Search

CN-121981356-A - Intelligent museum passenger flow scheduling method and system based on time sequence prediction

CN121981356ACN 121981356 ACN121981356 ACN 121981356ACN-121981356-A

Abstract

The invention relates to the technical field of intelligent scheduling, and particularly discloses a museum passenger flow intelligent scheduling method and system based on time sequence prediction, wherein the method comprises the steps of constructing a multi-element time sequence data sequence; the method comprises the steps of carrying out trend decomposition and periodic feature extraction on a multi-element time sequence data sequence, generating a feature vector cluster, carrying out multi-step cross-period rolling fusion prediction to obtain a passenger flow predicted value, constructing a passenger flow space-time distribution prediction tensor, taking an exhibition area with the passenger flow predicted value exceeding the capacity of a preset exhibition area as a potential congestion pressure source node, carrying out network diffusion deduction on the potential congestion pressure source node to obtain a dynamic influence deduction map of the exhibition area, carrying out strategy search in a multi-dimensional strategy space of the exhibition area based on the dynamic influence deduction map to generate a target scheduling strategy set, carrying out time sequence arrangement on the target scheduling strategy set, converting the target scheduling strategy set into an executable instruction sequence, and issuing the executable instruction sequence.

Inventors

  • ZHANG LI
  • Jia can
  • LI YONGQIANG

Assignees

  • 西安维真视界文化科技股份有限公司

Dates

Publication Date
20260505
Application Date
20260323

Claims (10)

