Automatic optimization system for construction schedules and quotation strategies based on contract success probability

Abstract

The present invention provides an order management system and method that predicts potential project risks and contract success probability by complexly processing customer data and external environment data, and automatically generates a design, construction schedule, and proposal strategy selected according to evaluation criteria based on the prediction results.

Description

Technology Field

[0001] The present invention relates to an artificial intelligence-based construction order management system and method, and more specifically, to a technology that comprehensively analyzes customer data and external environmental data to predict potential risks and success possibilities of a project, and automatically generates optimal designs, construction schedules, and proposal strategies based on the prediction results to maximize the order success rate and profitability of an interior or architectural construction company. Background Technology

[0002] With competition intensifying in the construction and interior design markets recently, various efforts are being made to increase the success rate of securing construction contracts. Traditional contract management methods relied primarily on the personal experience or intuition of assigned sales representatives to propose designs and quotations to clients. This approach resulted in inconsistent proposal quality and had clear limitations in identifying hidden client requirements or potential risks inherent in the project in advance.

[0003] Furthermore, some existing computerized systems were limited to simply statistically analyzing past project data or database-retrieving design portfolios. Consequently, they failed to effectively incorporate unstructured and unpredictable risk factors into bidding strategies, such as rapid external environmental changes like typhoons or logistical disruptions, complex project delays arising from the simultaneous deployment of multiple work teams, and shifts in the client's psychological state. As a result, low success rates, excessive proposal preparation costs, and frequent project losses due to unexpected risks occurred. The problem to be solved

[0004] The present invention was devised to solve the problems of the prior art as described above, and its main purpose is to solve the following problems.

[0005] The first task is to provide technical means to objectively predict the likelihood of contract success and the multidimensional risks inherent in a project based on customer data and external environmental data, while excluding the subjective judgment of the person in charge.

[0006] The second task is to go beyond the assessment of a single risk and proactively detect the destructive synergistic effects that occur when multiple independent risk factors arise simultaneously, and to reflect this in the construction schedule plan to minimize the possibility of project schedule collapse.

[0007] The third task is to automatically calculate reasonable and dynamic estimates corresponding to the risk level of each project by linking the physical risks of project execution with the customer's financial reliability.

[0008] The fourth task is to increase the final contract success rate through effective communication by inferring changes in market trends and customers' psychological states (decision fatigue) in real time, and automatically generating and delivering customized proposals optimized for customers' information processing capabilities.

[0009] The problems of the present invention are not limited to those mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below. means of solving the problem

[0011] According to an embodiment of the present invention for solving the above problem

[0012] A storage unit that stores data;

[0013] A communication unit that transmits and receives data to and from an external system; and

[0014] In an automatic optimization system for construction schedules and estimation strategies based on contract success probability, comprising a processing unit that controls the overall operation of the system,

[0015] The above processing unit is,

[0016] A customer information collection module that collects customer information including text or image data representing the customer's preferred style and physical information of the construction target space through the above communication unit;

[0017] A project order prediction module that performs a project order prediction including multiple risk indicators inherent in project execution and contract possibilities based on the collected customer information and past project data previously stored in the storage unit; and

[0018] A design construction strategy recommendation module that evaluates the stability of the proposed construction schedule and recommends an optimal construction strategy based on the above order forecast results;

[0019] Characterized by executing,

[0020] The above storage unit is,

[0021] 'Raw data lake' that stores raw data collected from external sources without modification;

[0022] A 'data warehouse' that stores data processed into a structured and analyzable form; and

[0023] 'Model and rule registry' that stores version-specific information on trained machine learning models and various judgment rules;

[0024] Characterized by being logically separated including,

[0025] The above processing unit is,

[0026] It is characterized by controlling the entire process to operate organically in a logical order, including a workflow management engine that defines the execution order and data dependencies of the customer information collection module, the order forecasting module, and the design and construction strategy recommendation module.

[0027] The above order forecasting module is,

[0028] It is characterized by calculating 'basic attractiveness' which evaluates the intrinsic value of a proposal and 'total risk' which evaluates the extrinsic and intrinsic instability of a project, and

[0029] The above total risk is,

[0030] 'Regional Impact Vulnerability' calculated based on disaster and social disruption data received in real time through an external public data application programming interface (API); and

[0031] 'Logistics bottleneck probability' calculated based on historical logistics data and real-time traffic information;

[0032] It is characterized by being calculated based on a risk indicator that includes at least one of the following,

[0033] The above order forecasting module is,

[0034] It is characterized by executing a 'risk veto' logic that forcibly adjusts the probability of final contract success to a 'very low' grade regardless of the evaluation result of the basic attractiveness calculated above, when the total risk calculated above corresponds to a 'risk' grade set based on the minimum value of risk indicators commonly observed in past project failure cases.

[0035] The above design and construction strategy recommendation module is,

[0036] It is characterized by evaluating robustness indicating the stability of the proposed construction schedule,

[0037] The above robustness is,

[0038] 'Work team synchronization vulnerability' representing risks arising from the complexity of process management within the project; and

[0039] 'Material Supply Weather Coupling' representing the risk arising from the combination of external supply chain instability and extreme weather events;

[0040] It is characterized by evaluating based on,

[0041] The above task team synchronization vulnerability is,

[0042] It is characterized by a structure of a merge node derived from project process management data in which multiple preceding tasks merge into a single subsequent task, and a calculation based on the average delay rate based on the historical data of each preceding task.

[0043] The above material supply weather combination diagram is,

[0044] Combine the main transportation route data of key materials with the probability of adverse weather forecasts on the corresponding route received via an external API,

[0045] The 'weather sensitivity' for each transport method is characterized by being determined by referring to a statistical database of past transport delay rates by transport method and weather condition stored in a storage unit, and

[0046] The above design and construction strategy recommendation module is,

[0047] If the condition is satisfied that the above work team synchronization vulnerability exceeds a first threshold and 'and (logical AND condition)' the above material supply weather coupling degree exceeds a second threshold, the robustness may be finally evaluated by additionally applying a 'synergy risk penalty' set to a value greater than the sum of the results of the individual risk evaluations, separate from the evaluation of individual risks.

