Hotel management method and system based on AI

Abstract

The invention discloses an AI-based hotel management method and system, and relates to the technical field of intelligent management with integration of artificial intelligence and hotel management, and the method comprises the steps: collecting and preprocessing hotel data, screening the hotel data with high correlation characteristics through employing an improved chaos grass chopper optimization algorithm, introducing external data, and carrying out the optimization of the hotel data. Using a Monte Carlo sampling method to simulate hotel material demand samples in a random demand scene; using a grey wolf optimization algorithm to optimize the inventory cost and the storage space utilization rate; according to the method, a supply chain network model is constructed, feature extraction is performed, supplier selection and logistics scheduling joint optimization is performed by using a mucous algorithm, and dynamic adjustment is performed on a supply chain based on emergencies, and feature screening is performed through an improved chaos grass chopper optimization algorithm, so that the quality of model input data is improved; a supply chain network model is constructed, and joint optimization of supplier selection and logistics scheduling is carried out by using a mucous algorithm, so that the overall operation efficiency of the supply chain is enhanced.

Description

Technical Field

[0001] This invention relates to the field of intelligent management technology that integrates artificial intelligence with hotel management, and in particular to an AI-based hotel management method and system. Background Technology

[0002] With the continuous development of artificial intelligence, big data analytics, and supply chain management technologies, in order to improve hotel management efficiency and optimize inventory allocation, more and more researchers and companies are beginning to try to introduce machine learning, optimization algorithms, and data-driven methods into the hotel operation and management process.

[0003] However, existing technologies still have significant shortcomings. Most systems still use traditional methods in the feature selection stage, which cannot effectively remove redundant or low-relevance features, resulting in low model training efficiency and prediction accuracy. Current hotel inventory forecasting methods generally use static demand forecasting models, which cannot fully account for random demand fluctuations and lack flexibility and robustness. Traditional inventory optimization often overlooks the limiting factors of storage space and fails to dynamically adjust inventory levels in accordance with real-time market demand; In terms of supply chain management, static or rule-driven scheduling mechanisms are still used, which are difficult to adapt to the challenges of frequent emergencies in complex environments. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an AI-based hotel management method and system to address the problem that most systems still use traditional methods in the feature selection stage, which cannot effectively remove redundant or low-relevance features, resulting in low model training efficiency and prediction accuracy. Current hotel inventory forecasting methods generally use static demand forecasting models, which cannot fully consider random demand fluctuations and lack flexibility and robustness. Traditional inventory optimization often ignores the limiting factors of storage space and fails to dynamically adjust inventory levels in combination with real-time market demand. In terms of supply chain management, static or rule-driven scheduling mechanisms are still used, which are difficult to adapt to the challenges of frequent emergencies in complex environments.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides an AI-based hotel management method, comprising: Hotel data was collected and preprocessed. An improved Chaotic Grasshopper optimization algorithm was used to filter hotel data with high correlation features. A random forest regression model was constructed to introduce external data into the hotel data with high correlation features. The Monte Carlo sampling method was used to simulate hotel material demand samples in random demand scenarios. Construct a TCN model to predict material demand, use the Grey Wolf optimization algorithm to optimize inventory costs and warehouse space utilization, and generate inventory adjustment instructions. We construct a supply chain network model, extract features, use slime mold algorithm to jointly optimize supplier selection and logistics scheduling, and dynamically adjust the supply chain based on unforeseen circumstances. Collect real-time data to update the model, store the collected and analyzed hotel data, and build a visual interface to display the analyzed hotel data.

[0007] As a preferred embodiment of the AI-based hotel management method of the present invention, the step of collecting and preprocessing hotel data, and simulating hotel material demand samples for random demand scenarios using an improved Chaotic Grasshopper optimization algorithm, external data, and Monte Carlo sampling method, includes: The hotel data includes hotel inventory, hotel operations, and external data; Each feature in the feature vector is binary encoded, the initial position of the grasshopper is generated using Logistic-Sine-Cosine chaotic mapping, the continuous positions of the grasshopper are converted into binary vectors using a logistic function, and the binary values ​​of all grasshoppers are used to generate a feature subset. Define hotel operation data and material consumption function, calculate fitness value, aim to maximize fitness value, use improved chaotic grasshopper optimization algorithm to update feature subset, calculate median of fitness value of all grasshoppers after each iteration, update grasshopper position for grasshoppers with fitness higher than median, otherwise update manta ray spiral foraging position. Add a Gaussian perturbation to the updated low-fitness grasshopper, and then use the Logistic-Sine-Cosine chaotic mapping to further update the low-fitness grasshopper with the added Gaussian perturbation; After reaching the maximum number of iterations, stop iterating and output c optimal feature subsets as the operational feature set, where c is the number of features of the hotel operational data with the highest fitness value corresponding to a single material consumption, set based on hotel requirements; For each type of material and its corresponding operational characteristics and external data, a subset dataset is generated, a random forest regression model is constructed, historical hotel data is collected and preprocessed as training data for the random forest regression model, and training is performed with the goal of minimizing the mean squared error. Based on external data to define scenario data, a trained random forest model is used to predict the incremental demand for materials in each scenario. The basic demand of the average material is adjusted, a Monte Carlo sampling method is used to generate random demand samples, and the 3σ principle is used to detect the coverage of the samples.

