An intelligent seat arrangement optimization method for a convention

Abstract

The application discloses a kind of intelligent seat arrangement optimization method of convention and exhibition, belong to the technical field of convention and exhibition management and intelligent scheduling.This method obtains the object identification data of participant and calculates social gravity coefficient and physical constraint coefficient, obtains seat space attribute and calculates seat basic evaluation value, to seated personnel to construct seat arrangement comprehensive evaluation value;New participant arrives to idle seat as candidate set to calculate marginal evaluation value increment and select initial seat, execute local evaluation of allowing peripheral fine-tuning, determine final seat according to reduction amplitude;Iterate to local variance convergence after recording convergence time and updating social gravity coefficient, protect special demand personnel by elastic binding coefficient;Generate control instruction to drive terminal to display seat arrangement result, use historical data to carry out cross-convention and exhibition optimization on preset weight.The application realizes real-time incremental seat arrangement under flow arrival, has self-evolution ability, improves seat arrangement efficiency and adaptation accuracy.

Description

Technical Field

[0001] This invention belongs to the field of exhibition management and intelligent scheduling technology, and in particular relates to an intelligent seating optimization method for exhibitions. Background Technology

[0002] Intelligent seating arrangement at conferences and exhibitions is a core aspect of the organization and management of large-scale conferences, industry summits, product launches, and other events, directly impacting the experience and communication effectiveness of attendees. Existing seating arrangements primarily fall into two categories: rule-based static allocation methods and optimization algorithm-based intelligent seating methods. Rule-based static allocation methods typically sort seats according to a single dimension such as attendee job title, guest level, or registration time, for example, placing important guests in the front row and ordinary attendees in the back. This type of method is simple to operate but lacks flexibility. Optimization algorithm-based intelligent seating arrangements mainly employ heuristic algorithms such as genetic algorithms, simulated annealing algorithms, or particle swarm optimization algorithms. The optimization objective is to maximize social proximity or minimize physical inconvenience, calculating the globally optimal seating allocation scheme in one go after obtaining all attendee information. Furthermore, some methods introduce weighted summation strategies to compromise between multiple objectives, balancing factors such as social relationships, field of vision, and accessibility by setting different weight coefficients. Related technical solutions can be found in Chinese patents CN118735486A and CN111091209A. With the digital transformation of the exhibition industry, intelligent seating systems are gradually being integrated with hardware devices such as exhibition management platforms, check-in systems, and electronic nameplates.

