Hotel self-service check-in registration management system and method

Abstract

The invention discloses a hotel self-service check-in registration management system and method, and relates to the technical field of hotel management, and the method comprises the steps: carrying out the phase-space reconstruction of a room state change event sequence according to the state transition intensity; determining a room state fluctuation mode set according to the calculated room state fluctuation entropy; calculating a request feature entropy according to the check-in request sequence; calculating a real-time room distribution risk index according to the request feature entropy and the room state fluctuation mode set; when the room allocation risk index exceeds a preset risk threshold value, determining a local correction area of a preset room state prediction model; performing local correction on the model according to the local correction region, and updating the idle room candidate set; determining a room allocation decision scheme according to the idle room candidate set and the check-in request sequence; according to the method, accurate identification and dynamic regulation and control of risks between hotel room state fluctuation and check-in requests are effectively realized, and the accuracy and response efficiency of room allocation decision making are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of hotel management, and particularly relates to a hotel self-check-in registration management system and method. Background Technique

[0002] With the rapid development of smart hotels and the continuous improvement of the requirements for check-in experience, the management of hotel room allocation is gradually moving towards automation and refinement. However, the currently commonly used hotel automatic check-in management systems generally adopt relatively simple and static room status monitoring and room allocation decision-making methods. Such methods usually do not fully explore the changing rules of room status over time, nor effectively capture the potential complex interactions between check-in demand data and room status. This results in insufficient accuracy in predicting room status changes, making it difficult to adapt to the dynamic and variable check-in demands in actual operation, further causing inaccurate allocation of guest room resources and a decline in check-in experience.

[0003] Specifically, the current technical solutions generally only perform static analysis on the immediate situation or historical records of room status, lacking in-depth exploration of the internal rules of the frequency and duration of status changes, and being unable to accurately grasp the usage cycle and fluctuation pattern of guest rooms. In addition, the traditional methods for handling check-in demands also mostly stay at the level of simple sequence recording and statistical analysis, ignoring the complexity, randomness of check-in requests, and the highly non-linear relationship that may exist between them and room status changes, thus leading to the inability to accurately predict the potential risks between room status and check-in requests.

[0004] At the same time, the existing hotel room status prediction models usually adopt global unified prediction, lacking a local dynamic adjustment mechanism for the fluctuation characteristics of different regions or specific room status, and being difficult to effectively meet the real-time optimization requirements of the room status prediction model in local high-risk areas. In this case, if there are drastic changes or abnormal situations in the local room status, it may lead to problems such as reduced room allocation efficiency, increased decision-making risks, and decreased check-in service quality. Summary of the Invention

[0005] The purpose of the present invention is to provide a hotel self-check-in registration management system and method to solve the problems in the above background technique.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] In the first aspect, the present invention provides a hotel self-check-in registration management method, including:

[0008] S101: Real-time collect the sequence of status change events of each room in the hotel, and calculate the status transition intensity according to the status transition frequency and status duration of each room;

[0009] S102: Perform phase space reconstruction on the sequence of room state change events according to the state transition intensity; determine the set of room state fluctuation patterns based on the clustering analysis results of the calculated room state fluctuation entropy;

[0010] S103: Collect the sequence of check-in requests sent by the self-service terminal in real time, calculate the request feature entropy according to the data complexity of the check-in request sequence; calculate the real-time room allocation risk index according to the non-linear cross-mapping relationship between the request feature entropy and the set of room state fluctuation patterns;

[0011] S104: When the real-time room allocation risk index exceeds the preset risk threshold, determine the local correction area of the preset room state prediction model according to the spatial distribution and propagation path of the risk index; perform local correction on the preset room state prediction model according to the local correction area, and update the set of candidate idle rooms;

[0012] S105: Determine the room allocation decision plan according to the cross-scale correlation analysis between the set of candidate idle rooms and the check-in request sequence; update the state transition intensity of each room in real time according to the determined room allocation decision plan.

[0013] In a second aspect, the present invention provides a hotel self-check-in registration management system, which is implemented based on the above-mentioned hotel self-check-in registration management method, and includes:

[0014] A collection module, which is used to collect the sequence of state change events of each room in the hotel in real time, and calculate the state transition intensity according to the state transition frequency and state duration of each room;

[0015] A reconstruction module, which is used to perform phase space reconstruction on the sequence of room state change events according to the state transition intensity; determine the set of room state fluctuation patterns based on the clustering analysis results of the calculated room state fluctuation entropy;

[0016] A calculation module, which is used to collect the sequence of check-in requests sent by the self-service terminal in real time, calculate the request feature entropy according to the data complexity of the check-in request sequence; calculate the real-time room allocation risk index according to the non-linear cross-mapping relationship between the request feature entropy and the set of room state fluctuation patterns;

[0017] An update module, which is used to determine the local correction area of the preset room state prediction model according to the spatial distribution and propagation path of the risk index when the real-time room allocation risk index exceeds the preset risk threshold; perform local correction on the preset room state prediction model according to the local correction area, and update the set of candidate idle rooms;

[0018] An allocation module, which is used to determine the room allocation decision plan according to the cross-scale correlation analysis between the set of candidate idle rooms and the check-in request sequence; update the state transition intensity of each room in real time according to the determined room allocation decision plan.