[0048] In one embodiment,

[0049] In a method for predicting orders performed by an automated optimization system for construction schedules and quotation strategies based on contract success probability,

[0050] (a) A step of collecting customer information including text or image data representing the customer's preferred style and physical information of the construction target space through a communication unit;

[0051] (b) A step of calculating 'basic attractiveness' which evaluates the intrinsic value of the proposal and 'total risk' which evaluates the extrinsic and intrinsic instability of the project, based on the collected customer information and past project data stored in the storage unit;

[0052] (c) a step of forcibly adjusting the probability of final contract completion to a ‘very low’ grade regardless of the evaluation result of the basic attractiveness calculated above, if the total risk calculated above corresponds to a pre-set ‘risk’ grade; and

[0053] (d) a step of determining the final contract success probability grade adjusted based on the above forced adjustment result or the above basic attractiveness;

[0054] It may include.

[0055] In one embodiment,

[0056] In the construction strategy recommendation method performed by the automatic optimization system for construction schedules and quotation strategies based on contract success probability,

[0057] (a) A step of calculating 'work team synchronization vulnerability' representing the risk arising from the complexity of internal project management for the proposed construction schedule;

[0058] (b) A step of calculating a ‘material supply weather coupling’ that represents the risk arising from the combination of external supply chain instability and extreme weather events with respect to the above construction schedule;

[0059] A step of determining whether the condition that the above work team synchronization vulnerability exceeds a first threshold and 'and (logical AND condition)' the above material supply weather coupling degree exceeds a second threshold is satisfied; and

[0060] (d) A step of determining the final robustness grade of the construction schedule by additionally applying a 'synergy risk penalty' separately from the evaluation of individual risks, if the above conditions are satisfied as a result of the judgment in step (c) above;

[0061] It may include.

[0062] In one embodiment,

[0063] In an integrated order management method performed by an automated optimization system for construction schedules and quotation strategies based on contract success probability,

[0064] (a) a step of collecting customer information through a communication unit, and a processing unit extracting key keywords from the customer information and converting them into structured data;

[0065] (b) A step in which the processing unit determines the project's 'contract success probability grade' based on the structured data and external risk data;

[0066] (c) A step in which the processing unit derives a plurality of construction strategy candidates based on the contract success probability grade, and determines a 'construction schedule robustness grade' by evaluating the possibility of simultaneous occurrence of internal process risk and external supply chain risk for each candidate; and

[0067] (d) A step in which the processing unit finalizes the proposal strategy selected according to evaluation criteria by comprehensively considering the contract success probability grade and the construction schedule robustness grade;

[0068] It may include. Effects of the invention

[0070] According to the present invention for solving the above-mentioned problems, the following effects can be expected.

[0071] Potential losses can be prevented in advance through the preemptive filtering of projects with critical risks. This invention prevents massive financial and time losses resulting from securing high-risk projects by utilizing a 'risk veto' logic that requires a reconsideration of the contract itself if uncontrollable external environmental risks or inherent complexity exceed a specific threshold, even if the proposal is highly attractive.

[0072] Through data-driven, objective decision-making, consistency within the organization can be enhanced regarding the company-wide order success rate. Moving away from the existing method that relied on the individual capabilities of a select few top performers, this invention supports the execution of the entire proposal process based on standardized and quantified data, thereby stably improving the order management capabilities of the entire organization.

[0073] It can enhance the ability to predict and respond to complex risk scenarios. By quantitatively evaluating the synergistic effects of complex risks, such as the simultaneous occurrence of internal process problems and external supply chain problems, and inducing the establishment of sufficient safeguards in the schedule plan, this invention enables the stable execution of a project even in unpredictable situations.

[0074] Customer satisfaction and final contract rates can be increased through customized adaptive communication. By inferring even psychological states such as customer decision fatigue and dynamically adjusting the quantity and method of information provision, this invention minimizes customer decision-making stress and effectively conveys the key value of the proposal, thereby providing a positive customer experience.

[0075] The effects according to the present invention are not limited to those exemplified above, and a wider variety of effects are included within the present invention. Brief explanation of the drawing

[0077] Figure 1 illustrates an overall relationship diagram according to the present invention. Figure 2 illustrates a flowchart between all components according to the present invention. Figure 3 illustrates a process 1 flowchart according to the present invention. Figure 4 illustrates a process 2 flowchart according to the present invention. Figure 5 illustrates a process 3 flowchart according to the present invention. Figure 6 illustrates a process 4 flowchart according to the present invention. FIG. 7 schematically illustrates an integrated dashboard screen according to the present invention. FIG. 8 schematically illustrates a new project registration screen according to the present invention. FIG. 9 schematically illustrates a project comprehensive analysis dashboard according to the present invention. FIG. 10 schematically illustrates an order prediction analysis screen according to the present invention. FIG. 11 schematically illustrates a construction schedule robustness analysis screen according to the present invention. FIG. 12 schematically illustrates a dynamic estimate calculation screen according to the present invention. FIG. 13 schematically illustrates a proposed complexity determination screen according to the present invention. FIG. 14 schematically illustrates an automatic generation proposal preview screen according to the present invention. FIG. 15 schematically illustrates an alternative option simulation screen according to the present invention. FIG. 16 schematically illustrates a customer interaction monitoring screen according to the present invention. FIG. 17 schematically illustrates an adaptive proposal notification screen according to the present invention. FIG. 18 schematically illustrates a project result feedback input screen according to the present invention. FIG. 19 schematically illustrates a prediction model management (MLOps) screen according to the present invention. FIG. 20 schematically illustrates a risk rule management screen according to the present invention. Specific details for implementing the invention

[0078] Hereinafter, various embodiments are described in more detail with reference to the attached drawings. The embodiments described in this specification may be modified in various ways. Specific embodiments may be depicted in the drawings and described in detail in the detailed description. However, specific embodiments disclosed in the attached drawings are intended only to facilitate understanding of various embodiments. Accordingly, the technical concept is not limited by specific embodiments disclosed in the attached drawings, and it should be understood that it includes all equivalents or substitutions that fall within the spirit and scope of the invention.

[0079] Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but these components are not limited by the aforementioned terms. The aforementioned terms are used solely for the purpose of distinguishing one component from another.

[0080] Artificial intelligence-related functions according to the present disclosure may be executed by a processor and memory of a processing unit. The processor may be implemented as a general-purpose computing processor or a dedicated processor for artificial intelligence computing.

[0081] Pre-configured operation rules or artificial intelligence models can be generated through learning. Here, being generated through learning means that the learning algorithm is configured to produce an output corresponding to the objective by updating the artificial intelligence model using multiple training data.