[0003] However, the existing technologies still have the following shortcomings. First, existing methods utilize only one dimension of participant identification, mainly limited to identity recognition information, failing to fully leverage the multi-dimensional semantic information contained in the participant identification, such as social relationship identifiers (e.g., historical interaction records, cooperation intentions) and physical constraint identifiers (e.g., mobility assistance needs, audiovisual assistance needs), for comprehensive seating arrangement. This results in seating arrangements that cannot accurately adapt to the differentiated needs of participants. Second, existing methods all adopt a global one-time seating strategy, requiring calculation only after all participant information is complete. This cannot handle the streaming scenario where participants arrive and sign in sequentially. If new participants arrive after seating is completed, manual arrangement or a global recalculation is necessary, resulting in high computational costs and impacting the on-site experience. Third, existing methods lack quantitative feedback and closed-loop correction mechanisms for seating arrangement effectiveness. The process terminates after seating is completed, making it impossible to quantify and evaluate the merits of the seating arrangement, let alone feed the actual seating effect back to subsequent seating processes for continuous optimization. Fourth, the existing methods cannot automatically drive seat-binding terminals to display information. The seating arrangement scheme still relies on manual operation from algorithm calculation to physical presentation, and a complete "collection-computation-control" technical closed loop has not yet been formed. Fifth, the weight coefficients between multiple objectives in the existing methods rely on manual experience to be set once, and cannot be dynamically adjusted according to the actual convergence efficiency during the seating process. Moreover, the seating data between different sessions are independent of each other, and the weight parameters cannot be learned and optimized from historical seating experience. Sixth, the existing methods only achieve one-time allocation for participants with special needs such as mobility impairments through constraints, and lack a protection mechanism for the seats of participants with special needs during subsequent dynamic adjustments. Their allocated prime seats may be reassigned in subsequent reshuffling or adjustments. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, this invention provides an intelligent seating arrangement optimization method for exhibitions, which solves six problems in the prior art, including single object identification dimension, inability to cope with streaming arrivals, lack of closed-loop feedback, inability to automatically drive hardware devices, reliance on manual weight setting and inability to learn across exhibitions, and lack of dynamic protection for personnel with special needs.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A method for optimizing intelligent seating arrangements at exhibitions includes the following steps: Step 1: Obtain the object identification data of the participants and calculate the social attraction coefficient and physical constraint coefficient for each participant; Step 2: Obtain the spatial attribute data of the seats and calculate the basic evaluation value of each seat; Step 3: Construct a comprehensive evaluation value for seating arrangement based on the seated personnel and occupied seats. The comprehensive evaluation value for seating arrangement is composed of the social attraction evaluation value, the physical constraint evaluation value, and the basic seat evaluation value. Step 4: When new participants arrive, use the available seats as the candidate set, calculate the marginal evaluation value increment, and select the L candidate seats with the smallest increment as the initial selection seats. Step 5: Perform local fine-tuning evaluation on each initially selected seat, allowing those already seated in the surrounding area to move within the preset space, and determine the final seat based on the reduction in local evaluation value; Step six: Assign new participants to their final seats. If the variance of local evaluation values ​​is greater than the preset convergence threshold, repeat step five until convergence is achieved. Step 7: Record the convergence time of each seat allocation adjustment, and update the social attraction coefficient of the participants involved in the adjustment based on the convergence time. Step 8: Generate a seat control command and send it to the seat binding terminal to drive the seat binding terminal to display the corresponding attendee information.

[0006] Preferably, the object identification data includes identity information, industry domain tags, interactive object identifiers from historical participation records, cooperation intention markers, special needs information, and estimated arrival time; The identity verification information includes name, ID number, and / or facial feature code; The special needs information includes mobility assistance needs identifiers and / or audiovisual assistance needs identifiers.

[0007] Preferably, the spatial attribute data includes two-dimensional plane coordinates, field of view obstruction coefficient, sound coverage quality, distance from the podium, and distance from the nearest exit; The field of view occlusion coefficient is a value that is greater than or equal to 0 and less than or equal to 1; The sound coverage quality is a value greater than or equal to 0 and less than or equal to 1.

[0008] Preferably, the social attraction coefficient is calculated as follows: The job level of the participants is mapped to the industry influence index, with the mapping rule being that ordinary participants are mapped to 1, department heads to 2, company leaders to 3, and industry experts to 4. Multiply the industry influence index by the first preset weight to obtain the first product term; The sum of the social connection strengths of the participant with all other participants is multiplied by the second preset weight to obtain the second product term, where the value of the social connection strength between the participant and any other participant is: 1 if there is an interaction record with the other participant in the historical meeting records, and 0 otherwise. Multiply the participant's cooperation intention activity level by the third preset weight to obtain the third product term, where the cooperation intention activity level is the number of cooperation intention tags marked as "high" divided by the total number of tags; Add the first product term, the second product term, and the third product term to obtain the social attraction coefficient of the participant; The first preset weight, the second preset weight, and the third preset weight are all greater than 0, and the sum of the three is equal to 1.

