Intelligent demand management replenishment system for construction materials based on predictive modeling

Abstract

The present application relates to the technical field of intelligent construction material management and wisdom logistics, in particular to a construction material intelligent demand management supply system based on predictive modeling; comprising data acquisition, trust quantification, entropy value analysis and strategy determination unit; the system compares the predicted data and the on-site reported data through the space-time graph neural network, and calculates the trust weight; the core is to calculate the scheduling entropy value according to the trust weight and the transport capacity resources, and compare it with the collapse threshold; when the entropy value triggers different states, the system adaptively generates cost priority or reverse saturation supply instructions; the present application establishes a dynamic credit potential analysis mechanism, effectively identifies malicious false reporting, and solves the problem of data distortion caused by interest game.

Description

Technical Field

[0001] This invention relates to the field of intelligent construction material management and smart logistics technology, specifically to an intelligent demand management and replenishment system for construction materials based on predictive modeling. Background Technology

[0002] In complex construction environments, network latency and inter-agency conflicts often lead to distorted material demand data reported by construction nodes. Existing material replenishment schemes generally lack dynamic quantitative assessment of data reliability, relying solely on simple numerical statistics for scheduling, making it difficult to distinguish between normal demand fluctuations and malicious misreporting. This passive response model cannot effectively balance economic costs and supply security when faced with panic buying by subcontractors due to anticipated shortages, and is prone to systemic logistical deadlocks or collapses due to localized data falsification. Therefore, how to construct a trust potential analysis mechanism based on multi-dimensional features and adaptively switch between cost-priority and inverse game replenishment strategies according to the system entropy state to ensure the robustness and stability of the logistics network in a data-distorted environment has become an urgent technical problem to be solved. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides a building materials intelligent demand management and replenishment system based on predictive modeling. Specifically, the technical solution of this invention includes:

[0004] The data acquisition and simulation center is used to retrieve material demand information for construction nodes and send the material demand information to the trust quantification calculation unit for credit potential analysis to obtain trust weights.

[0005] The system entropy analysis unit is used to perform thermodynamic state assessment on the trust weight, and to calculate the scheduling entropy value by combining the normalization ratio of transportation capacity resources and the demand volatility.

[0006] The strategy mode determination unit is used to compare and analyze the scheduling entropy value with the preset system crash threshold.

[0007] If the scheduling entropy value is less than the preset system crash threshold, a cost priority signal is generated.

[0008] If the scheduling entropy value is greater than or equal to the preset system collapse critical threshold, an inverse game signal is generated.

[0009] When generating a cost-priority signal, the adaptive supply generation unit calls the linear programming solver to generate the optimal supply instruction.

[0010] When generating an inverse game signal, the adaptive supply generation unit generates a reverse saturation supply instruction based on the trust weight.

[0011] Preferably, the credit potential analysis process is as follows:

[0012] Obtain the on-site reported data of construction nodes within the current monitoring period;

[0013] The spatiotemporal graph neural network model is used to generate baseline prediction data for the corresponding time.

[0014] Calculate the numerical deviation vector between the field-reported data and the baseline predicted data;

[0015] Perform time-dimensional integration on all numerical deviation vectors from historical moments to the current moment to obtain the cumulative deviation value;

[0016] The cumulative deviation value is normalized by using the preset maximum storage capacity parameter of the physical storage yard to obtain the normalized cumulative deviation value;

[0017] The normalized cumulative deviation value is substituted into the preset exponential decay function model for mapping processing to obtain the trust weight.

[0018] Preferably, the analysis process of the system entropy analysis unit is as follows:

[0019] Obtain the normalized proportion of transportation capacity resources occupied by construction nodes;

[0020] Calculate the natural logarithm of the normalized ratio of transport capacity resources;

[0021] The steady-state component is obtained by multiplying the normalized ratio of transport capacity resources, the natural logarithm, and the trust weight and then inverting the product.

[0022] Obtain the time derivative of the data reported from the construction site at each construction node to obtain the demand volatility;

[0023] Demand volatility is normalized using a pre-defined theoretical maximum consumption rate to obtain normalized volatility values.

[0024] Calculate the difference between the value 1 and the trust weight to obtain the distrust coefficient;

[0025] The wave dynamic component is obtained by multiplying the normalized fluctuation value with the confidence coefficient.

[0026] The scheduling entropy value is obtained by weighted summation of the steady-state component and the wave dynamic component.