[0082] An artificial intelligence model can be composed of multiple neural network layers. Each of the multiple neural network layers has multiple nodes and weight values, and performs neural network operations through calculations between the results of previous layers and the multiple weights. The multiple weights possessed by the multiple neural network layers can be optimized based on the learning results of the artificial intelligence model. For example, multiple weights can be updated so that the loss value or cost value obtained by the artificial intelligence model during the learning process is reduced or minimized. Additionally, to minimize the loss value or cost value, multiple weights can be updated in a direction that minimizes the gradient associated with the loss value or cost value. Artificial neural networks may include deep neural networks (DNNs), such as Convolutional Neural Networks (CNNs), Deep Neural Networks (DNNs), Recurrent Neural Networks (RNNs), Restricted Boltzmann Machines (RBMs), Deep Belief Networks (DBNs), Bidirectional Recurrent Deep Neural Networks (BRDNNs), or Deep Q-Networks, but are not limited to the examples mentioned above.

[0083] A network is a data transmission path between a system and an external device, and can be a closed network or an Internet-based network.

[0084] The terminal can be implemented as various forms of electronic devices, such as smartphones, tablet computers, desktop computers, and smart televisions.

[0085] In this specification, terms such as “comprising” or “having” are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof. When a component is described as being “connected” or “connected” to another component, it should be understood that it may be directly connected to or connected to that other component, or that there may be other components in between. On the other hand, when a component is described as being “directly connected” or “directly connected” to another component, it should be understood that there are no other components in between.

[0086] Meanwhile, in this specification, a software component may be referred to as a 'module' that performs at least one function or operation, and a hardware component may be referred to as a 'unit' that performs at least one function or operation. The module may be implemented by software, hardware, or a combination of hardware and software.

[0087] In addition, the operation embodiments and configurations described in a general manner without being explained in detail below follow the prior art and are omitted in order to focus on describing the purpose of the present invention and the resulting effects.

[0088] Furthermore, in describing the present invention, if it is determined that a detailed description of related known functions or configurations may unnecessarily obscure the essence of the invention, such detailed description is abbreviated or omitted.

[0089] An artificial intelligence-based construction order management system for implementing the present invention may be configured to include a processing unit (100) responsible for central control, a storage unit (200) for permanent storage of data, and a communication unit (300) for transmitting and receiving data with an external device.

[0090] The processing unit (100) is defined as a hardware component implemented with one or more processors responsible for central control of the system according to the present invention. The processing unit (100) comprehensively controls the entire process, including a workflow management module that defines the execution order, data dependencies, and exception handling of each software module. The workflow management module calls the order prediction module (130) after customer information collection and preprocessing are completed, and controls the order prediction module (130) to transmit the output value of the order prediction module (130) as an input value to the design construction strategy recommendation module (140). The processing unit (100) exchanges data with the storage unit (200) and the communication unit (300) via an internal bus.

[0091] The storage unit (200) is defined as one or more physical or logical storage media that permanently store all data and software required for the present invention. The storage unit (200) includes a plurality of logically separated data storage units to systematically manage the lifecycle of data. Specifically, it is configured to include (a) a 'raw data lake' that stores raw data collected from the outside without modification; (b) a 'data warehouse' that stores structured and refined data; and (c) a 'model and rule registry' that stores version information of learned machine learning models and various judgment rules. Through this, it performs the function of ensuring data consistency, version control, and reproducibility. The storage unit (200) responds to data read / write requests from the processing unit (100) and maintains data integrity through a database management system. External data received through the communication unit (300) transmits a command to be preferentially written to the raw data lake under the control of the processing unit (100).

[0092] The communication unit (300) is defined as a set of hardware interfaces responsible for data communication between the inside and outside of the system. The communication unit (300) may include an application programming interface (API) gateway for collecting external public data and social media data, and a secure database linkage unit for data linkage with an internal customer relationship management system and project management software.

[0093] The above user terminal unit (400) is a user interface device for checking the processing results of the system and providing input data, and can be implemented as a smartphone or tablet computer, etc. The above user terminal unit (400) communicates with a server through a communication unit (300) and can generate customer behavior data such as the number of times a proposal is viewed, the time to reply, and whether additional information is requested.

[0094] The customer information collection module (110) is defined as a software module that performs the function of collecting data from the outside and initial identification, and the data preprocessing module (120) is defined as a software module that performs the function of processing the collected raw data into an analyzable form. The customer information collection module (110) identifies the type of input data and tags initial additional information (metadata) and stores it in a raw data lake. Subsequently, the data preprocessing module (120) retrieves customer data from the raw data lake, extracts keywords from text and extracts features from images, and then stores the structured data set in a data warehouse.

[0095] The order prediction module (130), the design and construction strategy recommendation module (140), and the automatic estimation module (160) are defined as analysis and generation modules that work in conjunction with each other to derive a proposal package selected according to evaluation criteria. The design and construction strategy recommendation module (140) can limit the search range of the recommended strategy by using the 'contract success probability grade' received from the order prediction module (130) as an input value.

[0096] The result analysis module (170) is defined as a software module responsible for the model operation function of feeding actual project results back to the system and updating the prediction model through this. The result analysis module (170) does not immediately deploy the retrained new model to the operating environment, but instead performs a performance comparison evaluation with the existing operating model using a previously stored verification data set. Control can be exercised to approve the deployment of the new model only when the results of the comparison evaluation meet pre-set performance criteria.

[0097] In a scenario where a proposal process for a new customer begins, when a representative inputs the details of the consultation with the customer and on-site photos into the system, a communication unit (300) receives the data. Under the control of a processing unit (100), a customer information collection module (110) identifies the data as 'text' and 'image', adds additional information, and stores it in a raw data lake within a storage unit (200). A workflow management module within the processing unit (100) recognizes that data storage is complete and executes a data preprocessing module (120). The data preprocessing module (120) retrieves customer data from the raw data lake, extracts keywords from the text and features from the images, and then stores the structured data set in a data warehouse.

[0098] Once preprocessing is complete, the workflow management module executes the order prediction module (130) to evaluate the initial order probability. Subsequently, the design construction strategy recommendation module (140) generates multiple design candidate groups based on the initial probability and the preprocessed data. The module calls the automatic estimation module (160) in real-time for each candidate group to calculate an estimated estimate, and then transmits the design and estimate package back to the order prediction module (130) to re-evaluate the final contract probability of each candidate group. Through the above iterative process, a command is sent to confirm the design package with the highest probability as the final proposal.

[0099] After several months, when the project is successfully contracted, the person in charge inputs the results into the system. A communication unit (300) receives these 'success' results, and a processing unit (100) records them in a 'data warehouse'. A result analysis module (170) that runs periodically retrains the entire data, including these new success cases, to create a new prediction model. After comparing and evaluating the performance of the new model and the existing model, if performance improvement is confirmed, the module automatically sends a command to replace the operational model stored in the 'model and rule registry' with the new model to improve the overall prediction accuracy of the system.