[0019] In the above technical solution, the technical effects and advantages provided by the present invention are as follows:

[0020] By collecting and finely analyzing the state transition frequency and duration characteristics of each room in the hotel in real time, the present invention accurately constructs a key index of the room state transition intensity, thereby effectively excavating the periodic law and fluctuation characteristics of the room state change, providing accurate and reliable data support for subsequent room state prediction and decision-making analysis, and effectively solving the problem of insufficient excavation of the room state change law in the prior art.

[0021] Furthermore, by innovatively using the non-linear cross-mapping relationship between the room state fluctuation pattern and the feature entropy of the check-in request, a real-time room allocation risk index is constructed, which can timely and accurately identify potential risk conflict areas between the check-in request and the room state change, thereby realizing the effective quantification and dynamic monitoring of the room allocation risk, significantly improving the rationality and safety of the hotel check-in decision-making, and effectively overcoming the defect of poor check-in experience existing in the traditional static allocation method.

[0022] In addition, by introducing non-linear feature mining algorithms such as locality-sensitive hashing (LSH) and dynamic time warping (DTW), the local adaptive correction of the room state prediction model and the dynamic and accurate update of the candidate set of idle rooms are realized, enabling the room state prediction and check-in decision-making to always closely adapt to real-time local state fluctuations and abnormal events, effectively reducing the room state prediction error and improving the utilization efficiency of guest room resources, and comprehensively solving the prominent problem of slow response of the traditional global prediction model to local abnormal changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0024] Figure 1 It is a flowchart of a hotel self-service check-in registration management method of the present invention;

[0025] Figure 2 It is a framework diagram of a hotel self-service check-in registration management system of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art. The drawings are merely schematic illustrations of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus their repeated description will be omitted.

[0027] In addition, the described features, structures, or characteristics may be combined in any suitable manner in one or more example embodiments. In the following description, numerous specific details are provided to give a thorough understanding of the example embodiments of this disclosure. However, those skilled in the art will realize that one or more of the specific details may be omitted in practicing the technical solutions of this disclosure, or other methods, components, steps, etc. may be adopted. In other cases, well-known structures, methods, implementations, or operations are not shown or described in detail to avoid obscuring the aspects of this disclosure.

[0028] Example 1

[0029] As Figure 1 shown, this example embodiment discloses a hotel self-check-in registration management method, including:

[0030] S101: Real-time collect the sequence of status change events for each room in the hotel, and calculate the status transition intensity according to the status transition frequency and status duration of each room.

[0031] It should be noted that the sequence of status change events refers to a series of event data recorded in real time when the status of each room in the hotel changes, including: room number, room status type, and the exact time when the corresponding event occurs; specifically, each room status clearly includes but is not limited to the following types:

[0032] Idle status: The room has been cleaned and is ready for occupancy.

[0033] Occupied status: The guest has completed the check-in registration and the room is actually occupied.

[0034] To-be-cleaned status: Waiting for cleaning service after the guest checks out or changes rooms.

[0035] Maintenance status: The room is temporarily not open due to equipment failure and needs repair and maintenance.

[0036] Locked status: The front desk or the system temporarily locks the room and it cannot be allocated due to special requirements.

[0037] It should be understood that the accurate real-time acquisition of the sequence of state change events provides a precise data input basis for subsequent calculation of state transition intensity, thus effectively supporting subsequent complex decision-making analysis;

[0038] In implementation, calculating the state transition intensity according to the state transition frequency and state duration of each room includes:

[0039] Record the occurrence time of state transition events in each room in real time, and construct a state event time sequence matrix;

[0040] Specifically, the state event time sequence matrix is a two-dimensional matrix, where the horizontal axis is the room number, the vertical axis is each type of state transition that occurs in each room, and the elements in the matrix clearly record the accurate time of specific state transitions;

[0041] For example, taking hotel room 508 as an example, the recorded state event time sequence matrix is (as shown in Table 1 below):

[0042] Table 1: Data table of state event time sequence matrix

[0043]

[0044] Determine the duration sequence of each room state based on the state event time sequence matrix;

[0045] The duration sequence refers to a specific state (such as idle, occupied, to be cleaned, under repair or locked) in each hotel room. By recording the start time and end time of state events in real time, the duration of each occurrence of this state (i.e., the specific duration of each continuous existence of this state) is arranged in sequence to form a clear time sequence;

[0046] Perform Fourier transform on the duration sequence of each room state, and calculate the spectral distribution of the duration. The specific formula is:

[0047] ;

[0048] In the formula: is the duration of the nth time in the duration sequence of the room state; N is the number of data points of the corresponding room state duration; is the imaginary unit;