  1. 1. A method for intelligent scheduling of museum passenger flow based on time sequence prediction, the method comprising: C1, collecting basic passenger flow data and associated environmental factor data of an exhibition area in a museum to construct a multi-element time sequence data sequence; c2, carrying out trend decomposition and periodic feature extraction on the multi-element time sequence data to generate a feature vector cluster of the exhibition area; C3, performing multi-step cross-period rolling fusion prediction on the feature vector cluster to obtain a passenger flow predicted value, and constructing a passenger flow space-time distribution prediction tensor of the exhibition area according to the passenger flow predicted value; C4, taking an exhibition area with the passenger flow predicted value exceeding the capacity of a preset exhibition area as a potential congestion pressure source node according to the passenger flow space-time distribution prediction tensor, and carrying out network diffusion deduction on the potential congestion pressure source node based on a preset topological connection relation to obtain a dynamic influence deduction graph of the exhibition area; c5, performing strategy search in a multidimensional strategy space of the exhibition area based on the dynamic influence deduction graph to generate a targeted scheduling strategy set, wherein the multidimensional strategy space comprises a path adjustment instruction, an entrance management instruction and a service resource instruction; And C6, carrying out time sequence arrangement on the target scheduling strategy set, converting the target scheduling strategy set into an executable instruction sequence, and issuing the executable instruction sequence.
  2. 2. The intelligent scheduling method of museum passenger flow based on time sequence prediction as set forth in claim 1, wherein the performing trend decomposition and periodic feature extraction on the multi-element time sequence data sequence to generate the feature vector cluster of the exhibition area includes: trend decomposition is carried out on the multi-element time sequence data sequence, and a trend component, a seasonal component and a residual component are separated; performing fast Fourier transform extraction on the seasonal components to obtain main periodic frequency and periodic intensity, and taking the main periodic frequency and the periodic intensity as first periodic characteristics; carrying out mean analysis and variance analysis on the first-order differential sequence of the trend component, and taking a mean analysis result and a variance analysis result as trend change characteristics; And splicing and integrating the first periodic characteristic, the trend change characteristic and the noise statistical characteristic extracted from the residual component into a characteristic vector cluster of the exhibition area.
  3. 3. The intelligent scheduling method of museum passenger flow based on time sequence prediction as set forth in claim 1, wherein the performing multi-step cross-period rolling fusion prediction on the feature vector cluster to obtain a passenger flow predicted value, and constructing a passenger flow space-time distribution predicted tensor of the exhibition area according to the passenger flow predicted value includes: inputting the characteristic vector cluster into a pre-trained time convolution network to extract the time dependence of the exhibition area; Splicing the time dependency relationship with the attribute features in the associated environmental factor data to form fusion features; inputting the fusion features into a long-short-period memory network of the feature vector cluster, and selectively memorizing and forgetting time sequence information of the fusion features according to an update gate and a reset gate of the long-short-period memory network to generate a preliminary prediction sequence; And carrying out inverse normalization on the preliminary prediction sequence to obtain a passenger flow prediction value, and constructing a passenger flow space-time distribution prediction tensor of the exhibition area according to the passenger flow prediction value.
  4. 4. The intelligent scheduling method of museum passenger flow based on time sequence prediction as claimed in claim 1, wherein the predicting tensor according to the passenger flow space-time distribution, using the exhibition area with passenger flow predicted value exceeding the capacity of the preset exhibition area as a potential congestion pressure source node, and performing network diffusion deduction on the potential congestion pressure source node based on a preset topological connection relation to obtain a dynamic influence deduction graph of the exhibition area, comprises: According to the passenger flow space-time distribution prediction tensor, taking an exhibition area with the passenger flow prediction value exceeding the capacity of a preset exhibition area as a potential congestion pressure source node; constructing a topological connection relationship based on a physical communication channel of a building plane electronic graph in the museum; Based on the historical visitor visit line data of the exhibition area, counting a visitor natural transition probability matrix of the exhibition area; Marking the excess guest flow value of the potential congestion pressure source node as an initial pressure source; simulating a layer-by-layer conduction process of the exhibition area from the initial pressure source according to the topological connection relation and the guest natural transfer probability matrix; And in the layer-by-layer conduction process, recording the aggregate pressure value of the exhibition area and the pressure arrival prediction time of the exhibition area in real time, stopping simulation when the pressure values on all the conduction paths are lower than a preset influence threshold value, and outputting the labeled topology network as the dynamic influence deduction graph.
  5. 5. The intelligent scheduling method for museum passenger flow based on time sequence prediction as set forth in claim 4, wherein the simulating the layer-by-layer conduction process of the exhibition area comprises: establishing a pressure conduction iteration state list of the exhibition area, and storing the pressure value of the initial pressure source and the exhibition area node of the museum into the pressure conduction iteration state list; performing transition matrix probability reading on the exhibition area nodes in the pressure conduction iteration state list to obtain a transition probability value of the exhibition area; Calculating an exhibition area conduction expected pressure value of the museum based on the transition probability value and the exhibition area pressure value to be conducted; the calculation formula of the expected pressure value of the exhibition area conduction is as follows: Wherein, the Is the exhibition node The number of indices of the set is, Is the exhibition node The number of indices of the set is, The desired pressure value is conducted for the exhibition area, To be the pressure value to be conducted for the exhibition, To be from exhibition area node Node to exhibition area Is a natural transition probability of the historical tourists, To be from exhibition area node Node to exhibition area Is used for the diffusion damping factor of (a), Is natural constant as the base Is a function of the exponent of (c), As a diffusion attenuation coefficient for the exhibition region, Node from exhibition area for tourists in exhibition area Node to exhibition area The transfer time required; Taking the exhibition region conduction expected pressure value and the corresponding exhibition region node as source parameters of the layer-by-layer conduction process; and carrying out iterative loop on the layer-by-layer conduction process based on the source parameters until the pressure conduction quantity is lower than the influence threshold value, and stopping simulation.