[0008] As a preferred embodiment of the AI-based hotel management method of the present invention, the step of constructing a TCN model for material demand forecasting, using the Grey Wolf optimization algorithm to optimize inventory costs and warehouse space utilization, and generating inventory adjustment instructions includes: The data in the integrated dataset is normalized, and the scenario and simulated demand samples for each type of material are extracted and constructed into an input matrix according to time. Construct a TCN model, collect historical data for training, input the input matrix into the trained TCN model, perform N multi-scale predictions, generate a prediction distribution, calculate the mean and variance of the N prediction results, perform prediction correction, and use exponential moving average (EMA) smoothing to obtain the final prediction result. Construct a multi-objective function to optimize inventory costs and warehouse space utilization, set constraints, use the Grey Wolf MOGWO optimization algorithm to solve the multi-objective function, search for the Pareto optimal solution set, select the solution that meets the hotel's operational preferences, and generate inventory adjustment instructions. The system collects the latest inventory consumption time series data in real time, calculates the consumption Z-score for each material, sets an anomaly threshold and compares it with the Z-score. If the Z-score is greater than the anomaly threshold, it is judged as an anomaly, a record is generated for the anomaly material, and the automatic replenishment of the relevant material is suspended until manual verification is performed. Otherwise, it is judged as normal and replenishment continues.

[0009] As a preferred embodiment of the AI-based hotel management method of the present invention, the step of constructing a supply chain network model, extracting features, and using a slime mold algorithm for joint optimization of supplier selection and logistics scheduling includes: Collect and preprocess supplier data, and construct a directed graph with suppliers and hotels as nodes, logistics relationships from suppliers to hotels as edges, and delivery time and transportation cost as edge weights. To construct a GNN supply chain network model, historical supplier data and hotel data are collected, preprocessed, and then the model is trained. The trained GNN supply chain network model is output, and a directed graph is input for inference. From the embedding vectors of GNN nodes, supplier features are extracted, standardized, and then a supplier scoring formula and a total cost function are defined. Randomly generate slime mold populations, and initialize them, set constraints, and select anomaly handling options; The algorithm optimizes slime mold foraging by minimizing the total cost function. A maximum number of iterations is set, and the iteration stops after the maximum number of iterations is reached. The optimal solution for the current iteration is then output.

[0010] As a preferred embodiment of the AI-based hotel management method of the present invention, the step of dynamically adjusting the supply chain based on unforeseen circumstances includes: Collect real-time inventory data, supplier status, and logistics status; set rules for contingencies; and adjust purchase quantities accordingly. The system dynamically adjusts supply chain choices based on contingency rules and generates dynamic adjustment reports.

[0011] As a preferred embodiment of the AI-based hotel management method of the present invention, the step of collecting real-time data for model updating includes: Collect real-time consumption data for each type of material, calculate the real-time deviation between the real-time consumption data and the predicted value, set a deviation threshold using an empirical method, and if the real-time deviation is greater than the deviation threshold, incrementally train the TCN model based on the newly collected real-time data.

[0012] As a preferred embodiment of the AI-based hotel management method of the present invention, the step of storing, collecting, and analyzing the generated hotel data, and constructing a visual interface to display the analyzed hotel data, includes: The collected and analyzed data is stored in a distributed manner, and a visualization interface is built using D3.js to display the predictive performance chart, inventory optimization chart, and dynamic adjustment report.

[0013] Secondly, the present invention provides an AI-based hotel management system, comprising, The data collection and optimization module is used to collect and preprocess hotel data, and uses an improved Chaotic Grasshopper optimization algorithm to filter hotel data with highly relevant features. A sampling module is constructed to introduce external data into hotel data with high correlation characteristics to build a random forest regression model, and the Monte Carlo sampling method is used to simulate hotel material demand samples in random demand scenarios. The forecast generation module is used to build a TCN model for material demand forecasting, use the Grey Wolf optimization algorithm to optimize inventory costs and warehouse space utilization, and generate inventory adjustment instructions. The extraction and adjustment module is used to build a supply chain network model, extract features, use slime mold algorithm to jointly optimize supplier selection and logistics scheduling, and dynamically adjust the supply chain based on emergencies. The update module is used to collect real-time data for model updates; The storage visualization module is used to store the collected and analyzed hotel data and to build a visual interface to display the analyzed hotel data.

