score: ， in, For the second candidate set The overall matching score of each product; Will The products in the second candidate set are sorted from highest to lowest score, and the top few products with the highest scores are selected as feasible products.

5. The collaborative management method for order delivery of building ventilation and purification products according to claim 4, characterized in that, For each viable product: Calculate the required production time, obtain the standard production time of each component of the feasible product and the processing dependencies between components from the enterprise ERP system, obtain the queue of pending orders in front of the corresponding production line of each component from the enterprise production scheduling module, determine the queuing waiting time of each component in front of the corresponding production line, the queuing waiting time is equal to the sum of the cumulative processing time of all pending orders in front of that component; add the queuing waiting time of each component to its own standard production time to obtain the completion time of that component. Based on the processing dependencies between components, the maximum value is taken in parallel for components without dependencies, and the values ​​are accumulated serially for components with dependencies, to obtain the production time required for the feasible product. Calculate the funds required for its production, obtain the standard production funds required for each component of the feasible product from the enterprise's ERP system, and sum them up to obtain the funds required for production. Calculate the installation time, obtain the standard working hours of each installation process of the feasible product from the installation history database of the enterprise ERP system, construct a directed acyclic graph of installation processes based on the sequential dependencies between processes, determine the longest path from the start process to the end process as the critical path, and use the sum of the working hours of each process on the critical path as the installation time. Calculate the funds required for its installation. Obtain the number of workers and skill levels, equipment types and usage time required for each installation process of the feasible product from the enterprise ERP system. Calculate the labor cost according to the differentiated labor rate for different skill levels. Calculate the equipment amortization cost according to the equipment type and usage time. Add them together to obtain the funds required for installation.

6. The collaborative management method for order delivery of building ventilation and purification products according to claim 5, characterized in that, In step S3, the method for determining the optimal transportation method for each feasible product on each alternative route is as follows: Obtain spatial constraint point information for each candidate path, and for each spatial constraint point, construct a local voxel tensor based on the spatial constraint parameters. ,in For the first The first of the alternative paths The local voxel tensor of a spatially constrained point, wherein the local voxel tensor is a three-dimensional binary tensor, and a voxel value of 1 indicates that passage is allowed, and a voxel value of 0 indicates that an obstacle exists; The overall dimensions of each feasible product are converted into an overall voxel tensor. , For the first The overall body element tensor of a feasible product; According to the product structure tree, it is decomposed into an L-level set of sub-component voxel tensors: ， For the first The first feasible product The set of voxel tensors of sub-components, L is The maximum number of disassembly stages for a feasible product; Based on voxel Boolean convolution, the intersection volume ratio of the entire voxel tensor of the feasible product with the local voxel tensor of each alternative path is calculated. If all ratios are equal to 1, the optimal transportation method is determined to be original container transportation; if any ratio is less than 1, then: Calculate the intersection volume ratio of each sub-component voxel tensor in the voxel tensor set of each sub-component of the feasible product with each local voxel tensor on the alternative path level. If there is a level where all the calculated intersection volume ratios are equal to 1, then the transportation method to disassemble the sub-component at that level is determined to be the optimal transportation method. If, even at the maximum number of disassembly levels, there is still a case where the intersection volume ratio is less than 1, then it is determined that the alternative route cannot be used for transportation.

7. The collaborative management method for order delivery of building ventilation and purification products according to claim 6, characterized in that, In step S3, the transit time of each feasible product on each alternative route based on the optimal transportation mode includes logistics time and station operation time; the transportation funds include logistics transportation funds and station operation funds for the corresponding transportation mode.

8. The collaborative management method for order delivery of building ventilation and purification products according to claim 7, characterized in that, Step S4 includes the following: Enumerate all possible delivery plans. Each delivery plan consists of a feasible product, a transportation route, and the optimal transportation method for the current feasible product and transportation route. Specifically: ， in, For the first One delivery plan, For the first One feasible product, For the first Transportation routes, For the first The first feasible product in the first The optimal mode of transport on the transport route; for : Its total time cost ,in For the first Production time for each delivery plan For the first The transit time for each delivery solution. For the first Installation time for each delivery solution; Its total cost of capital ,in For the first The funds required for the production of each delivery plan For the first The funds required for transportation of each delivery plan. For the first Funding required for the installation of each delivery plan; Risk adjustments are made to the total time cost and total capital cost, specifically as follows: ， ， in, and These are the risk-adjusted values ​​for total time cost and total capital cost, respectively. and These are the time risk preference coefficient and the cost risk preference coefficient, respectively. and These are the standard deviations of total time cost and total cost of capital, respectively. ， in , and These are the standard deviations of time for the production stage, the transportation stage, and the installation stage, respectively; and , , ; ， in , and These are the standard deviations of costs in the production, transportation, and installation stages, respectively; and , , ; , and These are the preset coefficients of variation for the production, transportation, and installation stages, respectively. , , ; Based on the customer's expected delivery time span Select the one that meets the requirements from all delivery options. The feasible solutions constitute a set of feasible solutions; If the set of feasible solutions is not empty, then choose from it. The smallest possible solution is selected as the optimal solution; if the set of feasible solutions is empty, then the solution with the smallest possible solution is selected. The minimum solution is selected as the optimal solution, and a delayed delivery notification is sent to the customer. The calculation is then performed. and The difference is used as the delay time span.

Images