[0049] Determine the room state transition intensity according to the characteristic spectral peak of the spectral distribution and the state transition frequency;

[0050] Furthermore, determining the room state transition intensity according to the characteristic spectral peak of the spectral distribution and the state transition frequency includes:

[0051] Determine the dominant spectral peak according to the duration spectral distribution curve , the formula is as follows:

[0052] ;

[0053] Calculate the periodic intensity of each room state according to the frequency and amplitude corresponding to the dominant spectral peak , and the specific formula is:

[0054] ;

[0055] According to the periodic intensity and the state transition frequency , use weighted summation to determine the room state transition intensity , and the specific formula is as follows:

[0056] ;

[0057] In the formula: is the Fourier transform function, is the frequency parameter of the Fourier transform, and are the weighting coefficients and satisfy , which is determined specifically according to experimental data. For example, ; among them, the state transition frequency ; for example, taking 1 day as the statistical period, room 508 has 5 state transitions within 1 day, then the frequency ;

[0058] It should be understood that the calculated state transition intensity can effectively quantify the room state change law, and provide a quantitative index for subsequent accurate prediction of the room state fluctuation mode and real-time room state optimization decision-making.

[0059] S102: Perform phase space reconstruction on the room state change event sequence according to the state transition intensity; determine the set of room state fluctuation modes according to the clustering analysis results of the calculated room state fluctuation entropy;

[0060] It should be noted that the phase space reconstruction refers to using the one-dimensional time series composed of the room state transition intensity and converting it into the state trajectory of a high-dimensional space through non-linear dynamics methods to facilitate in-depth analysis of the room state fluctuation law;

[0061] Specifically, the room state change event sequence clearly comes from the state transition intensity sequence constructed in step S101, and each data point of the state transition intensity sequence specifically and clearly corresponds to the state transition intensity value of each hotel room within continuous time;

[0062] Exemplarily, the state transition intensity sequence obtained by room 508 on a certain day is:

[0063] ;

[0064] It should be understood that through phase space reconstruction, the internal non-linear structure of the room state fluctuations can be effectively revealed, providing clear trajectory feature data for subsequent room state pattern clustering analysis;

[0065] In implementation, the phase space reconstruction of the room state change event sequence according to the state transition intensity includes:

[0066] Calculate the autocorrelation function of the room state transition intensity, and determine the optimal delay time according to the first drop to 1 / e, , where e is a mathematical constant, and the specific formula is as follows:

[0067] ;

[0068] In the formula: is the t-th explicit data point in the room state transition intensity sequence, is the mean value of the sequence, and N is the length of the sequence;

[0069] Analyze the state transition intensity sequence by the Cao method to determine the optimal embedding dimension ;

[0070] Specifically, the Cao method defines an auxiliary quantity 𝐸(𝑚) to evaluate the embedding dimension, and the formula is:

[0071] ;

[0072] In the formula: is the vector constructed from the state transition intensity sequence with the embedding dimension m, is the vector closest to , represents the Euclidean distance;

[0073] Based on the optimal delay time and the optimal embedding dimension perform phase space reconstruction to obtain the room state phase space reconstruction trajectory. The specific reconstruction formula is as follows:

[0074] ;

[0075] Each vector specifically represents a state point in the phase space, and all state points form the phase space reconstruction trajectory;

[0076] For example, continuing the above example, the reconstructed trajectory vector specifically obtained for room 508 is:

[0077] ;

[0078] ;

[0079] And so on, forming a clear trajectory sequence;

[0080] In implementation, determining the set of room state fluctuation patterns according to the clustering analysis result of the calculated room state fluctuation entropy includes:

[0081] Calculating the fluctuation entropy of the phase space reconstruction trajectory of each room state, and the specific formula is:

[0082] ;

[0083] wherein, the phase space trajectory is equally divided into M grid regions, is the probability that the trajectory point falls into the i-th grid;

[0084] For example, the trajectory of room 508 is clearly divided into 10 grid regions, and the proportion of trajectory points in each grid is clearly 0.1, 0.2…, 0.05 respectively, and its fluctuation entropy is calculated;

[0085] Using the fluctuation entropy to calculate the entropy density distribution of the room state;

[0086] It should be noted that: the entropy density distribution adopts the Kernel Density Estimation (KDE):

[0087] ;

[0088] In the formula: is the clear fluctuation entropy of the i-th room, 𝐾 is the Gaussian kernel function, and h is the bandwidth parameter, which is determined according to empirical data;

[0089] Automatically identifying the position of the clustering center according to the entropy density distribution, and determining the set of room state fluctuation patterns;

[0090] It should be noted that: automatically identifying the clustering center of the room state fluctuation pattern according to the entropy density distribution, adopting the DBSCAN clustering algorithm; each clustering center in the clustering result corresponds to a clear room state fluctuation pattern;

[0091] Exemplarily, the following set of fluctuation patterns is identified through clustering:

[0092] Pattern A: High room state stability and low fluctuation entropy;

[0093] Pattern B: Severe room state fluctuations and high entropy value;

[0094] Pattern C: Strong regularity of room state fluctuations and medium entropy value.