[0014] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein the computer program, when executed by the processor, implements any step of the AI-based hotel management method as described in the first aspect of the present invention.

[0015] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the AI-based hotel management method as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: By collecting and preprocessing hotel data, this invention uses an improved chaotic grasshopper optimization algorithm to filter hotel data with high correlation features, constructs a random forest regression model to introduce external data into the hotel data with high correlation features, and uses the Monte Carlo sampling method to simulate hotel material demand samples in random demand scenarios; it constructs a TCN model to predict material demand, uses the gray wolf optimization algorithm to optimize inventory costs and warehouse space utilization, and generates inventory adjustment instructions; it constructs a supply chain network model, performs feature extraction, uses the slime mold algorithm to jointly optimize supplier selection and logistics scheduling, and dynamically adjusts the supply chain based on emergencies; it collects real-time data to update the model, stores the collected and analyzed hotel data, and constructs a visual interface to display the analyzed hotel data, thereby improving the system's intelligence and decision-making efficiency, enhancing resource allocation efficiency and inventory control level, and strengthening the resilience and dynamic scheduling capabilities of the supply chain. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the AI-based hotel management method in Example 1.

[0019] Figure 2 This is a schematic diagram of the file encryption system in Example 1.

[0020] Figure 3 The flowchart is shown in Example 1 for the improved Chaos Grasshopper optimization algorithm.

[0021] Figure 4 This is a flowchart of the supply chain dynamic response and adjustment process in Example 1. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Example 1, referring to Figures 1 to 4 This is the first embodiment of the present invention, which provides an AI-based hotel management method, including the following steps: S1. Collect and preprocess hotel data, use the improved Chaotic Grasshopper optimization algorithm to filter hotel data with high correlation features, construct a random forest regression model to introduce external data into the hotel data with high correlation features, and use the Monte Carlo method sampling method to simulate hotel material demand samples in random demand scenarios. Specifically, hotel data is collected and preprocessed, and feature vectors are generated from the preprocessed hotel data; The hotel data includes hotel inventory, hotel operations, and external data; The hotel inventory data includes consumption data for food ingredients and daily necessities; The hotel's operational data includes occupancy rate, food and beverage order volume, and room cleaning frequency. The external data includes weather and holiday data; The preprocessing includes imputing missing values ​​using linear interpolation, detecting and deleting outliers using the Z-score method, and performing standardization. Each feature in the feature vector is binary encoded (each feature has two states, "selected" and "unselected"); The initial position of the grasshopper is generated using a Logistic-Sine-Cosine chaotic mapping, with the following formula: , , in, Let be the initial position of the i-th grasshopper in the j-th feature dimension. and represents the upper and lower bounds (0 and 1) of the feature dimension. Let be the chaotic variable of the i-th grasshopper in the j-th feature dimension. Based on Updated chaotic variables; Use a logical function to convert the continuous positions of the grasshopper into a binary vector, formula: , in, Let be the binary value of the i-th grasshopper in the j-th feature dimension. It is a logical function; Generate a feature subset from the binary values ​​of all grasshoppers. ; Define hotel operation data and material consumption functions, calculate fitness value, formula: , in, Let i be the fitness value of the i-th hopper. For the feature subset of the i-th grasshopper, For the j-th feature The Pearson correlation coefficient between (e.g., occupancy rate) and material consumption y (e.g., beef consumption). For the total number of features, and The weights for balancing correlation and feature quantity are set based on the actual situation. With the goal of maximizing fitness, the improved chaotic grasshopper optimization algorithm is used to update the feature subset. After each iteration, the median of the fitness values ​​of all grasshoppers is calculated. Grasshoppers with fitness values ​​higher than the median are updated in position, otherwise the manta ray spiral foraging position is updated. The grasshopper's position is updated using the following formula: , in, For the i-th high-fitness grasshopper in the first... The position of the next iteration. This is the shrinkage coefficient, which decreases with each iteration. The maximum number of iterations, and These are the maximum and minimum shrinkage coefficients, set using a fixed-value method. For social interaction functions, For the first Only highly adaptable grasshoppers in the first The position of the next iteration. For the i-th high-fitness grasshopper and the i-th Only the Euclidean distance of highly adaptable grasshoppers, For the first The optimal position (maximum fitness value) in the next iteration. and These are the upper and lower limits of the update position for i highly fit grasshoppers, respectively; The manta ray spiral foraging location is updated using the following formula: , in, For the i-th low-fitness grasshopper in the first... The position of the next iteration. The numbers are random and are uniformly distributed within the range of 0-1. The spiral shape parameter is a fixed value set manually. It is the natural logarithm; Add Gaussian perturbation to the updated low-fitness grasshopper to enhance its global search capability. Use Logistic-Sine-Cosine chaotic mapping to further update the low-fitness grasshopper with added Gaussian perturbation. After reaching the maximum number of iterations, stop iterating and output c optimal feature subsets as the operational feature set, where c is the number of features of the hotel operational data with the highest fitness value corresponding to a single material consumption, set based on hotel requirements; For each material and its corresponding operational characteristics, along with external data, generate a subset dataset, using the formula: , in, For timestamps, For supplies In time The standardized value, For supplies The corresponding number One operational characteristic in time Standardized values ​​(such as restaurant order volume). For the first External data features in time The standardized value (if it is a holiday feature, it is marked as 1 for holidays, otherwise it is 0, and other data are represented by standardized values). Construct a random forest regression model, formula: , in, For the random forest regression model, This is random error; Historical hotel data was collected and preprocessed to serve as training data for a random forest regression model, with the goal of minimizing the mean squared error. Defining scenario data based on external data: The holiday label dataset is (0,1), where 0 represents no holidays and 1 represents holidays. Based on the weather forecast, the temperature is divided into three ranges: low temperature range, normal temperature range, and high temperature range; Based on weather forecasts, humidity is divided into three ranges: low humidity, normal humidity, and high humidity (relative humidity). The system matches holidays, temperature ranges, and humidity ranges to simulate 18 different scenarios (such as holidays, high temperature, and high humidity). For each scenario, it sets external feature values ​​(such as holiday 1, temperature 0.8, humidity 0.8) and operational feature mean values ​​(the operational feature is taken as the mean value to focus on the impact of external data and reduce the complexity of scenario definition), forming a scenario set (including each scenario and its corresponding external feature value and operational feature mean value). The incremental demand for resources in each scenario is predicted using a pre-trained random forest model, using the following formula: , in, For supplies In the scene incremental demand, In the scene The One external feature value, In the scene The Mean of each operational characteristic; The basic demand for goods is adjusted based on incremental demand, using the formula: , in, For scenario-based For supplies Adjusted value of the mean For supplies The mean is obtained based on historical regression statistics; Based on the adjusted material mean, a random demand sample is generated using the Monte Carlo sampling method, formula: , in, For supplies In the scene The Secondary sampling requires samples, Using random forest regression to capture the influence of external data For supplies The standard deviation is used to preserve historical volatility characteristics. Let be a standard normally distributed random variable. This represents the total number of times the phrase was used. The 3σ principle is used to detect sample coverage, using the formula: , , in, and These are the minimum and maximum values ​​in the sample required for sampling; If the lower limit exceeds the range Or the upper limit exceeds the range. The standard deviation of the materials was recalculated based on historical data, and the total number of samples was increased for resampling. The data, including timestamps, material types, operational characteristics, external characteristics, scenarios, and simulation requirements, are integrated to generate an integrated dataset.