[0027] Preferably, the generation process of the reverse saturation replenishment command is as follows:

[0028] Obtain the historical average consumption and preset saturation attack coefficient of the construction node;

[0029] The basic saturation level is obtained by multiplying the historical average consumption by the preset saturation attack coefficient.

[0030] Divide the base saturation amount by the trust weight to obtain the trust compensation supply amount;

[0031] Obtain the maximum storage capacity of the pre-set physical stockpile at the construction node;

[0032] The amount of trust compensation supply is numerically compared with the maximum storage capacity of the preset physical storage yard.

[0033] If the trust compensation supply amount is less than the preset maximum physical storage capacity, the trust compensation supply amount will be set as a reverse saturation supply instruction.

[0034] If the trust compensation supply amount is greater than or equal to the preset maximum physical storage capacity, the preset maximum physical storage capacity will be set as the reverse saturation supply command.

[0035] Preferably, the preset exponential decay function model is configured with a trust decay sensitivity coefficient;

[0036] Obtain historical data on supply disruptions and quality incidents;

[0037] Reverse calibration analysis of historical supply interruption quality incident data;

[0038] If historical supply disruption incident data shows that small deviations can lead to major incidents, the sensitivity coefficient for trust decay should be increased.

[0039] Preferably, the process for obtaining the demand volatility is as follows:

[0040] Perform differential calculation on the material demand vector reported in real time at construction nodes;

[0041] Obtain the change in demand per unit time;

[0042] The change in demand is set as the demand volatility.

[0043] Preferably, the preset system collapse threshold is determined based on the logistics deadlock critical point derived from the simulation sand table; the maximum storage capacity of the physical storage yard is derived from the building information model database.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1. This system constructs a data acquisition and simulation center and a trust quantification calculation unit, uses a spatiotemporal graph neural network model to generate objective benchmark prediction data, and compares it with the data reported on-site. The system performs time-dimensional integral calculation on the numerical deviation vector between the two, and normalizes and performs exponential decay mapping in combination with the preset maximum storage capacity of the physical storage yard, thereby establishing a dynamic credit potential analysis mechanism. This mechanism can effectively distinguish between normal demand fluctuations and malicious false reporting, solve the data distortion problem caused by interest games in isolated projects, and ensure that scheduling decisions are based on real and reliable quantitative trust weights.

[0046] 2. This system introduces a system entropy analysis unit, unifying heterogeneous dimensions such as logistics dispersion, data unreliability, and demand volatility into a single scheduling entropy index. By weighting and evaluating the normalization ratio of transport capacity resources, trust weight, and demand volatility, the system can perceive the stability of the logistics network in real time. This multi-dimensional feature fusion evaluation method overcomes the limitations of existing solutions that rely solely on single numerical statistics, accurately predicting whether the system is approaching the collapse threshold, and providing a scientific quantitative basis for subsequent strategy switching.

[0047] 3. This system achieves adaptive switching between cost-priority mode and inverse game mode through a strategy mode determination unit. When the scheduling entropy value is lower than the preset system collapse threshold, the system uses a linear programming solver to generate the optimal supply instruction to minimize economic costs. When the entropy value exceeds the threshold, indicating that the system is in a high-risk state, it automatically switches to inverse game mode. This dual-modal control mechanism effectively balances the economic benefits under normal conditions with the supply security under extreme conditions, prevents systemic logistics deadlock caused by the accumulation of local risks, and significantly improves the robustness of the logistics network.

[0048] 4. This system adopts a reverse saturation replenishment strategy under the inverse game model, generating reverse saturation replenishment instructions based on trust weights. The system utilizes the counterintuitive control logic that the lower the trust level, the greater the replenishment compensation, and quickly fills the node inventory under the constraint of the maximum physical storage capacity. This method of eliminating the subcontractors' psychological shortage expectations through excessive saturation supply of physical quantities can effectively curb malicious hoarding behavior caused by panic, force construction nodes to return to a rational state of truthful reporting as needed, and block the impact of data fraud on the logistics system. Attached Figure Description

[0049] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0050] Figure 1 This is a system block diagram of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0052] Example 1:

[0053] Please see Figure 1 A predictive modeling-based intelligent demand management and replenishment system for building materials includes:

[0054] The data acquisition and simulation center is used to retrieve material demand information for construction nodes and send the material demand information to the trust quantification calculation unit for credit potential analysis to obtain trust weights.