[0095] S103: collecting a check-in request sequence sent by a self-service terminal in real time, calculating a request feature entropy according to the data complexity of the check-in request sequence; calculating a real-time room allocation risk index according to a nonlinear cross-mapping relationship between the request feature entropy and a room status fluctuation pattern set;

[0096] It should be noted that the check-in request sequence refers to the check-in request event sequence issued by the user recorded in real time by the self-service terminal, specifically including data fields such as request time, requested room type, check-in duration, and customer preferences; the request sequence is transmitted to the backend server in real time in a preset data format to ensure the real-time and integrity of subsequent data processing;

[0097] For example, a clear check-in request sequence is as follows:

[0098] Request time: 2024-06-01 08:15:23, Room type: Single room, Stay duration: 2 nights, Customer preference: Window side;

[0099] Request time: 2024-06-01 08:15:35, Room type: Double room, Length of stay: 1 night, Customer preference: Quiet location;

[0100] Request time: 2024-06-01 08:15:47, Room type: Suite, Stay duration: 3 nights, Customer preference: High floor;

[0101] It should be understood that calculating the real-time room allocation risk index through the nonlinear cross-mapping relationship between the request feature entropy and the room status fluctuation pattern helps to accurately assess the potential conflict risk between the current check-in request and the room status fluctuation;

[0102] Specifically, the complexity of the check-in request sequence is calculated using the permutation entropy, and the formula is as follows:

[0103] ;

[0104] Where: represents the embedding dimension of permutation entropy, represents the jth arrangement mode, shows the relative frequency of the jth arrangement pattern in the request sequence;

[0105] For example, if in the most recent 10 requests at the current moment, the frequencies of occurrence of the arrangement pattern of room types are 0.4, 0.3, 0.2, and 0.1, respectively, then the calculated arrangement entropy is specifically:

[0106] ;

[0107] In implementation, calculating the real-time room allocation risk index according to the non-linear cross mapping relationship between the request feature entropy and the room state fluctuation mode set includes:

[0108] Performing phase space embedding on the room state fluctuation mode set and the request feature entropy respectively, and the specific formula is:

[0109] The embedding of the request feature entropy sequence is:

[0110] ;

[0111] The embedding of the room state fluctuation mode set is:

[0112] ;

[0113] In the formula: is the request feature entropy at the t-th moment, is the room state fluctuation mode identifier at the t-th moment;

[0114] Exemplarily, for example, the embedding vector at a certain moment:

[0115] The request feature entropy vector may be [1.24, 1.35, 1.28, 1.45];

[0116] The room state fluctuation mode set vector may be [Mode A, Mode C, Mode B, Mode A], and is encoded as a numerical value using the corresponding identifier;

[0117] Using the Convergent Cross Mapping algorithm (CCM) to calculate the cross-prediction error of the room state fluctuation mode on the request feature entropy;

[0118] Among them, the use of the Convergent Cross Mapping algorithm (CCM) to calculate the cross-prediction error includes:

[0119] Predicting the request feature entropy sequence based on the reconstructed sequence of the room state fluctuation mode embedded in the phase space;

[0120] It should be noted that: predicting the request feature entropy sequence based on the reconstructed sequence of the room state fluctuation mode embedded in the phase space, specifically using the k-nearest neighbor reconstruction method, to find the k nearest neighbors of the embedding vector is, for , and the prediction vector is the weighted average of the corresponding of these nearest neighbors;

[0121] For example, when k = 3, the prediction vector is:

[0122] ;

[0123] In the formula: is the reciprocal of the distance;

[0124] Calculate the non - linear prediction error of the reconstructed sequence for the request feature entropy sequence. The specific formula is as follows:

[0125] ;

[0126] In the formula: N is the sequence length, is the Euclidean distance;

[0127] Determine the stable value of the cross - prediction error through the convergence characteristics of the prediction error with respect to the reconstruction dimension;

[0128] It can be understood that: by gradually increasing the embedding dimension m, observe the cross - prediction error whether it converges to a definite stable value; define the error value when it first reaches the stable state as the definite stable value of the cross - prediction error;

[0129] Determine the real - time room allocation risk index according to the cross - prediction error. The specific formula is:

[0130] ;

[0131] In the formula: is the slope parameter, controlling the sensitivity of the exponent to the error, is the central threshold parameter, obtained through pre - data analysis;

[0132] It should be understood that through the above - described detailed and clear calculation of the request feature entropy and determination of the cross - mapping risk index, the accurate assessment of the risk between room state fluctuations and request demands is effectively realized, providing a clear and reliable data basis for real - time room state regulation.