[0026] By introducing a Logistic-Sine-Cosine chaotic sequence to generate individual initial positions, the diversity and uniformity of population initialization are significantly improved, thus avoiding the algorithm from getting trapped in local optima. A logistic function maps continuous spatial position values ​​to discrete 0s or 1s, realizing the transformation from the search space to the feature subset encoding space. The Pearson correlation coefficient is used to measure the correlation between operational data and material consumption, and a penalty term for the number of features is introduced to achieve a feature selection criterion with strong correlation but low redundancy. By using grasshopper position updates and manta ray spiral foraging updates for grasshoppers with fitness above and below the median, respectively, the algorithm achieves... An integrated search strategy enhances an individual's ability to escape from the global space through Gaussian perturbation and further strengthens diversity with chaotic sequences, thereby reshaping individuals trapped in local optima. By focusing on the feature subsets corresponding to high-fitness individuals and selecting c optimal features, the strategy ensures that the selected features are highly correlated and representative of specific material needs. By fusing and standardizing operational features with external environmental data, a refined subset dataset for time series modeling is constructed, providing a foundation for personalized modeling of different materials. By introducing the Monte Carlo method to construct random samples based on predictions while maintaining volatility, and by using the 3σ principle to detect rationality, the strategy enhances the data's ability to simulate reality.