[0055] The system entropy analysis unit is used to perform thermodynamic state assessment on the trust weight, and to calculate the scheduling entropy value by combining the normalization ratio of transportation capacity resources and the demand volatility.

[0056] The strategy mode determination unit is used to compare and analyze the scheduling entropy value with the preset system crash threshold.

[0057] If the scheduling entropy value is less than the preset system crash threshold, a cost priority signal is generated.

[0058] If the scheduling entropy value is greater than or equal to the preset system collapse critical threshold, an inverse game signal is generated.

[0059] When generating a cost-priority signal, the adaptive supply generation unit calls the linear programming solver to generate the optimal supply instruction.

[0060] When generating an inverse game signal, the adaptive supply generation unit generates a reverse saturation supply instruction based on the trust weight.

[0061] This embodiment is a construction material intelligent demand management and supply system based on predictive modeling. It addresses the problem of distorted material demand data caused by network latency and interest game in isolated projects by constructing a dynamic trust and game control mechanism. The system includes a data acquisition and simulation center, a system entropy analysis unit, a strategy mode determination unit, and an adaptive supply generation unit.

[0062] The data acquisition and simulation center retrieves material demand information for construction nodes. The information is then transmitted to the trust quantification calculation unit. This unit introduces the concepts of reliability engineering decay function and field theory potential energy accumulation to perform credit potential energy analysis on the subcontractor's historical misrepresentation behavior and calculate the trust weight. ;

[0063] The system entropy analysis unit performs a thermodynamic state assessment on the trust weight; this unit combines the normalized ratio of transport capacity resources, the trust weight, and the demand volatility to perform a weighted calculation, outputting the scheduling entropy value. This unifies the two heterogeneous dimensions of logistics dispersion and data unreliability into a single scalar indicator.

[0064] The strategy mode determination unit acts as a mode switching switch, which determines the scheduling entropy value. Compared to the preset system crash threshold Perform comparative analysis; when the scheduling entropy value When the system is in a stable state, it generates a cost-priority signal; when the scheduling entropy value... At this time, the system is in a high-risk state and generates an inverse game signal;

[0065] The adaptive supply generation unit executes different strategies based on the received signals; when it receives a cost-priority signal, it calls the linear programming solver to generate the optimal supply command that minimizes economic costs. Specifically, the following linear programming model is constructed:

[0066] The objective function is to minimize the total generalized cost. :

[0067]

[0068] in, The cost of transporting goods per unit distance per unit quantity. For transportation distance, This refers to the unit price for material purchases.

[0069] The constraints include:

[0070] Node physical storage capacity constraints:

[0071]

[0072] in, For nodes Current inventory levels;

[0073] System total capacity limit constraint:

[0074]

[0075] in, This represents the maximum number of transport vehicles that the system can currently dispatch.

[0076] Upon receiving an inverse game signal, a reverse saturation replenishment instruction is generated based on the trust weights. This configuration eliminates subcontractors' expectations of shortages by saturating supply; it also enables the system to adaptively switch between economic and emergency modes, preventing systemic logistical collapses caused by localized data falsification.

[0077] Example 2:

[0078] The credit potential analysis process is as follows:

[0079] Obtain the on-site reported data of construction nodes within the current monitoring period;

[0080] The spatiotemporal graph neural network model is used to generate baseline prediction data for the corresponding time.

[0081] Calculate the numerical deviation vector between the field-reported data and the baseline predicted data;

[0082] Perform time-dimensional integration on all numerical deviation vectors from historical moments to the current moment to obtain the cumulative deviation value;

[0083] The cumulative deviation value is normalized by using the preset maximum storage capacity parameter of the physical storage yard to obtain the normalized cumulative deviation value;

[0084] The normalized cumulative deviation value is substituted into the preset exponential decay function model for mapping processing to obtain the trust weight.

[0085] The data acquisition and simulation center performs a credit potential analysis process, aiming to quantify the credibility of construction node behaviors through a non-linear penalty mechanism.