[0133] S104: When the real - time room allocation risk index exceeds the preset risk threshold, determine the local correction area of the preset room state prediction model according to the spatial distribution and propagation path of the risk index; perform local correction on the preset room state prediction model according to the local correction area, and update the set of candidate idle rooms;

[0134] It should be noted that the meaning of the real - time room allocation risk index exceeding the preset risk threshold is: the real - time risk index (with a value range of 0 to 1) calculated through step S103 is greater than the preset risk threshold (for example, 0.7), indicating that there is a relatively high risk in the current room allocation of the hotel, and it is necessary to perform local correction on the preset room state prediction model;

[0135] Exemplarily, if the calculated value of the real - time risk index at a certain moment is clearly 0.78, exceeding the set risk threshold of 0.7, the local correction mechanism is triggered at this time;

[0136] It should be understood that this step aims to determine the local areas that need to be corrected based on the spatial propagation characteristics of the risk index, so as to effectively and specifically optimize the preset room status prediction model, and further accurately update the room candidate set in real time to avoid room allocation conflicts;

[0137] It should be noted that the preset room status prediction model is a non-linear time series prediction model based on the Long Short-Term Memory network (LSTM), which is mainly used to make short-term predictions on the future state changes of each room in the hotel, so as to assist the real-time room allocation decision-making process;

[0138] Specifically, the input data of the preset room status prediction model includes but is not limited to the following features: the historical room status sequence (such as idle, occupied, to be cleaned, under maintenance, locked), the room status transition intensity sequence, the pattern categories determined in the room status fluctuation pattern set (Pattern A, Pattern B, Pattern C, etc.), the historical data of the real-time room allocation risk index; the output of the preset room status prediction model is: the state probability distribution of each room in a future specific time window (for example, the next 1 hour or the next 3 hours), including the specific probability values of each state (idle, occupied, to be cleaned, under maintenance, locked);

[0139] Exemplarily, for room 508, the preset room status prediction model inputs the historical status sequence (idle → occupied → to be cleaned → idle), the status transition intensity sequence (such as 2.014, 1.856…), the current pattern type (such as Pattern B), and the real-time risk index (such as 0.78). After prediction by the LSTM network, the output results are specifically: within the next 1 hour: the probability of being idle is 0.75, the probability of being occupied is 0.15, the probability of being under maintenance is 0.05, and the probability of being to be cleaned is 0.05; within the next 3 hours: the probability of being idle is 0.50, the probability of being occupied is 0.35, the probability of being under maintenance is 0.10, and the probability of being to be cleaned is 0.05;

[0140] Furthermore, the preset room status prediction model specifically adopts the following LSTM network structure: the input layer is clearly set to 4 nodes (corresponding to the above 4 input features); the LSTM hidden layer uses 2 stacked layers, each layer containing 64 hidden units; the output layer is a fully connected layer, and 5 nodes are output corresponding to the predicted probabilities of each room state; the activation function is clearly the softmax function to ensure that the sum of the output probability values is clearly 1;

[0141] It should be understood that the above-mentioned clear LSTM network obtains the weight parameters through training with historical data, and uses the error correction data in the local correction area to clearly update the network parameters during the real-time application process to achieve real-time improvement of the model prediction performance;

[0142] In implementation, determining the local correction area of the preset room status prediction model according to the spatial distribution and propagation path of the risk index includes:

[0143] Construct a risk index space network based on the real-time room allocation risk index;

[0144] It should be noted that: the risk index space network is a weighted network with each room in the hotel as a node and the spatial proximity or functional relevance (such as on the same floor, in the same area, or of the same room type) between rooms as edges;

[0145] Specifically, the nodes clearly represent rooms, such as 101, 102, 103, etc., and the edges clearly represent the spatial or functional association between two rooms. For example, rooms 101 and 102 are adjacent to form an edge; the vertical adjacency between room 101 and the upstairs room 201 also clearly forms an edge.

[0146] The weight of each node is specifically the current real-time risk index of this room, ranging from 0 to 1;

[0147] Exemplarily, an example of a local network node and edge is as follows:

[0148] Node 101 (risk index 0.8), Node 102 (risk index 0.6), Node 103 (risk index 0.9);

[0149] The edges are: (101,102), (101,103), (102,103), indicating the proximity or functional association between rooms;

[0150] Calculate the degree centrality and betweenness centrality of the nodes in the risk index space network;

[0151] Among them, the degree centrality calculation formula is:

[0152] ;

[0153] In the formula: is the number of nodes connected to node i, and N is the total number of network nodes;

[0154] Among them, the betweenness centrality calculation formula is:

[0155] ;

[0156] In the formula: represents the number of shortest paths between node j and node k, represents the number of these shortest paths passing through node i;

[0157] Exemplarily, if node 101 is clearly adjacent to 4 rooms and there are 10 rooms in total in the network, then the degree centrality of node 101 is clearly ;

[0158] Determine the local correction area of the preset room status prediction model according to the nodes whose centrality exceeds the preset threshold;