[0027] S2. Construct a TCN model to predict material demand, use the Grey Wolf optimization algorithm to optimize inventory costs and warehouse space utilization, and generate inventory adjustment instructions. Specifically, the data in the integrated dataset is normalized, and the scenario and simulated demand samples for each type of material are extracted and constructed into an input matrix according to time. Construct a TCN model, collect historical data for training, input the input matrix into the trained TCN model, perform N multi-scale predictions, generate a prediction distribution, calculate the mean and variance of the N prediction results, and perform prediction correction. The formula is:

[0028] in, This represents the corrected predicted demand for material m. To predict the mean value of material m, The confidence level is , which represents the standard normal distribution value (95% confidence interval). To predict the variance of material m, T is the time index; The corrected prediction results are smoothed using the exponential moving average (EMA) to reduce noise and obtain the final prediction results. Based on the final prediction results, a multi-objective function is constructed to optimize inventory costs and warehouse space utilization. The formula is: , , in, Total inventory cost For storage costs, This represents the total inventory of materials. Let m be the inventory level of material m. The unit storage cost of material m is obtained through hotel billing. For shortage costs, For the final prediction result, The unit shortage cost (e.g., a loss of 10 yuan per kg due to a beef shortage) is estimated based on historical data through regression. To maximize warehouse space utilization, Used storage space for supplies Total warehouse space; Set constraints: , in, and These represent the minimum and maximum limits of the inventory of material m, respectively. The MOGWO optimization algorithm is used to solve the multi-objective function, search the Pareto optimal solution set (containing multiple sets of inventory solutions), select the solution that meets the hotel's operational preferences (such as low cost bias: beef inventory 75kg, cost 500 yuan, space utilization rate 60%), and generate inventory adjustment instructions. The system collects the latest inventory consumption time series data in real time, calculates the consumption Z-score for each material, sets an anomaly threshold and compares it with the Z-score. If the Z-score is greater than the anomaly threshold, it is judged as an anomaly, a record is generated for the anomaly material, and the automatic replenishment of the relevant material is suspended until manual verification is performed. Otherwise, it is judged as normal and replenishment continues.

[0029] By constructing a TCN (Temporal Convolutional Network) model and training it with historical data, the system efficiently captures the evolutionary characteristics of material demand over time. By constructing a multi-objective optimization function (total inventory cost + warehouse space utilization) based on the prediction results, it achieves balanced decision-making among multiple operational objectives. The function quantifies economic and physical factors, including unit storage cost, shortage cost, and inventory level, realizing a closed-loop mapping from pure technical indicators to operating costs. By continuously collecting inventory consumption time series data and calculating Z-scores (standardized deviation index), it achieves dynamic monitoring of various material consumption behaviors, enhancing the system's self-correction and diagnostic capabilities.

[0030] S3. Construct a supply chain network model, extract features, use slime mold algorithm to jointly optimize supplier selection and logistics scheduling, and dynamically adjust the supply chain based on emergencies. Specifically, collect and preprocess supplier data; The supplier data includes material procurement prices, delivery times, credit scores (based on historical data), and location information; Construct a directed graph by using suppliers and hotels as nodes, logistics relationships between suppliers and hotels as edges, and delivery time and transportation cost as edge weights. To construct a GNN supply chain network model, historical supplier data and hotel data are collected, preprocessed, and then the model is trained. The trained GNN supply chain network model is output, and a directed graph is input for inference. From the embedding vectors of GNN nodes, supplier features (procurement cost, delivery time, and reliability (replaced by credit score) are extracted, standardized, and then a supplier scoring formula is defined: , in, The overall score for supplier g, To standardize procurement costs, To standardize delivery times, To standardize supplier reliability, , and These are the weights for procurement cost, delivery time, and supplier reliability, respectively, summed to one, with initial values ​​set by hotel preferences, all greater than 0; Define the total cost function, formula: , in, Let the total cost function be... For the number of suppliers, For suppliers Unit transportation cost to the hotel The distance from supplier g to the hotel. For transportation time, The time penalty coefficient is obtained by regression estimation based on historical data; Randomly generate slime mold populations and perform initialization processing, including initializing each individual slime mold population. This process consists of three parts: one part is supplier selection, which is encoded in binary, with 1 indicating supplier selection and 0 indicating no supplier selection. The second part is the logistics route. Based on the map, the K-shortest path algorithm is used to generate the initial route, the path length is limited, and feasible routes are randomly generated. The three parts are weighted according to procurement cost, delivery time, and supplier reliability; Define the constraints, formula: , in, Due to inventory constraints, Due to logistics capacity constraints, Due to time constraints, A binary variable representing whether to select supplier g. For supplier g's inventory level, For the hotel's supply needs, This is the maximum capacity of the transport vehicle. The maximum operating transport time; Configure exception handling options: If no supplier combination meets the inventory constraints, re-collect supplier data to generate slime mold populations; If the route is not feasible (e.g., exceeding vehicle capacity), randomly replace the supplier or shorten the route; The objective is to minimize the total cost function for slime mold foraging optimization, using the following formula: , in, For the first slime mold in the first The position of the next iteration (supplier selection + route + weight). For the first The optimal solution of the next iteration (lowest) ), and These are the global search factor and the local contraction factor, respectively. The global search factor decreases linearly from 1.0 to 0.0, and the local contraction factor is... Random values ​​within, As a weighting factor, The numbers are random and uniformly distributed. Within the range, For the first Ranking of fitness of slime molds ( (The lower the better) For the size of slime mold populations, and For two randomly selected slime mold sites, p is a random number that controls the search direction. The global search probability is set to a fixed value based on empirical methods. and Set the upper and lower bounds of the search space, respectively. The reason is to ensure that the algorithm achieves a balance between exploring new solutions (diversity) and developing current solutions (convergence). For partial development, For random exploration; Set a maximum number of iterations. Once the maximum number of iterations is reached, stop iterating and output the optimal solution for the current iteration.