[0086] Acquire on-site reported data from construction nodes via IoT sensors or manual terminals during the current monitoring period. ,in, Representing time variables; constructing graph structures ,in, For the set of construction nodes, The set of transport routes connecting nodes; adjacency matrix Based on the physical Euclidean distance between nodes Build, satisfy ,in, This is the distance attenuation coefficient;

[0087] Spatiotemporal graph neural networks use graph convolutional layers to extract spatial features, as shown in the formula:

[0088]

[0089] in, , It is the identity matrix. For degree matrix, For learnable weight parameters, The activation function is used; the model input features include historical material demand sequences and date attribute features; the specific fusion method is as follows: the date attribute features are one-hot encoded and mapped into a high-dimensional vector, and after dimensional alignment through a fully connected layer, they are concatenated with the historical material demand sequences along the feature channel dimension to form a fused input tensor. The loss function uses mean squared error loss. Conduct training;

[0090] Using a spatiotemporal graph neural network model, baseline prediction data for corresponding time points is generated based on topological dependencies. As an objective verification benchmark; calculate the numerical deviation vector between the field-reported data and the benchmark prediction data. ;

[0091] Historical moments up to the current moment Integrate all numerical deviation vectors over time to obtain the cumulative deviation value. ; Utilizing preset physical storage yard maximum capacity parameters derived from the BIM model database The cumulative deviation value is normalized to eliminate the influence of node size differences;

[0092] The normalized cumulative deviation value is substituted into the preset exponential decay function model for mapping processing, and the trust weight is calculated. The formula is as follows:

[0093]

[0094] in, This represents the absolute value of the average deviation per unit time, which is controlled by a constant parameter. Divide by and normalize to obtain the dimensionless deviation rate; This is a dimensionless trust decay sensitivity coefficient; the formula introduces a time normalization factor. This eliminates the problem of cumulative error divergence caused by increased monitoring time, ensuring that the trust weight reflects the average performance level of the node throughout its entire lifecycle.

[0095] Example 3:

[0096] The analysis process of the system entropy analysis unit is as follows:

[0097] Obtain the normalized proportion of transportation capacity resources occupied by construction nodes;

[0098] Calculate the natural logarithm of the normalized ratio of transport capacity resources;

[0099] The steady-state component is obtained by multiplying the normalized ratio of transport capacity resources, the natural logarithm, and the trust weight and then inverting the product.

[0100] Obtain the time derivative of the data reported from the construction site at each construction node to obtain the demand volatility;

[0101] Demand volatility is normalized using a pre-defined theoretical maximum consumption rate to obtain normalized volatility values.

[0102] Calculate the difference between the value 1 and the trust weight to obtain the distrust coefficient;

[0103] The wave dynamic component is obtained by multiplying the normalized fluctuation value with the confidence coefficient.

[0104] The scheduling entropy value is obtained by weighted summation of the steady-state component and the wave dynamic component.

[0105] The process of obtaining demand volatility is as follows:

[0106] Perform differential calculation on the material demand vector reported in real time at construction nodes;

[0107] Obtain the change in demand per unit time;

[0108] The change in demand is set as the demand volatility.

[0109] The system entropy analysis unit and the demand volatility acquisition process aim to provide a unified assessment of logistics dispersion and demand volatility.

[0110] The process of obtaining demand volatility involves performing time difference calculations on the material demand vectors reported in real time at construction nodes. Obtain the change in demand per unit time and set it as the demand volatility. This characterizes the degree of sudden and drastic change in node demand;

[0111] The system entropy analysis unit calculates the entropy based on the trust-weighted scheduling entropy state equation. The corrected formula is as follows:

[0112]

[0113] The revised calculation logic is as follows:

[0114] Before that, it is necessary to clarify The calculation definition is as follows: For the first At the current moment, each construction node... The normalized proportion of occupied transportation capacity resources is calculated using the following formula:

[0115] ,in, For the first The amount of transportation resources currently used by each node. This represents the total current operating capacity of the system; among which, A preset numerical stability constant is used to prevent overflow in logarithmic operations;

[0116] First item As an information entropy component, since lead to Since the value is negative, a negative sign is introduced to make it positive, representing the degree of dispersion in the distribution of logistics.

[0117] Second item This represents the volatility risk component, and is a non-negative value.

[0118] Scheduling entropy The physical meaning of this is: it characterizes the dispersion of transportation capacity resources and the fluctuation risk of demand for goods; when the weighted sum of the two increases, it means that the system is in a disordered state of excessively dispersed transportation capacity and drastically fluctuating demand; after the two are added together, the scheduling entropy value... The higher the threshold, the more unstable the system, thus maintaining consistency with the logic of triggering a reverse game signal when the threshold is exceeded.