[0100] The contract success probability rating process of the present invention transmits a command to determine the final probability based on two independent criteria: 'Basic Attractiveness,' which evaluates the intrinsic value of a proposal, and 'Total Risk,' which evaluates the extrinsic and intrinsic instability of project execution. As a judgment criterion related to Basic Attractiveness, the 'Cultural Fit Index' is defined as a judgment criterion for quantifying the potential for subjective and emotional satisfaction of customers. The reason for evaluating the Cultural Fit Index in this invention is that its value is calculated based on the empirical fact that, according to the analysis of past successful project data, proposals that satisfied customers' emotional needs had a statistically significantly higher final contract rate, even if their technical completeness was similar. Additionally, the 'Technical Proposal Score' is a judgment criterion for quantifying the objective and technical completeness of a proposal, and transmits a command to evaluate the competitiveness of the proposal itself through objective indicators such as budget appropriateness, reliability of the proposed technology, and similarity of the portfolio.

[0101] As a criterion for judging overall risk, 'Regional Shock Vulnerability' is defined as a standard for quantifying geopolitical and natural disaster risks in the project execution region, and evaluates in advance the potential impact that external environmental factors beyond the contractor's control will have on the project. 'Logistics Bottleneck Probability' is a criterion for quantifying the risk of delays that may occur during the procurement process of key materials and equipment, and evaluates the stability of the supply chain based on historical data and current conditions regarding specific ports, roads, and customs procedures. 'Permit Procedure Delay' is a criterion for quantifying the risk caused by delays in administrative processing by government agencies, predicting the likelihood of such delays based on the average and deviation of past similar permit processing times. 'Cross-Feature Interaction Entropy' is a criterion for quantifying the inherent complexity and uncertainty of the project itself; the reason for borrowing the concept of entropy from information theory is to technically model the principle that overall project uncertainty increases as the interdependence and unpredictability among the various variables constituting the project decrease. This transmits a command to be utilized as an indicator for evaluating the possibility of the existence of 'unknown risks'.

[0102] Considering that the attractiveness of the proposal and project risk affect the likelihood of final contract success in different ways, the above processing procedure may be composed of three stages: 'first evaluation (attractiveness assessment), second evaluation (risk classification), and third evaluation (risk-based adjustment).' For example, if the above overall risk grade is determined to be a pre-set risk grade, the above processing unit (100) controls the final contract success probability grade to be adjusted to the risk grade regardless of the above basic attractiveness grade.

[0103] The processing unit (100) extracts key design keywords from customer preference style text and reference images collected through the customer information collection module (110) in order to obtain a cultural suitability index. In one embodiment, the processing unit (100) may perform morphological analysis on the text, generate noun phrases as candidates, construct a graph of the co-occurrence relationships between the candidates, and then apply a TextRank algorithm to sort them in order of importance. Additionally, the processing unit (100) may extract color distribution, contours, and object features from the reference images, and generate image-based keywords by matching the features with a keyword dictionary. The processing unit (100) selects the top 10 keywords by integrating the text-based keywords and the image-based keywords. The processing unit (100) calculates the Jaccard similarity between the selected set of keywords and the set of keywords of successful case projects already stored in the storage unit (200). For example, if the number of intersection elements is 8 and the number of union elements is 27, the above culture fitness index can be calculated as the value obtained by dividing 8 by 27 (approximately 0.30).

[0104] The processing unit (100) can receive data on disaster text messages from the Ministry of the Interior and Safety, typhoon forecasts from the Korea Meteorological Administration, and assembly notification status from local police stations, which are received in real-time via a public data API, in order to obtain regional impact vulnerability. The processing unit (100) converts the disaster warning level issued within the last 30 days and the intensity of social disruption factors scheduled within the next 14 days into values ​​between 0 and 1 according to a pre-set risk mapping table, targeting the area scheduled for construction. If different types of risks exist simultaneously, the processing unit (100) can calculate the largest value among the converted values ​​as the regional impact vulnerability. For example, if the value corresponding to a preliminary typhoon warning is 0.5 and the value corresponding to an assembly notification is 0.3, the regional impact vulnerability can be calculated as 0.5.

[0105] The processing unit (100) calculates mutual information for all possible pairs of variables from proposal data of past awarded projects in order to obtain cross-feature interaction entropy. Based on the distribution of all calculated mutual information values, it calculates information entropy to normalize the cross-feature interaction entropy to a value between 0 and 1, and determines that the higher this value, the lower the correlation between variables and the higher the unpredictability.

[0106] The processing unit (100) can determine follow-up actions based on the final derived contract success probability rating. For example, if the contract success probability rating is 'high', the project is designated as a priority management target and a notification is generated. If the contract success probability rating is 'medium', guidance is provided to request supplementary materials regarding identified risk factors from the customer. If the contract success probability rating is 'low', an internal warning signal is generated. If the contract success probability rating is 'risky', a report is generated to stop the project proposal or to set the addition of specific clauses to the contract as a condition.

[0107] In a scenario for a proposal for a luxury apartment interior project located in Gangnam-gu, Seoul, the processing unit (100) determines the 'basic attractiveness grade' as 'high' based on customer preference and proposal competitiveness. Subsequently, since the regional shock vulnerability, logistics bottleneck probability, licensing procedure delay, and cross-feature interaction entropy of the Gangnam-gu area are all stable, the 'total risk grade' is determined as 'safe' according to the stored risk assessment rule ("classified as 'safe' grade if all risk indicators are less than 0.4"). Finally, since the risk grade is 'safe', the basic attractiveness grade is used as is, and a command is transmitted to confirm the final 'contract success probability grade' as 'high'.

[0108] If data for a specific risk indicator cannot be received from an external API, the processing unit (100) can ensure the stability of the prediction by using the average value of the region over the past three months as a substitute value or the maximum value over the past three months as a substitute value. In addition, the risk assessment rule set and threshold can be automatically adjusted through a decision tree model retrained using actual data from projects completed every quarter.

[0110]

[0111]

[0112] The example table illustrated in Table 2 is calculated according to the 'Overall Risk Class Determination Rule'. In one example, the overall risk class is classified as 'Risk' if the regional shock vulnerability exceeds 0.75, the logistics bottleneck probability exceeds 0.8, or the cross-feature interaction entropy exceeds 0.85. Additionally, if each risk indicator is 0.5 or higher and is below the risk threshold value, the overall risk class is classified as 'Caution'. Examples 3, 5, and 10 illustrate cases where, despite the base attractiveness rating being 'High' or 'Medium', a specific risk indicator exceeds a threshold value, causing the overall risk class to be classified as 'Risk', and as a result, the final contract success probability rating is adjusted to 'Risk'.