[0159] It should be understood that: according to the nodes whose centrality exceeds the preset threshold (for example, the nodes ranked in the top 20% of degree centrality or betweenness centrality) and their adjacent areas, as the local correction area of the preset room status prediction model;

[0160] Exemplarily, if the betweenness centrality of Room 101 is clearly the highest, and its adjacent rooms include 102 and 103, then 101, 102, and 103 are determined as the local correction area of the preset room status prediction model;

[0161] Specifically, the local correction of the preset room status prediction model according to the local correction area and the update of the candidate set of vacant rooms include:

[0162] Extract the historical prediction error data of the room status fluctuations in the local correction area;

[0163] Exemplarily, if the prediction of Room 101 at a certain past moment is "vacant", but the actual situation is "occupied", then the prediction error at that moment is recorded as "prediction - actual = vacant - occupied";

[0164] Adopt the Locality-Sensitive Hashing (LSH) algorithm to retrieve the historical error feature patterns similar to the current prediction error data;

[0165] Specifically, vectorize the error data to form a clear feature vector; through the LSH algorithm, quickly and efficiently find the historical error data feature vector with a relatively close Euclidean distance to the current error pattern;

[0166] Exemplarily, if the current error feature vector is [0, 1, -1], and the historical vector [0, 1, -1.1] is retrieved, then it is clearly a similar error pattern;

[0167] Perform time series alignment on the retrieved historical error feature patterns using the Dynamic Time Warping (DTW) algorithm to determine the optimal error correction sequence;

[0168] Specifically, the DTW alignment distance is calculated as:

[0169] ;

[0170] In the formula: are the current and historical error feature vectors, and K is the length of the clearly aligned sequence;

[0171] Based on the determined optimal error correction sequence, perform real-time correction on the local parameters of the preset room status prediction model;

[0172] Specifically, model parameters (such as weights and biases) are locally updated through a gradient correction method (e.g., least squares optimization); the model prediction accuracy is improved within a local area;

[0173] Recalculate the probability distribution of available rooms in the local area according to the corrected preset room status prediction model, and then update the candidate set of available rooms;

[0174] It should be understood that: after calculating the probability distribution of available rooms, rooms with an available probability exceeding a preset first probability threshold (e.g., 0.9) are re-added to the candidate set of available rooms; rooms with a probability lower than a preset second probability threshold (e.g., 0.2) are excluded to avoid misallocation;

[0175] Exemplarily, if the available probability of room 101 is clearly increased to 0.95 after correction, then room 101 clearly enters the updated candidate set of available rooms;

[0176] S105: Determine a room allocation decision plan based on the cross-scale correlation analysis of the candidate set of available rooms and the check-in request sequence; update the state transition intensity of each room in real time according to the determined room allocation decision plan;

[0177] In implementation, the determining a room allocation decision plan based on the cross-scale correlation analysis of the candidate set of available rooms and the check-in request sequence includes:

[0178] Perform multi-scale coarse-graining processing on the candidate set sequence of available rooms and the check-in request sequence;

[0179] It should be noted that: performing multi-scale coarse-graining processing on the candidate set sequence of available rooms and the check-in request sequence means: defining the initial scale as the original data scale, for example, a time granularity of 1 minute or 5 minutes; gradually and clearly increasing the scale to multi-scales such as 10 minutes, 15 minutes, 30 minutes, etc., and performing coarse-graining step by step; within each scale, clearly average or accumulate the sequence data to achieve scale transformation;

[0180] Exemplarily, the initial scale is 5 minutes. A candidate set sequence of a certain room (the available status of room 101) may be: [available, checked in, available, available, checked in]. After coarse-graining to a 15-minute scale, calculate the proportion of available rooms within 15 minutes (such as 2 / 3, clearly indicating 66.7%);

[0181] Calculate the cross-scale correlation between the two sequences at each scale;

[0182] It should be noted that: cross-scale correlation is a clear indicator used to describe the time correlation intensity between the candidate sequence of available rooms and the check-in request sequence at different scales;

[0183] Specifically, the multiscale mutual information (MMI) method is adopted to calculate the cross-scale correlation:

[0184] ;

[0185] In the formula: respectively represent the candidate sequence of vacant rooms and the occupancy request sequence after coarse-graining processing, represents the sequence at scale s the probability of joint occurrence, , are the probabilities of individual events respectively;

[0186] Exemplarily, at a 15-minute scale, the MMI value of the cross-scale correlation between the vacant room sequence and the occupancy request sequence may be clearly calculated as 0.85;

[0187] Determine the optimal scale and the room allocation decision plan at the corresponding scale according to the cross-scale correlation;

[0188] Specifically, select the scale at which the cross-scale correlation MMI value reaches the maximum or is close to the maximum as the optimal scale;

[0189] At the optimal scale, based on the time alignment principle, clearly match the candidate set of vacant rooms to the occupancy requests within the corresponding time period; form a room allocation decision plan, specifically a list of the corresponding relationship between room numbers and requests;