[0009] Preferably, the calculation method for the basic seat evaluation value is as follows: Multiply the difference between 1 and the field of view occlusion coefficient by the fourth preset weight to obtain the fourth product term; Subtract the difference in sound coverage quality from 1 and multiply it by the fifth preset weight to obtain the fifth product term; Multiply the normalized distance to the podium by the sixth preset weight to obtain the sixth product term; Multiply the difference between 1 and the accessibility coefficient by the seventh preset weight to obtain the seventh product term; Add the fourth, fifth, sixth, and seventh product terms together to obtain the basic evaluation value of the seat. The accessibility coefficient is set as follows: if the distance to the nearest exit is less than or equal to 2 meters, it is 1; if it is greater than 2 meters but less than or equal to 5 meters, it is 0.5; if it is greater than 5 meters, it is 0. The fourth preset weight, the fifth preset weight, the sixth preset weight, and the seventh preset weight are all greater than 0, and their sum equals 1.

[0010] Preferably, the social attraction evaluation value is calculated as follows: For all seated individuals, randomly select two different individuals, multiply their social attraction coefficients, divide by the square of the distance between their seats, and sum the results of all pairwise combinations. The distance between seats is a normalized value obtained by dividing the Euclidean distance in planar coordinates by the maximum diagonal length of the seating area in the exhibition venue.

[0011] Preferably, the physical constraint evaluation value is calculated as follows: For each person who is already seated, multiply the physical constraint coefficient by the environmental attribute value of their current seat, and sum the products of all seated persons. The environmental attribute values ​​consist of the seat's field of vision obstruction coefficient, sound coverage quality, and distance from the podium; The basic seat evaluation value is obtained by summing the basic seat evaluation values ​​of all occupied seats.

[0012] Preferably, the value of L is greater than or equal to 3 and less than or equal to 10; The preset threshold value ranges from greater than or equal to 0.01 to less than or equal to 0.1. The preset space range is a rectangular area centered on the seat being evaluated, consisting of one row in front and behind, and three seats on each side. The variance of the local evaluation value is calculated as follows: taking the final seat as the center, the rectangular area formed by one row in front and behind and three seats on each side as the local area, and calculating the variance of the comprehensive evaluation value of all occupied seats in the local area.

[0013] Preferably, the specific method for updating the social gravity coefficient is as follows: Multiply the current social attraction coefficient of the participant by a coefficient, which is 1 minus a product value, where the product value is a preset learning rate multiplied by the arithmetic mean of the participant's convergence time and then divided by a preset maximum acceptable convergence time threshold. The arithmetic mean of the convergence time is the arithmetic mean of the convergence times of all seat allocation adjustments in which the participant participated. The preset learning rate is greater than 0 and less than 1, and the preset maximum acceptable convergence time threshold is a positive integer. The preset learning rate has a value range of greater than or equal to 0.01 and less than or equal to 0.1, and the preset maximum acceptable convergence time threshold has a value range of greater than or equal to 5 and less than or equal to 20.

[0014] Preferably, the following steps are also included: All seating arrangement process data for each exhibition is stored in a historical database. The seating arrangement process data includes object identification data, initial and final evaluation values ​​for each seat allocation adjustment, convergence time for each seat allocation adjustment, and the identification of the participants and their seats involved. Before the execution of subsequent exhibitions, the data in the historical database is used as training samples, and the first to seventh preset weights are optimized and updated with the goal of minimizing the average convergence time. The optimized weights are then used as calculation parameters for subsequent exhibitions.

[0015] The technical effects and advantages of the intelligent seating arrangement optimization method for exhibitions proposed in this invention are as follows: 1. By obtaining object identification data and calculating social attraction coefficient and physical constraint coefficient in step one, this invention achieves accurate modeling of participants from multiple dimensions such as identity recognition, social relationship, and physical constraint. This enables seating optimization to simultaneously take into account the social proximity needs, physical comfort needs, and special assistance needs of participants, overcoming the shortcomings of existing seating methods that allocate seats based on only a single identity information.

[0016] 2. In step two, this invention calculates the basic evaluation value of a seat based on its field of vision obstruction coefficient, sound coverage quality, distance from the podium, and distance from the nearest exit. It also dynamically sets a preset space range based on the total number of seats. This achieves a precise match between the physical attributes of the seats and the needs of the attendees, ensuring that the seat allocation results meet individual differences and are suitable for different scales of conferences and exhibitions.