[0113] The judgment rules used in this invention are based on the technical necessity that proceeding with a contract solely on the excellence of a proposal can lead to significant financial losses, as the construction and interior design industries are characterized by high initial investment costs and long project durations, making them highly vulnerable to unexpected external risks. The risk veto logic of this invention is derived from these industrial characteristics, and the judgment criteria transmit commands that are objectively established based on statistical analysis of data from hundreds of past successful and failed projects and risk management theories.

[0114] This invention can reduce the potential for financial and time losses by identifying high-risk projects in advance. Furthermore, it can improve the reproducibility of decision-making by transitioning order acceptance decisions, which previously relied excessively on the experience of personnel, into a consistent, data-driven process. Moreover, it can strengthen an organization's risk management system by subdividing and evaluating project risks using multiple indicators.

[0115] The threshold values ​​for all grade classifications used in this specification are dynamically set based on the percentile of the corresponding indicator value in the entire project database. The threshold value for the 'Risk' grade is set not based on statistical criteria, but on the minimum value of risk indicators commonly observed in past project failure cases or absolute risk criteria established by experts, thereby transmitting a command to prevent the recurrence of catastrophic failures.

[0116] The construction schedule robustness rating process of the present invention may be configured to evaluate synergy risks separately from individual risks in order to identify and quantify potential vulnerabilities inherent in the project schedule.

[0117] As a judgment criterion, work team synchronization vulnerability can be defined as a criterion for quantifying the delay risk arising from the complexity of human resource management and process management within the project. Since delays in bottleneck sections where the results of multiple tasks are combined in project schedule management can cause a cascading effect on the entire subsequent process, the processing unit (100) can be controlled to identify the bottleneck section and calculate the possibility of delay in the bottleneck section.

[0118] The logical processing procedure of the present invention takes into account that a situation in which two or more different types of risks exist independently and a situation in which said risks exist simultaneously may have different levels of risk. For example, the processing unit (100) may control the activation of a synergy risk penalty when a first risk indicator exceeds a first threshold and simultaneously a second risk indicator exceeds a second threshold (logical conjunction condition).

[0119] The processing unit (100) can calculate the work team synchronization vulnerability using a program evaluation and review technique (PERT) chart generated from project management data to obtain judgment criteria data. For example, the number of merge nodes in which multiple preceding tasks join as one subsequent task in the PERT chart and the past delay rate of each preceding task can be weighted and summed to obtain a value between 0 and 1.

[0120] Additionally, the processing unit (100) calculates the material supply weather coupling degree by combining the main transport route data of the main material and the probability of weather deterioration forecasts for the next 15 days on the corresponding route received via the Meteorological Agency API. At this time, the 'weather sensitivity' for each transport method is not a fixed constant, but is determined by referring to the 'database of past transport delay rates by transport method and weather conditions' stored in the storage unit (200). For example, if the average delay rate of maritime transport was 80% when a typhoon warning was issued in the past, the sensitivity of maritime transport to typhoons is set to 0.8. For example, if the main material, Italian tiles, are transported by sea (weather sensitivity of 0.8 based on the database) and the probability of a typhoon occurring on the route is forecasted to be 50%, a command is transmitted to calculate the material supply weather coupling degree as 0.8 × 0.5 = 0.4.

[0121] The processing unit (100) determines follow-up actions based on the 'construction schedule robustness grade' finally derived. If the grade is 'stable', the proposed schedule is confirmed as is. If the grade is 'average', 2 to 3 days of buffer time are automatically added for the identified vulnerable processes, and notifications are automatically sent to the relevant departments to look for alternative suppliers for high-risk materials. If the grade is 'vulnerable', fundamental schedule readjustments are required, such as redesigning the process sequence or changing concurrent work sections to sequential work, and if adjustment is impossible, a 'high risk of schedule collapse' warning is sent to the project manager.

[0122] In a scenario for a specific construction schedule proposal, the processing unit (100) sets the initial robustness score to 100 points. Subsequently, if the work team synchronization vulnerability index is 0.6, which exceeds the first threshold of 0.5, the processing unit (100) subtracts a first deduction from the robustness score. In one embodiment, the first deduction may be calculated by multiplying the basic deduction constant 50 by (the value obtained by subtracting the threshold from the current index), and if the current index is 0.6 and the threshold is 0.5, the first deduction may be calculated as 5 points.

[0123] In the above scenario, if the material supply weather coupling index is 0.45, which exceeds the second threshold of 0.4, the processing unit (100) calculates the second deduction value by multiplying the basic deduction constant 40 by (the value obtained by subtracting the threshold from the current index) and adjusts the robustness score. Additionally, if both the work team synchronization vulnerability and the material supply weather coupling exceed their respective thresholds (logical AND condition), the processing unit (100) may apply an additional deduction based on the synergy risk penalty.

[0124] To ensure the reliability of the above logic, for new work teams lacking sufficient data, the industry average delay rate or the average delay rate of other teams performing similar processes may be applied as an initial value, and the indicators can be updated by accumulating actual data as the project progresses. Additionally, the weighting of the synergy risk penalty can be periodically learned and adjusted using schedule delay data from past projects.

[0125]

[0126] The judgment rules used in this invention are technically grounded in the 'system dynamics' theory, which posits that process delays in construction projects are amplified to unpredictable levels when multiple causes act simultaneously, rather than by a single cause. Although individual minor issues may appear controllable, their simultaneous occurrence can undermine the stability of the entire system; therefore, evaluating them independently is essential for objective risk management.

[0127] By adopting these components, the present invention possesses the technical effect of going beyond merely identifying schedules that may be delayed to proactively detecting combinations of complex risks that could lead to a worst-case scenario known as 'schedule collapse.' Furthermore, it transmits commands that enable effective schedule management by clearly distinguishing the causes of risk from internal and external sources and presenting customized response strategies for each.

[0128] The threshold values ​​for the 'Stable', 'Moderate', and 'Vulnerable' grades used in this specification are dynamically set by applying percentiles based on the distribution of the final robustness scores of the entire completed project. For example, by classifying the top 30% as 'Stable' and the bottom 30% as 'Vulnerable', a command is transmitted to allow the evaluation criteria to be flexibly adjusted according to changes in the risk level of the entire project portfolio.

[0129] The dynamic estimate determination process of the present invention can be configured to calculate a final estimate by combining the physical risks of the project and the financial reliability of the customer.