[0031] By modeling suppliers and hotels as nodes in a graph and using delivery time and transportation cost as edge weights, a directed graph structure is constructed. This graph structure provides explicit topological input for the subsequent GNN, enabling the model to learn the influence relationships between nodes based on graph convolution. By integrating unit transportation cost, distance, time, and a time penalty coefficient estimated based on historical regression into a function, a comprehensive assessment of logistics costs is achieved. The regression estimation of the time penalty term makes the model closer to historical operational patterns. Three variables (the binary code of supplier selection, the logistics cost generated by the K shortest path, and the logistical cost) are initialized using a randomly generated slime mold population. By defining path and weight variables, the algorithm achieves diverse coverage of the solution space, avoiding getting trapped in local optima. By formulating constraints such as inventory, vehicle capacity, and transportation time, it enables the legality screening and practical feasibility verification of the solution space. By defining abnormal cases of insufficient supplier combinations or infeasible paths and setting regeneration or path adjustment mechanisms, it improves the robustness of the algorithm and enables adaptive control of the execution process. By constructing a slime mold foraging model that includes global search, local contraction, and fitness guidance, and combining iterative control and search space setting, it achieves the global minimization of the total cost function under multiple constraints.

[0032] Furthermore, collect real-time inventory data, supplier status, and logistics status, and set rules for unforeseen circumstances: Rule 1 for emergencies: If the demand for materials suddenly increases, reallocate the procurement volume and prioritize suppliers with shorter delivery times. Rule 2 for unforeseen circumstances: If the best supplier's inventory is insufficient to meet the purchase quantity (e.g., the inventory has quality problems), the second best supplier shall be selected. Rule 3 for emergencies: In the event of a shortage of logistics and transportation vehicles, priority should be given to materials with high demand, and other materials should be outsourced. Based on the adjustment of procurement quantities according to Emergency Response Rule 1 and Emergency Response Rule 2, the formula is: , in, This refers to the adjusted procurement quantity of material m. This represents the original purchase quantity of material m. Let m be the inventory level of supplier g for material m. For the sudden increase in material m, This is an indicator function that indicates whether the emergency situation rule is triggered. and These are respectively Emergency Rule 1 and Emergency Rule 2; Based on the rules for handling emergencies, the supply chain selection is dynamically adjusted, generating a dynamic adjustment report, including the reason for the adjustment (e.g., "sudden increase in beef demand, supplier C becomes ineffective"), the adjustment result (e.g., "supplier A purchases 50kg of beef, supplier B purchases 9.33kg of beef, logistics resources are allocated to 120kg"), and suggestions (e.g., "contact supplier D to make supplementary purchases").

[0033] By setting rules for unexpected situations, the system enables standardized identification and pre-defined response logic for abnormal events. Through formulas for adjusting procurement quantities based on rules one and two for unexpected situations, the system achieves refined adjustments to the procurement plan, taking into account supplier inventory, sudden increases in procurement volume, and rule triggering conditions. By generating dynamic adjustment reports, the system makes the adjustment process transparent, providing managers with a reference for decision-making.

[0034] S4. Collect real-time data to update the model, store the collected and analyzed hotel data, and build a visual interface to display the analyzed hotel data. Specifically, real-time consumption data for each type of material is collected, the real-time deviation between the real-time consumption data and the predicted value is calculated, and a deviation threshold is set using an empirical method. If the real-time deviation is greater than the deviation threshold, the TCN model is incrementally trained based on the newly collected real-time data.

[0035] By collecting real-time material consumption data and calculating the real-time deviation between the real-time material consumption data and the predicted value, a real-time feedback mechanism for model performance is realized, providing timely and clear indicators for model optimization. This avoids errors caused by relying on static model predictions for inventory optimization. By setting a deviation threshold using an empirical method, if the real-time deviation exceeds the deviation threshold, the TCN model will be incrementally trained based on newly collected real-time data, enabling the model to dynamically adapt to business changes and seasonal fluctuations, and continuously maintain the timeliness and accuracy of predictions.

[0036] Furthermore, the collected and analyzed data will be stored in a distributed manner, and a visualization interface will be built using D3.js to display the predictive performance chart, inventory optimization chart, and dynamic adjustment report. The prediction performance graph refers to a line graph of the real-time prediction bias of the TCN model. The inventory optimization chart displays real-time inventory utilization, material inventory, and inventory adjustment instructions.