[0017] 3. By constructing a comprehensive evaluation value for seating arrangement, calculating the marginal evaluation value increment, performing local fine-tuning evaluation, and iteratively converging until the variance is no greater than a preset convergence threshold in steps three to six, this invention enables incremental allocation and fine-tuning optimization of only local areas when new attendees arrive, without the need for a global rearrangement of all already seated attendees. This significantly reduces the time complexity of seating arrangement calculations while ensuring the stability and continuity of the seating arrangement scheme.

[0018] 4. By recording the convergence time of each seat allocation adjustment in step seven and updating the social attraction coefficient of the participants involved in the adjustment based on the convergence time, this invention achieves real-time quantitative feedback of the seating arrangement effect and dynamic correction of the social attraction coefficient. This enables the system to automatically adjust the parameter input of subsequent seating arrangements based on the convergence efficiency of the actual seating arrangement process, overcoming the shortcomings of existing methods that terminate after the seating arrangement is completed and cannot self-optimize.

[0019] 5. This invention utilizes a cross-exhibition knowledge transfer mechanism, using seating process data from historical databases as training samples, and optimizes and updates preset weights with the goal of minimizing average convergence time. This enables the system to continuously learn from historical exhibition seating experience, achieving a self-evolutionary capability where seating efficiency continuously improves with increasing usage frequency.

[0020] 6. This invention protects the seats of attendees with special needs by using an elastic binding coefficient. When subsequent local adjustments involve moving their seats, the system prioritizes avoiding movement, ensuring the stability of seats for attendees with special needs while enjoying personalized seating services. This overcomes the defect in existing methods where the prime seats of attendees with special needs are easily reassigned during subsequent optimizations.

[0021] 7. In step eight, the present invention generates seat control instructions based on the final seat allocation results and drives the seat binding terminal to display the attendee information, forming a complete technical closed loop of "data acquisition - seat calculation - control output". This enables the seat optimization results to be directly applied to the physical seat terminal equipment, realizing the automated execution of the seat allocation scheme from algorithm output to physical presentation. Attached Figure Description

[0022] Figure 1 This is an overall flowchart of the intelligent seating arrangement optimization method for exhibitions according to the present invention; Figure 2This is a flowchart of the local fine-tuning evaluation and iterative convergence process of the present invention; Figure 3 This is a diagram showing the system module architecture of the present invention; Figure 4 This is a flowchart illustrating the cross-exhibition knowledge transfer process of the present invention. Figure 5 This is a schematic diagram illustrating the composition of the comprehensive evaluation value of the seating arrangement in this invention. Detailed Implementation

[0023] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0024] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include," "contain," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "includes..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0025] refer to Figure 1This invention provides an intelligent seating optimization method for exhibitions, belonging to the field of exhibition management and intelligent scheduling technology. It addresses the technical problems of existing seating methods, such as a single object identification dimension, inability to handle continuous arrivals, lack of closed-loop self-evolution, and lack of dynamic protection for personnel with special needs. The method first acquires the object identification data of attendees to calculate social attraction coefficients and physical constraint coefficients, and acquires seat space attribute data to calculate basic seat evaluation values. A comprehensive seating evaluation value is constructed using already seated attendees and occupied seats. When new attendees arrive, the marginal evaluation value increment is calculated using vacant seats as a candidate set, and initial seats are selected. A local evaluation is performed on the initial seats, allowing for minor adjustments to surrounding seats. The final seats are determined based on the magnitude of the evaluation value reduction. Iterative fine-tuning continues until the local evaluation value variance converges. The convergence time is recorded, and the social attraction coefficients of the attendees are updated. Simultaneously, a flexible binding coefficient protects personnel with special needs. Finally, seat control commands are generated to drive the terminal to display the seating results, and historical data is used to optimize and update preset weights across exhibitions with the goal of minimizing the average convergence time. This invention realizes real-time incremental seating arrangement in streaming arrival scenarios. Through local fine-tuning, feedback correction, and cross-exhibition learning, the system has the ability to self-evolve. Through control commands, the algorithm is automatically executed to hardware, which effectively improves the calculation efficiency, adaptation accuracy, and intelligence level of seating arrangement.