[0130] As a criterion for judgment, the site noise interaction coefficient can be defined as a standard for quantifying the risk of work stoppages and cost increases that may arise due to environmental sensitivity around the construction site. The aforementioned site noise interaction coefficient can be evaluated by comprehensively considering not only the noise magnitude but also the presence of surrounding facilities, local government noise regulation ordinances, and complaint data from similar past sites. Additionally, the probability of a logistics bottleneck can be defined as a standard for quantifying the risk of delays that may occur during the transportation of key materials and equipment.

[0131] The logical processing procedure of the present invention is based on the technical foundation of a "two-way risk transfer" model that evaluates the "site risk" borne by the construction company and the "financial risk" arising from the customer in a mutually complementary manner. Unlike conventional technology, which unilaterally reflects only the construction company's risk in the price, the present invention applies the technical concept that high customer trust directly reduces the project's "financial volatility risk." Smooth cash flow ensures the ability to respond immediately without additional financing costs in the event of unexpected site risks, thereby having the effect of reducing the construction company's actual risk management costs. Therefore, converting a portion of these savings into a price discount for highly trustworthy customers is considered a reasonable pricing policy. To use an analogy with a ramen recipe, a command is transmitted to operate in a manner similar to how a regular customer who always pays the full price is discounted even if the store's operating costs rise, while a higher price is charged to a customer with a history of credit purchases to ensure the store's financial stability.

[0132] The processing unit (100) can calculate the field noise interaction coefficient to obtain judgment criteria data by multiplying the noise excess ratio, which is calculated by dividing the expected work noise level by the allowable standard of the competent local government ordinance, and an environmental sensitivity weight based on the number of noise-sensitive facilities within a radius of 500 meters. For example, if the expected noise is 80 decibels, the allowable standard is 70 decibels, and there are 2 sensitive facilities, the noise excess ratio is the value obtained by dividing 80 by 70 (approx. 1.14), and if the environmental sensitivity weight is 1.5, the field noise interaction coefficient can be calculated as approximately 1.71.

[0133] In particular, payment reliability speed is quantified by calculating the 'average payment delay days' by averaging the difference between the 'payment agreement date' specified in the contract for each progress payment and the 'actual payment date' when the actual payment was deposited, based on the customer's past transaction data. In a more preferred embodiment, the 'standard deviation of payment delay days' may be used as an additional judgment criterion along with the 'average payment delay days'. For example, even if the average delay days are the same, for customers whose standard deviation value exceeds a preset threshold, the grade is downgraded by one level to 'low' to additionally reflect the risk of payment unpredictability. For example, if the payment delay days in the past three transactions were 2 days, 0 days, and 4 days, respectively, a command is sent to calculate the average payment delay days as (2+0+4) / 3 = 2 days.

[0134] The processing unit (100) determines follow-up actions based on the final derived estimate. In the case of a high-risk project where the risk surcharge rate exceeds a specific standard, it automatically sends an order to the purchasing team to increase the advance payment ratio in the contract from the standard ratio and proceed with the pre-purchase of major materials. If the customer's payment reliability speed is determined to be 'low', it sends an order to automatically insert a clause into the contract mandating the use of a third-party escrow service to prevent the risk of uncollectible receivables in advance.

[0135] In a scenario for a specific project, the processing unit (100) calculates the base cost as 100 million won. Subsequently, as a result of evaluating on-site noise and logistics risks, the base risk surcharge rate is determined to be 8 percent by referring to a risk-surcharge rate mapping table. The mapping table has a two-dimensional matrix structure, and the rows may consist of grades of on-site noise interaction coefficients, and the columns may consist of grades of logistics bottleneck probabilities. Each cell may contain a surcharge rate calculated based on past data.

[0136] Next, as a result of analyzing the customer's past transaction history, it is determined that the average payment delay period is 2 days, satisfying the criteria for a 'high' grade. Accordingly, the processing unit (100) recalculates the final applicable premium rate by applying a 'trust discount rate' to the basic risk premium rate. At this time, the 'trust discount rate' is not a fixed value but is calculated through a continuous function such as (maximum discount rate 20%) × (1 - (average delay period / 'low' grade standard period 10 days)), and a discount rate of 16% is calculated. Accordingly, the final applicable premium rate is recalculated as 8% × (1 - 0.16) = 6.72%. Finally, a command is transmitted to calculate the final estimate as 106,720,000 won by applying the final applicable premium rate to the basic cost.

[0137] In the above scenario, if the customer's average payment delay date is 12 days, which corresponds to the 'low' grade standard, the processing unit (100) recalculates the final applicable premium rate by applying the 'credit risk premium rate' to the basic risk premium rate. The premium rate is also calculated as 6% according to the function (maximum premium rate 30%) × ((average delay date - 'low' grade standard date 10 days) / 10 days). As a result, the final applicable premium rate becomes 8% × (1 + 0.06) = 8.48%, and the final estimate is adjusted upward to 108.48 million won, and a command is transmitted to ensure that the customer's financial risk is directly reflected in the price.

[0138] To ensure the reliability of the above logic, new customers with no transaction history are initially assigned a payment reliability speed of 'Medium,' but the rating is adjusted using credit scores received from external credit rating agencies as a supplementary indicator. Furthermore, the risk-premium mapping table is automatically updated every quarter by statistically analyzing the actual risk costs incurred in completed projects, thereby continuously reflecting risk costs resulting from changes in market conditions.

[0139]

[0140] The judgment rules used in the present invention take into account that physical risks occurring during project execution and financial risks occurring during the payment collection process after completion can have a direct impact on the company's profit and loss. Therefore, the process of comprehensively evaluating the two types of risks and linking them to price can reinforce the rationality and objectivity of the estimate calculation.

[0141] By adopting these components, the present invention possesses the technical effect of utilizing an estimation system as a means of managing the comprehensive risks of a project, going beyond merely calculating costs. Through this, it is possible to prevent contracts with high-risk, low-reliability customers in advance or set prices corresponding to their risks, while conversely, providing price benefits to low-risk, high-reliability customers to establish long-term partnerships.

[0142] The threshold values ​​for the 'high', 'medium', and 'low' grades used in this specification are dynamically set by applying percentiles based on the distribution of the 'average payment delay days' of the entire customer database. For example, by classifying the bottom 30% as 'high' and the top 30% as 'low', a command is transmitted to allow the evaluation criteria to be flexibly adjusted according to changes in the payment propensity of the entire customer.

[0143] The adaptive proposal template selection process of the present invention can be configured to dynamically adjust the amount and depth of information delivered to the customer according to the customer's psychological state and market interest.