[0190] Exemplarily, if the calculation result shows that the MMI value at the 15-minute scale is the largest at 0.85, then determine the 15-minute scale as the optimal scale and form a decision plan:

[0191] For example: Room 101 is clearly allocated to the occupancy request between 15:30 and 15:45 (single room, staying for 1 day); Room 103 is clearly allocated to another request during the same period (double room, staying for 2 days);

[0192] It should also be noted that: after executing the room allocation decision, the status sequence of the corresponding room is updated in real time. For example, the status of Room 101 clearly changes from "vacant" to "booked"; the state transition frequency and state duration are recalculated in real time according to the updated status sequence; the state transition intensity of each room is updated in real time using the calculation formula in step S101;

[0193] Exemplarily, the status change sequence of Room 101 clearly adds a new transition (vacant → booked), and the state transition intensity of this room is recalculated and clearly updated in real time. For example, the updated state transition intensity is 0.75.

[0194] Embodiment 2

[0195] As Figure 2 shown, parts not detailed in this embodiment are as shown in Embodiment 1. This embodiment discloses a hotel self-check-in registration management system, including:

[0196] An acquisition module 201, configured to collect in real time the state change event sequence of each room in the hotel, and calculate the state transition intensity according to the state transition frequency and state duration of each room;

[0197] A reconstruction module 202, configured to perform phase space reconstruction on the room state change event sequence according to the state transition intensity; determine the set of room state fluctuation patterns according to the clustering analysis result of the calculated room state fluctuation entropy;

[0198] A calculation module 203, configured to collect in real time the check-in request sequence sent by the self-service terminal, calculate the request feature entropy according to the data complexity of the check-in request sequence; calculate the real-time room allocation risk index according to the non-linear cross-mapping relationship between the request feature entropy and the set of room state fluctuation patterns;

[0199] An update module 204, configured to, when the real-time room allocation risk index exceeds the preset risk threshold, determine the local correction area of the preset room state prediction model according to the spatial distribution and propagation path of the risk index; perform local correction on the preset room state prediction model according to the local correction area, and update the set of candidate idle rooms;

[0200] An allocation module 205, configured to determine the room allocation decision plan according to the cross-scale correlation analysis between the set of candidate idle rooms and the check-in request sequence; update the state transition intensity of each room in real time according to the determined room allocation decision plan.

[0201] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula closest to the real situation. The selection of preset parameters, weights, and thresholds in the formulas is set by those skilled in the art according to the actual situation.

[0202] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired network or a wireless network. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that contains one or more collections of available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs) or semiconductor media. The semiconductor media can be a solid-state drive.

[0203] Only some exemplary embodiments of the present invention have been described above by way of illustration. Without doubt, for those of ordinary skill in the art, the described embodiments can be modified in various different ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of the claims of the present invention.

Claims

1. A hotel self-check-in registration management method, characterized in that, Including: S101: Real-time collect the sequence of status change events for each room in the hotel, and calculate the state transition intensity according to the state transition frequency and state duration of each room; S102: Perform phase space reconstruction on the sequence of room status change events according to the state transition intensity; Determine the set of room status fluctuation patterns according to the clustering analysis results of the calculated room status fluctuation entropy; S103: Real-time collect the sequence of check-in requests sent by the self-service terminal, calculate the request feature entropy according to the data complexity of the check-in request sequence; calculate the real-time room allocation risk index according to the non-linear cross-mapping relationship between the request feature entropy and the set of room status fluctuation patterns; S104: When the real-time room allocation risk index exceeds the preset risk threshold, determine the local correction area of the preset room status prediction model according to the spatial distribution and propagation path of the risk index; perform local correction on the preset room status prediction model according to the local correction area, and update the set of candidate idle rooms; S105: Determine the room allocation decision plan according to the cross-scale correlation analysis between the set of candidate idle rooms and the check-in request sequence; update the state transition intensity of each room in real time according to the determined room allocation decision plan.

2. The hotel self-check-in registration management method according to claim 1, characterized in that, The calculating the state transition intensity according to the state transition frequency and state duration of each room includes: Real-time record the occurrence time of each room status transition event, and construct a state event time series matrix; Determine the duration sequence of each room status based on the state event time series matrix; Perform Fourier transform on the duration sequence of each room status, and calculate the spectral distribution of the duration; Determine the room status transition intensity according to the characteristic spectral peak of the spectral distribution and the state transition frequency.

3. The hotel self-check-in registration management method according to claim 2, characterized in that The determining the room status transition intensity according to the characteristic spectral peak of the spectral distribution and the state transition frequency includes: Determine the dominant spectral peak according to the duration spectrum distribution curve , the formula is as follows: ； Calculate the periodic intensity of each room state according to the frequency and amplitude corresponding to the dominant spectral peak , and the specific formula is as follows: ； According to the periodic intensity and the state transition frequency , the weighted sum is used to determine the room state transition intensity , and the specific formula is as follows: ； In the formula: is the Fourier transform function, is the frequency parameter of the Fourier transform, and are the weighting coefficients and satisfy .