[0037] By introducing dynamic prediction using the TCN model, an automatic adjustment mechanism based on error thresholds, and a distributed storage and visualization decision support system, the inventory management system has been transformed from "passive response" to "proactive prediction + self-evolution." Each step works in synergy, from data collection and analysis to model updates and user feedback, forming a closed-loop intelligent inventory optimization system that ultimately significantly improves the accuracy of inventory allocation, system response speed, and overall resource utilization efficiency.

[0038] This embodiment also provides an AI-based hotel management system, including: The data collection and optimization module is used to collect and preprocess hotel data, and uses an improved Chaotic Grasshopper optimization algorithm to filter hotel data with highly relevant features. A sampling module is constructed to introduce external data into hotel data with high correlation characteristics to build a random forest regression model, and the Monte Carlo sampling method is used to simulate hotel material demand samples in random demand scenarios. The forecast generation module is used to build a TCN model for material demand forecasting, use the Grey Wolf optimization algorithm to optimize inventory costs and warehouse space utilization, and generate inventory adjustment instructions. The extraction and adjustment module is used to build a supply chain network model, extract features, use slime mold algorithm to jointly optimize supplier selection and logistics scheduling, and dynamically adjust the supply chain based on emergencies. The update module is used to collect real-time data for model updates; The storage visualization module is used to store the collected and analyzed hotel data and to build a visual interface to display the analyzed hotel data.

[0039] This embodiment also provides a computer device applicable to AI-based hotel management methods, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the AI-based hotel management method proposed in the above embodiment.

[0040] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0041] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the AI-based hotel management method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0042] In summary, this invention improves system intelligence and decision-making efficiency, enhances resource allocation efficiency and inventory control, and strengthens supply chain resilience and dynamic scheduling capabilities by: collecting and preprocessing hotel data; using an improved Chaotic Grasshopper optimization algorithm to filter hotel data with high correlation features; constructing a Random Forest Regression Model to introduce external data into the hotel data with high correlation features; using the Monte Carlo sampling method to simulate hotel material demand samples under random demand scenarios; constructing a TCN model for material demand prediction; using the Gray Wolf optimization algorithm to optimize inventory costs and warehouse space utilization, and generating inventory adjustment instructions; constructing a supply chain network model for feature extraction; using the Slime Mold algorithm for joint optimization of supplier selection and logistics scheduling; dynamically adjusting the supply chain based on unforeseen circumstances; collecting real-time data for model updates; storing and analyzing the collected hotel data; and constructing a visual interface to display and analyze the hotel data.

[0043] 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, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An AI-based hotel management method, characterized in that: include, Hotel data was collected and preprocessed. An improved Chaotic Grasshopper optimization algorithm was used to filter hotel data with high correlation features. A random forest regression model was constructed to introduce external data into the hotel data with high correlation features. The Monte Carlo sampling method was used to simulate hotel material demand samples in random demand scenarios. Construct a TCN model to predict material demand, use the Grey Wolf optimization algorithm to optimize inventory costs and warehouse space utilization, and generate inventory adjustment instructions. Construct a supply chain network model, extract features, use slime mold algorithm to jointly optimize supplier selection and logistics scheduling, and dynamically adjust the supply chain based on unforeseen circumstances; Collect real-time data to update the model, store the collected and analyzed hotel data, and build a visual interface to display the analyzed hotel data.

2. The AI-based hotel management method as described in claim 1, characterized in that: The process of collecting and preprocessing hotel data, and simulating hotel material demand samples for stochastic demand scenarios using an improved chaotic grasshopper optimization algorithm, external data, and Monte Carlo sampling methods, includes: The hotel data includes hotel inventory, hotel operations, and external data; Each feature in the feature vector is binary encoded, the initial position of the grasshopper is generated using Logistic-Sine-Cosine chaotic mapping, the continuous positions of the grasshopper are converted into binary vectors using a logistic function, and the binary values ​​of all grasshoppers are used to generate a feature subset. Define hotel operation data and material consumption function, calculate fitness value, aim to maximize fitness value, use improved chaotic grasshopper optimization algorithm to update feature subset, calculate median of fitness value of all grasshoppers after each iteration, update grasshopper position for grasshoppers with fitness higher than median, otherwise update manta ray spiral foraging position. Add a Gaussian perturbation to the updated low-fitness grasshopper, and then use the Logistic-Sine-Cosine chaotic mapping to further update the low-fitness grasshopper with the added Gaussian perturbation; After reaching the maximum number of iterations, stop iterating and output c optimal feature subsets as the operational feature set, where c is the number of features of the hotel operational data with the highest fitness value corresponding to a single material consumption, set based on hotel requirements; For each type of material and its corresponding operational characteristics and external data, a subset dataset is generated, a random forest regression model is constructed, historical hotel data is collected and preprocessed as training data for the random forest regression model, and training is performed with the goal of minimizing the mean squared error. Based on external data to define scenario data, a trained random forest model is used to predict the incremental demand for materials in each scenario. The basic demand of the average material is adjusted, a Monte Carlo sampling method is used to generate random demand samples, and the 3σ principle is used to detect the coverage of the samples.