[0144] As a judgment criterion, 'social buzz volatility' is defined as a standard for quantifying social interest and topicality regarding a specific topic related to the proposal content. By measuring changes in the mention volume, search volume, and positive / negative sentiment index of related keywords on social media and online communities, it predicts how much information customers are likely to want regarding the topic. Additionally, 'decision fatigue score' is defined as a standard for quantifying the psychological burden and limitations in information processing ability felt by customers while proceeding with the proposal process, and a command is transmitted to infer the customer's 'cognitive overload' state based on specific interaction data.

[0145] The logical processing procedure of the present invention takes into account that there are limits to human information reception capacity, and that the quality of decision-making may deteriorate as the amount of information provided increases, especially in a state of high cognitive fatigue. Accordingly, the processing unit (100) can be controlled to switch to an information reduction mode that reduces the amount of information provided regardless of the customer interest level when the decision fatigue level exceeds a preset fatigue standard. In the information reduction mode, a summary proposal template can be selected to reduce the volume of the proposal and the number of detailed items.

[0146] The processing unit (100) may receive social buzz volatility through an external social media analysis API to obtain judgment criteria data. For example, the processing unit (100) receives the standard deviation of the daily mention volume change rate and the average value of the daily positive sentiment index for the main keywords of the proposal over the past 7 days. The processing unit (100) may calculate social buzz volatility by normalizing each of the two indicators and applying a pre-set weight (e.g., change rate weight 0.6, sentiment index weight 0.4).

[0147] In particular, the decision fatigue score is determined by receiving inputs such as the number of email and message exchanges with the customer after sending the proposal, the number of proposal views, and the average reply time, and classifying them as 'high' or 'low' through a pre-trained decision fatigue prediction model. The 'number of proposal views' is measured by embedding a unique web beacon in the proposal link sent to the customer and counting the signal transmitted to the server each time the customer views the link. Additionally, the decision fatigue prediction model learns the relationship patterns between 'communication frequency' and 'average reply time'. For instance, it enhances the accuracy of determination by reflecting time-series characteristics, classifying patterns with high frequency but short reply times as the 'high interest stage,' and patterns with high frequency but gradually increasing reply times as the 'fatigue accumulation stage.' For instance, if 35 emails were exchanged over the past 7 days and the customer's average reply time was measured at 30 hours, the model sends a command to determine the customer's decision fatigue score as 'high.'

[0148] The processing unit (100) changes the content and structure of the proposal according to the finally selected template. When a detailed analysis type template is selected, a proposal including a detailed specification comparison table by material and multiple visual materials can be generated and transmitted. When a brief summary type template is selected, summary information such as final design concept images and total estimated amount can be presented in a limited amount.

[0149] In a scenario for generating a proposal for a specific customer, the processing unit (100) first evaluates customer interest and fatigue. As input data, the social buzz volatility of the proposal content, 'eco-friendly interior,' is 0.8, satisfying the criteria for a 'high' grade. On the other hand, email exchanges with the customer occurred 25 times in the last 3 days and the average reply time was long at 40 hours. As a result of applying these input values ​​to a pre-trained logistic regression model, the output probability was calculated to be 0.92. Since this probability value exceeds 0.5, which is the threshold for a 'high' grade, the decision fatigue score is finally determined to be a 'high' grade.

[0150] Among the stored rules, for example, if the condition 'if the customer fatigue level is high, select a brief summary template regardless of the customer interest level' is satisfied, the processing unit (100) can select a brief summary template.

[0151] To ensure the reliability of the above logic, the decision fatigue prediction model is retrained regularly every month as new project data accumulates to continuously improve prediction accuracy. The training data for the model uses correct labels based on results where sales representatives tagged customer responses to the proposal process after past project completion as 'smooth,' 'difficult,' etc. Additionally, the system ensures flexibility by including exception handling rules that temporarily provide additional information equivalent to a 'detailed analysis template,' regardless of the current fatigue level, when a customer sends an explicit signal requesting information, such as by clicking a 'request details' button.

[0152]

[0153] The judgment rules used in this invention derive their technical necessity from research findings in behavioral economics and cognitive psychology, which indicate that consumers in modern society are exposed to a vast amount of information and easily experience psychological fatigue during complex decision-making processes. In particular, since customer "decision paralysis" or "information avoidance" are major causes of contract failure during the interior design contracting process—a high-involvement product—systematically managing these issues and reducing the customer's cognitive burden is an objective and rational approach.

[0154] By adopting these components, the present invention possesses the technical effect of implementing 'adaptive communication' tailored to the customer's psychological state, going beyond merely conveying information. Through this, it is possible to prevent customer churn caused by information overload and effectively convey the key value of the proposal to positively induce final decision-making. Furthermore, it provides the effect of increasing the efficiency of proposal work by reducing internal resources required for preparing unnecessary detailed materials.

[0155] The threshold values ​​for the 'high' and 'low' grades used in this specification are set based on the output probability values ​​of the decision fatigue prediction model. For example, if the output value of the logistic regression model is 0.5 or higher, it can be classified as 'high', and if it is less than 0.5, it can be classified as 'low'. This threshold value is not fixed, but rather a command is transmitted to maintain the model's discrimination performance in an optimal state by performing recipient operation characteristic curve analysis using past contract success and failure case data, finding the point where accuracy and recall are adjusted, and dynamically updating it.

[0156] The entire operation process of the artificial intelligence-based construction order management system according to one embodiment of the present invention is explained in detail through the case of a proposal for an old apartment interior remodeling project by new customers, the 'Kim couple'.

[0157] First, the sales representative inputs the initial consultation details with the "Mr. and Mrs. Kim" into the system. The communication unit (300) receives text information stating "prefers eco-friendly solid wood flooring and a minimalist style, but has a limited budget," the address of the apartment to be constructed, and on-site photo data. Under the control of the processing unit (100), the customer information collection module (110) identifies the data by type and transmits a command to store it in the "raw data lake" within the storage unit (200) along with a unique identifier called "Mr. and Mrs. Kim Project." Subsequently, the "workflow management module" within the processing unit (100) detects the input of new data and automatically executes the data preprocessing module (120). The data preprocessing module (120) retrieves the "Mr. and Mrs. Kim Project" data from the "raw data lake." The above module extracts key keywords such as 'eco-friendly', 'solid wood', 'minimalism', and 'budget sensitive' from text information through natural language processing, and analyzes on-site photos to convert the area, structure, and condition of existing finishing materials into structured data. The processed data is stored in a 'data warehouse' in a state ready for analysis.