4. The hotel self-check-in registration management method according to claim 3, wherein, The performing phase space reconstruction on the sequence of room status change events according to the state transition intensity includes: Calculate the autocorrelation function of the room state transition intensity and determine the optimal delay time according to the first drop to 1 / e, where e is the mathematical constant , and e is the mathematical constant Determining the Optimal Embedding Dimension by Analyzing the State Transition Intensity Sequence with the Cao Method ; Based on the optimal delay time and the optimal embedding dimension perform phase space reconstruction to obtain the phase space reconstruction trajectory of the room state.

5. The hotel self-check-in registration management method according to claim 4, characterized in that The determining the set of room status fluctuation patterns according to the clustering analysis results of the calculated room status fluctuation entropy includes: Calculate the fluctuation entropy of the phase space reconstruction trajectory of each room status; Calculate the entropy density distribution of the room status using the fluctuation entropy; Automatically identify the position of the clustering center according to the entropy density distribution, and determine the set of room status fluctuation patterns.

6. The hotel self-check-in registration management method according to claim 5, characterized in that The calculating the real-time room allocation risk index according to the non-linear cross-mapping relationship between the request feature entropy and the set of room status fluctuation patterns includes: Perform phase space embedding on the set of room status fluctuation patterns and the request feature entropy respectively, and the specific formulas are: The embedding of the request feature entropy sequence is: ； The embedding of the set of room status fluctuation patterns is: ； Wherein: is the request feature entropy at the t-th moment, is the room state fluctuation mode identifier at the t-th moment; Use the convergent cross-mapping algorithm to calculate the cross-prediction error of the set of room status fluctuation patterns for the request feature entropy; Among them, the calculating the cross-prediction error using the convergent cross-mapping algorithm includes: Predict the request feature entropy sequence based on the phase space embedded reconstructed sequence of the set of room status fluctuation patterns; Calculate the non-linear prediction error of the reconstructed sequence for the request feature entropy sequence; Determine the stable value of the cross-prediction error through the convergence characteristics of the prediction error with respect to the reconstruction dimension; Determine the real-time room allocation risk index according to the cross-prediction error.

7. The hotel self-check-in registration management method according to claim 6, characterized in that The determining the local correction area of the preset room status prediction model according to the spatial distribution and propagation path of the risk index includes: Construct a risk index space network based on the real-time room allocation risk index; Calculate the degree centrality and betweenness centrality of the nodes in the risk index space network; Determine the local correction area of the preset room state prediction model according to the nodes whose centrality exceeds the preset threshold.

8. The hotel self-check-in registration management method according to claim 7, characterized in that, The local correction of the preset room state prediction model according to the local correction area and the update of the idle room candidate set include: Extract the historical prediction error data of the room state fluctuation in the local correction area; Use the locally sensitive hashing algorithm to retrieve the historical error feature patterns similar to the current prediction error data; Perform time series alignment on the retrieved historical error feature patterns using the dynamic time warping algorithm to determine the optimal error correction sequence; Perform real-time correction on the local parameters of the preset room state prediction model based on the determined optimal error correction sequence; Recalculate the probability distribution of idle rooms in the local area according to the corrected preset room state prediction model, and then update the idle room candidate set.

9. The hotel self-check-in registration management method according to claim 8, characterized in that, The determination of the room allocation decision plan according to the cross-scale correlation analysis between the idle room candidate set and the check-in request sequence includes: Perform multi-scale coarse-graining processing on the idle room candidate set sequence and the check-in request sequence; Calculate the cross-scale correlation between the two sequences at each scale; Determine the optimal scale and the room allocation decision plan at the corresponding scale according to the cross-scale correlation.

10. A hotel self-check-in registration management system, which is implemented based on the hotel self-check-in registration management method described in any one of claims 1-9, characterized in that, Include: The acquisition module is used to collect the sequence of state change events of each room in the hotel in real time, and calculate the state transition intensity according to the state transition frequency and state duration of each room; The reconstruction module is used to perform phase space reconstruction on the sequence of room state change events according to the state transition intensity; Determine the set of room state fluctuation patterns according to the clustering analysis results of the calculated room state fluctuation entropy; The calculation module is used to collect the check-in request sequence sent by the self-service terminal in real time, and calculate the request feature entropy according to the data complexity of the check-in request sequence; Calculate the real-time room allocation risk index according to the non-linear cross-mapping relationship between the request feature entropy and the set of room state fluctuation patterns; The update module is used to, when the real-time room allocation risk index exceeds the preset risk threshold, determine the local correction area of the preset room state prediction model according to the spatial distribution and propagation path of the risk index; perform local correction on the preset room state prediction model according to the local correction area, and update the idle room candidate set; The allocation module is used to determine the room allocation decision plan according to the cross-scale correlation analysis between the idle room candidate set and the check-in request sequence; update the state transition intensity of each room in real time according to the determined room allocation decision plan.

Images