3. The AI-based hotel management method as described in claim 2, characterized in that: The construction of the TCN model for material demand forecasting, the use of the Grey Wolf optimization algorithm to optimize inventory costs and warehouse space utilization, and the generation of inventory adjustment instructions include: The data in the integrated dataset is normalized, and the scenario and simulated demand samples for each type of material are extracted and constructed into an input matrix according to time. Construct a TCN model, collect historical data for training, input the input matrix into the trained TCN model, perform N multi-scale predictions, generate a prediction distribution, calculate the mean and variance of the N prediction results, perform prediction correction, and use exponential moving average (EMA) smoothing to obtain the final prediction result. Construct a multi-objective function to optimize inventory costs and warehouse space utilization, set constraints, use the Grey Wolf MOGWO optimization algorithm to solve the multi-objective function, search for the Pareto optimal solution set, select the solution that meets the hotel's operational preferences, and generate inventory adjustment instructions. The system collects the latest inventory consumption time series data in real time, calculates the consumption Z-score for each material, sets an anomaly threshold and compares it with the Z-score. If the Z-score is greater than the anomaly threshold, it is judged as an anomaly, a record is generated for the anomaly material, and the automatic replenishment of the relevant material is suspended until manual verification is performed. Otherwise, it is judged as normal and replenishment continues.

4. The AI-based hotel management method as described in claim 3, characterized in that: The construction of the supply chain network model, feature extraction, and the use of the slime mold algorithm for joint optimization of supplier selection and logistics scheduling include: Collect and preprocess supplier data, and construct a directed graph with suppliers and hotels as nodes, logistics relationships from suppliers to hotels as edges, and delivery time and transportation cost as edge weights. To construct a GNN supply chain network model, historical supplier data and hotel data are collected, preprocessed, and then the model is trained. The trained GNN supply chain network model is output, and a directed graph is input for inference. From the embedding vectors of GNN nodes, supplier features are extracted, standardized, and then a supplier scoring formula and a total cost function are defined. Randomly generate slime mold populations, and initialize them, set constraints, and select anomaly handling options; The algorithm optimizes slime mold foraging by minimizing the total cost function. A maximum number of iterations is set, and the iteration stops after the maximum number of iterations is reached. The optimal solution for the current iteration is then output.

5. The AI-based hotel management method as described in claim 4, characterized in that: The dynamic adjustment of the supply chain based on unforeseen circumstances includes: Collect real-time inventory data, supplier status, and logistics status; set rules for contingencies; and adjust purchase quantities accordingly. The system dynamically adjusts supply chain choices based on contingency rules and generates dynamic adjustment reports.

6. The AI-based hotel management method as described in claim 5, characterized in that: The process of collecting real-time data for model updates includes: Collect real-time consumption data for each type of material, calculate the real-time deviation between the real-time consumption data and the predicted value, set a deviation threshold using an empirical method, and if the real-time deviation is greater than the deviation threshold, incrementally train the TCN model based on the newly collected real-time data.

7. The AI-based hotel management method as described in claim 6, characterized in that: The process of storing, collecting, and analyzing hotel data, and constructing a visual interface to display the analyzed hotel data, includes: The collected and analyzed data is stored in a distributed manner, and a visualization interface is built using D3.js to display the predictive performance chart, inventory optimization chart, and dynamic adjustment report.

8. An AI-based hotel management system, based on the AI-based hotel management method according to any one of claims 1 to 7, characterized in that: include, The data collection and optimization module is used to collect and preprocess hotel data, and uses an improved Chaotic Grasshopper optimization algorithm to filter hotel data with highly relevant features. A sampling module is constructed to introduce external data into hotel data with high correlation characteristics to build a random forest regression model, and the Monte Carlo sampling method is used to simulate hotel material demand samples in random demand scenarios. The forecast generation module is used to build a TCN model for material demand forecasting, use the Grey Wolf optimization algorithm to optimize inventory costs and warehouse space utilization, and generate inventory adjustment instructions. The extraction and adjustment module is used to build a supply chain network model, extract features, use slime mold algorithm to jointly optimize supplier selection and logistics scheduling, and dynamically adjust the supply chain based on emergencies. The update module is used to collect real-time data for model updates; The storage visualization module is used to store the collected and analyzed hotel data and to build a visual interface to display the analyzed hotel data.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the AI-based hotel management method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the AI-based hotel management method according to any one of claims 1 to 7.

Images