[0119] Example 4:

[0120] The process of generating the reverse saturation replenishment command is as follows:

[0121] Obtain the historical average consumption and preset saturation attack coefficient of the construction node;

[0122] The basic saturation level is obtained by multiplying the historical average consumption by the preset saturation attack coefficient.

[0123] Divide the base saturation amount by the trust weight to obtain the trust compensation supply amount;

[0124] Obtain the maximum storage capacity of the pre-set physical stockpile at the construction node;

[0125] The amount of trust compensation supply is numerically compared with the maximum storage capacity of the preset physical storage yard.

[0126] If the trust compensation supply amount is less than the preset maximum physical storage capacity, the trust compensation supply amount will be set as a reverse saturation supply instruction.

[0127] If the trust compensation supply amount is greater than or equal to the preset maximum physical storage capacity, the preset maximum physical storage capacity will be set as the reverse saturation supply command.

[0128] The generation process of the reverse saturation supply instruction is based on the inverse game mechanism, which eliminates the psychological expectation of shortage through the oversaturation of physical quantities;

[0129] Obtain the historical average consumption of construction nodes and preset saturation attack coefficient The base saturation level is obtained by multiplying the historical average consumption by the preset saturation attack coefficient. Divide the base saturation amount by the trust weight to obtain the trust compensation supply amount. Due to trust weight The lower the node's trust level, the greater the calculated trust compensation amount.

[0130] Obtain the maximum storage capacity of the pre-set physical storage yard for the construction node. The trust compensation supply amount is compared with the preset maximum physical storage capacity, and the reverse saturation supply control logic is executed.

[0131]

[0132] If the trust compensation supply amount is less than the preset maximum physical storage capacity, the trust compensation supply amount is set as a reverse saturation supply instruction; if the trust compensation supply amount is greater than or equal to the preset maximum physical storage capacity, the preset maximum physical storage capacity is set as a reverse saturation supply instruction. This strategy, in a high-risk system state, eliminates the expected shortage by quickly filling the physical storage capacity, forcing subcontractors to return to on-demand reporting. To prevent all nodes from simultaneously triggering maximum saturation supply, causing a collapse in the system's total capacity, the system further executes a total capacity safety net constraint logic:

[0133] Calculate the total demand for reverse saturation supply instructions generated at the current moment. ;

[0134] like ,in, To determine the system's total capacity limit, all supply instructions for all nodes will be scaled and adjusted proportionally.

[0135] Qiactual(t)=Qifinal(t)·Vtotal(t)Qtotal_demand(t)

[0136] This step ensures that while the saturation attack eliminates psychological expectations, the physical logistics network does not deadlock due to exceeding its capacity limit.

[0137] Example 5:

[0138] The preset exponential decay function model is configured with a trust decay sensitivity coefficient;

[0139] Obtain historical data on supply disruptions and quality incidents;

[0140] Reverse calibration analysis of historical supply interruption quality incident data;

[0141] If historical supply disruption incident data shows that small deviations can lead to major incidents, the sensitivity coefficient for trust decay should be increased.

[0142] The preset exponential decay function model is configured with a trust decay sensitivity coefficient. To ensure the safety of parameter settings and the independence of variables, the coefficient is configured through reverse calibration analysis.

[0143] Establish a historical accident calibration dataset This dataset contains material supply deviation samples corresponding to historical cold-seam accident records. and the severity level of the accident ; here and It is only used during the parameter calibration phase and is unrelated to real-time data during system operation; Perform regression analysis; if the analysis results show a small sample of material supply deviation... With a higher accident severity level If a significant positive correlation exists, then increase the trust decay sensitivity coefficient. This self-learning mechanism ensures that for high-risk nodes, even a small deviation in demand will cause a sharp drop in trust weight, thereby triggering a more stringent replenishment strategy.

[0144] Example 6:

[0145] The preset system crash threshold is determined based on the logistics deadlock critical point derived from the simulation sand table; the maximum storage capacity of the physical storage yard is derived from the building information model database.

[0146] Preset system crash threshold Based on digital twin simulation sandbox analysis, before system deployment, multiple rounds of stress tests were conducted to simulate the operation of the logistics network under different trust levels, identifying entropy inflection points that could lead to deadlock or paralysis of the logistics network. Maximum storage capacity of physical storage yard It reads directly from the project's building information model database to ensure that the physical constraint parameters are consistent with the actual site conditions.