[0158] Once data preprocessing is complete, the workflow management module transmits a command to execute four major processes in an organically linked manner. The order prediction module (130) is executed first, and due to the high degree of agreement between the keywords 'eco-friendly' and 'minimalism' and past success cases, the 'cultural suitability index' is evaluated as 'high'. However, because the consultation time is during typhoon season and the apartment is old, the licensing process is expected to be complex, so the 'regional shock vulnerability' and 'licensing process delay' indicators are evaluated as 'caution' grades. As a result, the initial 'contract success probability grade' is determined to be 'average'. Subsequently, the design construction strategy recommendation module (140) receives the probability of the above 'average' grade and performs a schedule evaluation. Since the main supplier of the preferred material, 'eco-friendly solid wood flooring,' is Southeast Asia and maritime transport is essential, the 'material supply weather coupling degree' combined with the typhoon forecast is calculated as 'high'. In addition, a bottleneck section requiring simultaneous flooring and electrical wiring work was identified, so the 'Work Team Synchronization Vulnerability' is also assessed as 'Caution'. Based on the overall assessment, the 'Construction Schedule Robustness Grade' is determined to be 'Vulnerable'.

[0159] Next, the automatic estimation module (160) is executed. The module first calculates the basic cost, and then, based on the results of the schedule robustness assessment for the derived 'vulnerable' grade, adds risk management costs to the basic cost to prepare for potential transportation delays and process interruptions due to typhoons, as well as for securing material inventory. Additionally, since the construction site address is near a school, the 'site noise interaction coefficient' is evaluated as 'high,' resulting in an additional surcharge. Since the 'Mr. and Mrs. Kim' are new customers, the 'payment reliability speed' is rated as 'medium.' As a result, a 'final applied surcharge rate' higher than the standard surcharge rate is calculated, determining a final estimate that is higher than expected. Finally, the proposal generation module (150) is executed. Since the 'social buzz volatility' regarding 'eco-friendly interior' is high, it is determined that providing detailed information is advantageous. However, by analyzing the pattern of repetitive questions and prolonged deliberation regarding the budget during the initial consultation process, the 'decision fatigue score' is evaluated as 'caution.' Based on the combined results of the two indicators, the system sends a command to select a 'standard proposal template' that is neither too detailed nor too simple.

[0160] The proposal generation module (150) generates a final proposal by synthesizing the analysis results of the four processes above. The proposal automatically includes (a) a standard design in a 'minimalism' style, (b) a somewhat high calculated estimate amount, and (c) a note explaining the reason for the increased estimate, along with a proposal for 'domestic laminate flooring' which has low risk and is inexpensive as an alternative option. The generated proposal is automatically sent via email to 'Mr. and Mrs. Kim' through the communication unit (300). Subsequently, a behavioral pattern in which 'Mr. and Mrs. Kim' view the proposal link and, in particular, click the 'domestic laminate flooring' option multiple times to check detailed information is detected through the communication unit (300) and the customer information collection module (110). This interaction data is immediately transmitted to the processing unit (100).

[0161] The workflow management module re-executes the proposal complexity determination process based on new customer behavior data. The system recognizes that the customer is showing high interest in a specific alternative, and this time, by selecting a brief summary template, it generates a second proposal summarizing the final quote and design changes when applying the specific alternative, and can request approval from the person in charge to send it.

[0162] After reviewing the second proposal, the person in charge delivers it to the "Mr. and Mrs. Kim," and the customer is satisfied and signs the final contract. The person in charge inputs the final result, "Contract Success," into the system. The "Success" result received through the communication unit (300) is recorded in the "Data Warehouse" of the storage unit (200) along with all data of the "Mr. and Mrs. Kim Project." The result analysis module (170), which runs periodically, retrains the entire data including this new success case to update the prediction models. In particular, it learns a new rule pattern: "When proposing imported materials during the typhoon season, a strategy of clearly disclosing risks and proposing price-competitive domestic alternative materials together increases the contract success rate of 'budget-sensitive' customers." After verifying the new model enhanced through this learning result by comparing it with the existing model, if performance improvement is confirmed, the module automatically transmits a command to replace the operational model stored in the "Model and Rule Registry" of the storage unit (200) with the new model to improve the overall prediction accuracy of the system and the sophistication of the proposal strategy.

[0163] Although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above. Various modifications are possible by those skilled in the art without departing from the essence of the invention as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present invention. Explanation of the symbols

[0164] Processing unit (100) Customer information collection module (110) Data preprocessing module (120) Order prediction module (130) Design construction strategy recommendation module (140) Proposal generation module (150) Automatic estimation module (160) Result analysis module (170) Storage unit (200) Communication unit (300)

Claims

Claim 1 An automatic optimization system for construction schedules and quotation strategies based on contract success probability, comprising: a storage unit for storing data; a communication unit for transmitting and receiving data with an external system; and a processing unit for controlling the overall operation of the system, wherein the processing unit comprises: a customer information collection module that collects customer information including text or image data representing a customer's preferred style and physical information of a construction target space through the communication unit; an order prediction module that determines a 'contract success probability grade' based on the customer information and past project data stored in the storage unit; and a design construction strategy recommendation module that limits the search range of a recommendation strategy using the contract success probability grade as an input value, derives a plurality of construction strategy candidate groups according to the limited search range, and determines a 'construction schedule robustness grade' for the construction schedules of the construction strategy candidate groups based on 'work team synchronization vulnerability,' which represents a risk arising from the complexity of process management within the project, and 'material supply weather coupling,' which represents a risk arising from the combination of supply chain instability and extreme weather events outside the project. An automatic estimation module that calculates a final estimate by combining the physical risks of the project and the financial reliability of the customer, determines a basic risk premium rate by referring to a 'risk-premium rate mapping table' based on judgment criteria including site noise interaction coefficients and logistics bottleneck probabilities, and calculates a 'final applicable premium rate' by adjusting the basic risk premium rate according to the payment reliability speed;A system for automatically optimizing construction schedules and estimation strategies based on contract success probability, wherein the processing unit determines the construction schedule and estimation strategy selected according to evaluation criteria based on the contract success probability grade, the construction schedule robustness grade, and the final applicable surcharge rate, and the design construction strategy recommendation module determines the construction schedule robustness grade by additionally applying a 'synergy risk penalty' separately from the evaluation of individual risks when a logical AND condition is satisfied in which the work team synchronization vulnerability exceeds a first threshold and the material supply weather coupling degree simultaneously exceeds a second threshold. Claim 2 delete

Images