  6. 6. The intelligent scheduling method of museum passenger flow based on time sequence prediction as set forth in claim 5, wherein the performing policy search in a multidimensional policy space of the exhibition area based on the dynamic influence deduction graph to generate a target scheduling policy set, the multidimensional policy space including a path adjustment instruction, an entrance management instruction and a service resource instruction includes: carrying out graph structural feature analysis on the dynamic influence deduction graph, and identifying a pressure-bearing high-value node set in the dynamic influence deduction graph and a key conduction path set of the initial pressure source; Matching a path strategy space of the key conduction path set in a preset path strategy database; Constructing a node strategy space of the exhibition area node based on the bearing high-value node set and the space layout parameters of the exhibition area node; carrying out Cartesian product combination on the path strategy space and the node strategy space to generate an initial candidate strategy combination; sequentially inputting the initial candidate strategy combination into a preset strategy deduction engine, adjusting the passage state and the node admittance flow in the topological connection relation according to a path adjustment instruction, an entrance management instruction and a service resource instruction in the initial candidate strategy combination, then starting to simulate the layer-by-layer conduction process of the exhibition area again from the initial pressure source, and outputting a new influence deduction graph after disturbance; taking the new influence deduction image after disturbance as a disturbance effect image of the exhibition area; Comparing and analyzing the disturbance effect graph with the dynamic influence deduction graph, and analyzing the performance index and implementation cost quantification value of the candidate strategy combination according to a comparison result; And comprehensively evaluating the candidate strategy combination based on the efficiency index and the implementation cost quantification value, and taking the high-score strategy combination in the comprehensive evaluation result as a target scheduling strategy set.
  7. 7. The intelligent scheduling method for museum passenger flow based on time sequence prediction as set forth in claim 6, wherein the constructing a node policy space of the exhibition node based on the set of bearing high-value nodes and the spatial layout parameters of the exhibition node includes: collecting a current people flow distribution thermodynamic diagram of a real-time monitoring image in the pressure-bearing high-value node set; Converting the exhibition area layout plan of the real-time monitoring image into a space grid map, and superposing the current people flow distribution thermodynamic diagram on the space grid map to generate a local high-density grid region; Based on the position of the local high-density grid area and the channel position relation of the space grid map, carrying out dispersion strategy analysis on the exhibition area to obtain dynamic dispersion strategy options of the exhibition area nodes; Parameterizing the dynamic dispersion policy options, and carrying out structural encapsulation on parameterized results to obtain a node policy space of the exhibition area node.
  8. 8. The intelligent scheduling method for museum passenger flow based on time sequence prediction as set forth in claim 6, wherein the comprehensively evaluating the candidate policy combination based on the performance index and the implementation cost quantization value and taking a high-score policy combination in the comprehensive evaluation result as a target scheduling policy set comprises: comparing the total pressure value of the disturbance effect graph on the key conduction path set with the total pressure value of the corresponding path in the dynamic influence deduction graph, and determining a path pressure relief rate index based on a comparison result; Comparing the number of nodes with the node pressure value lower than a preset safety threshold value in the pressure-bearing high-value node set with the number of nodes with the node pressure value lower than the preset safety threshold value on the disturbance effect graph, and determining a node safety improvement rate index according to a comparison result; Performing cost evaluation on the candidate strategy combination, and determining an implementation cost index of the candidate strategy combination; and carrying out weighted scoring on the path pressure relief rate index, the node safety improvement rate index and the implementation cost index to obtain a comprehensive score of the candidate strategy combination, and taking the high-score strategy combination in the comprehensive evaluation as a target scheduling strategy set.
  9. 9. The intelligent scheduling method for museum passenger flow based on time sequence prediction as set forth in claim 6, wherein the time sequence scheduling of the target scheduling policy set is converted into an executable instruction sequence, and the issuing of the executable instruction sequence includes: analyzing the target scheduling strategy set to obtain a logic dependency item, a hardware resource identifier and a software interface identifier of a strategy instruction in the target scheduling strategy set; constructing a directed dependency graph among the strategy instructions based on the logic dependency items; performing time axis arrangement on the topology ordering result of the directed dependency graph and the estimated execution time length of the strategy instruction to generate an initial timing sequence arrangement scheme; performing resource conflict detection on the initial timing sequence scheme, and identifying instruction conflict pairs of the initial timing sequence scheme; based on the instruction priority rule of the exhibition area, eliminating resources of the instruction conflict pair to generate a final timing sequence arrangement scheme; Compiling the strategy instruction into a device identifiable control instruction according to the execution sequence of the instruction in the final timing arrangement scheme, packaging the identifiable control instruction and a resource allocation list corresponding to the identifiable control instruction into an executable instruction sequence, and issuing the executable instruction sequence.
  10. 10. A timing prediction-based intelligent scheduling system for museum guest flow, for implementing the timing prediction-based intelligent scheduling method for museum guest flow of claim 1, the system comprising: the system comprises a sequence construction module, a display module and a display module, wherein the sequence construction module is used for acquiring basic passenger flow data and associated environmental factor data of an exhibition area in a museum so as to construct a multi-element time sequence data sequence; The trend decomposition and feature extraction module is used for carrying out trend decomposition and periodic feature extraction on the multi-element time sequence data to generate a feature vector cluster of the exhibition area; The rolling prediction and tensor construction module is used for carrying out multi-step cross-period rolling fusion prediction on the characteristic vector cluster to obtain a passenger flow predicted value, and constructing passenger flow space-time distribution prediction tensors of the exhibition area according to the passenger flow predicted value; the congestion deduction and influence analysis module is used for predicting tensors according to the passenger flow space-time distribution, taking an exhibition area with the passenger flow predicted value exceeding the capacity of a preset exhibition area as a potential congestion pressure source node, and carrying out network diffusion deduction on the potential congestion pressure source node based on a preset topological connection relation to obtain a dynamic influence deduction diagram of the exhibition area; The strategy searching and generating module is used for carrying out strategy searching in a multidimensional strategy space of the exhibition area based on the dynamic influence deduction graph to generate a target scheduling strategy set, wherein the multidimensional strategy space comprises a path adjustment instruction, an entrance management instruction and a service resource instruction; And the timing sequence arrangement and instruction execution module is used for performing timing sequence arrangement on the target scheduling strategy set, converting the target scheduling strategy set into an executable instruction sequence and issuing the executable instruction sequence for execution.