[0147] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A construction material intelligent demand management and replenishment system based on predictive modeling, characterized in that, include: The data acquisition and simulation center is used to retrieve material demand information for construction nodes and send the material demand information to the trust quantification calculation unit for credit potential analysis to obtain trust weights. The system entropy analysis unit is used to perform thermodynamic state assessment on the trust weight, and to calculate the scheduling entropy value by combining the normalization ratio of transportation capacity resources and the demand volatility. The strategy mode determination unit is used to compare and analyze the scheduling entropy value with the preset system crash threshold. If the scheduling entropy value is less than the preset system crash threshold, a cost priority signal is generated. If the scheduling entropy value is greater than or equal to the preset system collapse critical threshold, an inverse game signal is generated. When generating a cost-priority signal, the adaptive supply generation unit calls the linear programming solver to generate the optimal supply instruction. When generating an inverse game signal, the adaptive supply generation unit generates a reverse saturation supply instruction based on the trust weight.

2. The intelligent demand management and supply system for building materials based on predictive modeling according to claim 1, characterized in that, The credit potential analysis process is as follows: Obtain the on-site reported data of construction nodes within the current monitoring period; The spatiotemporal graph neural network model is used to generate baseline prediction data for the corresponding time. Calculate the numerical deviation vector between the field-reported data and the baseline predicted data; Perform time-dimensional integration on all numerical deviation vectors from historical moments to the current moment to obtain the cumulative deviation value; The cumulative deviation value is normalized by using the preset maximum storage capacity parameter of the physical storage yard to obtain the normalized cumulative deviation value; The normalized cumulative deviation value is substituted into the preset exponential decay function model for mapping processing to obtain the trust weight.

3. The intelligent demand management and supply system for building materials based on predictive modeling according to claim 1, characterized in that, The analysis process of the system entropy analysis unit is as follows: Obtain the normalized proportion of transportation capacity resources occupied by construction nodes; Calculate the natural logarithm of the normalized ratio of transport capacity resources; The steady-state component is obtained by multiplying the normalized ratio of transport capacity resources, the natural logarithm, and the trust weight and then inverting the product. Obtain the time derivative of the data reported from the construction site at each construction node to obtain the demand volatility; Demand volatility is normalized using a pre-defined theoretical maximum consumption rate to obtain normalized volatility values. Calculate the difference between the value 1 and the trust weight to obtain the distrust coefficient; The wave dynamic component is obtained by multiplying the normalized fluctuation value with the confidence coefficient. The scheduling entropy value is obtained by weighted summation of the steady-state component and the wave dynamic component.

4. The intelligent demand management and supply system for building materials based on predictive modeling according to claim 1, characterized in that, The process of generating the reverse saturation replenishment command is as follows: Obtain the historical average consumption and preset saturation attack coefficient of the construction node; The basic saturation level is obtained by multiplying the historical average consumption with the preset saturation attack coefficient. Divide the base saturation amount by the trust weight to obtain the trust compensation supply amount; Obtain the maximum storage capacity of the pre-set physical stockpile at the construction node; The amount of trust compensation supply is numerically compared with the maximum storage capacity of the preset physical storage yard. If the trust compensation supply amount is less than the preset maximum physical storage capacity, the trust compensation supply amount will be set as a reverse saturation supply instruction. If the trust compensation supply amount is greater than or equal to the preset maximum physical storage capacity, the preset maximum physical storage capacity will be set as the reverse saturation supply command.

5. The intelligent demand management and supply system for building materials based on predictive modeling according to claim 2, characterized in that, The preset exponential decay function model is configured with a trust decay sensitivity coefficient. Obtain historical data on supply disruptions and quality incidents; Reverse calibration analysis of historical supply interruption quality incident data; If historical supply disruption incident data shows that small deviations can lead to major incidents, the sensitivity coefficient for trust decay should be increased.

6. The intelligent demand management and replenishment system for building materials based on predictive modeling according to claim 3, characterized in that, The process for obtaining the demand volatility is as follows: Perform differential calculation on the material demand vector reported in real time at construction nodes; Obtain the change in demand per unit time; The change in demand is set as the demand volatility.

7. The intelligent demand management and supply system for building materials based on predictive modeling according to claim 1, characterized in that, The preset system collapse threshold is determined based on the logistics deadlock critical point derived from the simulation sand table; the maximum storage capacity of the physical storage yard is derived from the building information model database.

Images