Description

Intelligent museum passenger flow scheduling method and system based on time sequence prediction Technical Field The invention relates to the technical field of intelligent scheduling, in particular to a museum passenger flow intelligent scheduling method and system based on time sequence prediction. Background In the prior art, museum passenger flow scheduling generally depends on manual experience or simple passenger flow statistics, lacks deep mining and fusion analysis of multi-element time sequence data, is difficult to capture periodic change and trend characteristics of passenger flow, causes insufficient passenger flow prediction precision, and cannot provide reliable basis for scheduling decisions. In the prior art, the identification of potential congestion is often limited to a single exhibition area, dynamic deduction capability of diffusion conduction of passenger flow pressure between exhibition areas is lacking, chain reaction of congestion is difficult to predict, and quantitative evaluation of passenger flow pressure propagation paths is difficult. In the prior art, a fixed rule or a template is mostly adopted for generating a scheduling strategy, targeted strategy searching and optimization are lacked, and path adjustment instructions, entrance management and control instructions and service resource instructions are difficult to dynamically adjust in real time, so that the scheduling measure effect is poor. In the prior art, the instruction execution lacks timing coordination arrangement, resource conflict is easy to cause, and the ordering and the high efficiency of multi-strategy parallel execution are difficult to ensure. Therefore, development of a method and a system for intelligent scheduling of museum passenger flow based on time sequence prediction are needed to solve the problems of inaccurate passenger flow prediction, congestion deduction and missing, rough strategy generation and disordered instruction execution in the prior art, and improve the intelligent level and emergency response efficiency of museum passenger flow scheduling. Disclosure of Invention The invention provides a museum passenger flow intelligent scheduling method and system based on time sequence prediction, which are used for solving the problems in the background technology. In order to achieve the above purpose, the invention provides a method for intelligent scheduling of museum passenger flow based on time sequence prediction, which comprises the following steps: C1, collecting basic passenger flow data and associated environmental factor data of an exhibition area in a museum to construct a multi-element time sequence data sequence; c2, carrying out trend decomposition and periodic feature extraction on the multi-element time sequence data to generate a feature vector cluster of the exhibition area; C3, performing multi-step cross-period rolling fusion prediction on the feature vector cluster to obtain a passenger flow predicted value, and constructing a passenger flow space-time distribution prediction tensor of the exhibition area according to the passenger flow predicted value; C4, taking an exhibition area with the passenger flow predicted value exceeding the capacity of a preset exhibition area as a potential congestion pressure source node according to the passenger flow space-time distribution prediction tensor, and carrying out network diffusion deduction on the potential congestion pressure source node based on a preset topological connection relation to obtain a dynamic influence deduction graph of the exhibition area; c5, performing strategy search in a multidimensional strategy space of the exhibition area based on the dynamic influence deduction graph to generate a targeted scheduling strategy set, wherein the multidimensional strategy space comprises a path adjustment instruction, an entrance management instruction and a service resource instruction; And C6, carrying out time sequence arrangement on the target scheduling strategy set, converting the target scheduling strategy set into an executable instruction sequence, and issuing the executable instruction sequence. In a preferred embodiment, the trend decomposition and the periodic feature extraction are performed on the multi-element time series data sequence to generate a feature vector cluster of the exhibition area, which includes: trend decomposition is carried out on the multi-element time sequence data sequence, and a trend component, a seasonal component and a residual component are separated; performing fast Fourier transform extraction on the seasonal components to obtain main periodic frequency and periodic intensity, and taking the main periodic frequency and the periodic intensity as first periodic characteristics; carrying out mean analysis and variance analysis on the first-order differential sequence of the trend component, and taking a mean analysis result and a variance analysis result as trend change characteristics;