[0026] Example 1 Purpose of implementation: This embodiment aims to illustrate the complete implementation process of the present invention in a small industry seminar scenario with a relatively small number of participants (80 people), focusing on verifying: (1) the dynamic setting logic of the system for the preset space range when the total number of seats is less than 100; (2) the execution flow of incremental allocation and local fine-tuning in a streaming check-in scenario; and (3) the complete link of convergence time recording and social gravity coefficient correction in a single local field reconstruction. Through this embodiment, it is demonstrated that the present invention can achieve real-time seating optimization when participants arrive one after another in a small exhibition scenario.

[0027] Implementation steps: Step one: The central processing unit (CPU) acquires the object identification data of 80 attendees collected by the check-in sensor. Taking one attendee—Zhang, the R&D director of a technology company—as an example, his object identification data includes: identity information (name "Zhang", ID number, facial feature code), industry domain label ("Artificial Intelligence"), interaction object identifiers from past attendee records (interaction records with 5 other attendees in past exhibitions), cooperation intention markers (3 marked "high", 1 marked "medium", and 2 marked "low"), special needs information (no need for mobility assistance), and estimated arrival time (8:50 AM). The CPU calculates Zhang's social attraction coefficient and physical constraint coefficient based on the above object identification data. Zhang's job level is department head, corresponding to an industry influence index of 2. He has historical interaction records with 5 other participants, with a sum of social connection strengths of 5. The calculation method for his cooperation intention activity level is as follows: The central processing unit counts the number of cooperation intention tags marked "high" in Zhang's data, which is 3. Dividing this number by the total number of cooperation intention tags (6), we get a cooperation intention activity level of 0.5. The central processing unit multiplies Zhang's industry influence index 2 by a first preset weight, which has a value of 0.3, resulting in a first product term of 0.6. It multiplies the sum of Zhang's social connection strengths with all other participants (5) by a second preset weight, which has a value of 0.4, resulting in a second product term of 2.0. Finally, it multiplies Zhang's cooperation intention activity level of 0.5 by a third preset weight, which has a value of 0.3, resulting in a third product term of 0.15. The central processing unit (CPU) adds the first product term 0.6, the second product term 2.0, and the third product term 0.15 to obtain Zhang's social attraction coefficient as 2.75. The CPU then determines Zhang's physical constraint coefficient based on his specific needs. Since Zhang has no specific needs, the physical constraint coefficient is set to the default value of 0.5.

[0028] Step two: The central processing unit acquires the spatial attribute data of 80 seats in the exhibition venue. This spatial attribute data includes each seat's two-dimensional planar coordinates, row number, column number, field-of-view obstruction coefficient, sound coverage quality, distance from the podium, and distance from the nearest exit. Taking the seat in the 5th row, 3rd column as an example, its field-of-view obstruction coefficient is 0.1, its sound coverage quality is 0.85, its normalized distance from the podium is 0.30, and its distance from the nearest exit is 3.5 meters. The central processing unit (CPU) calculates the basic evaluation value of the seat: CPU Calculation 1 minus the seat's field-of-view obstruction coefficient of 0.1 equals 0.9, multiplied by the fourth preset weight (0.25), resulting in a fourth product term of 0.225; CPU Calculation 1 minus the seat's sound coverage quality of 0.85 equals 0.15, multiplied by the fifth preset weight (0.25), resulting in a fifth product term of 0.0375; the normalized distance from the seat to the podium of 0.30 is multiplied by the sixth preset weight (0.25), resulting in a sixth product term of 0.075; the distance from the seat to the nearest exit is 3.5 meters, greater than 2 meters and less than or equal to 5 meters, with an accessibility coefficient of 0.5. CPU Calculation 1 minus 0.5 equals 0.5, multiplied by the seventh preset weight (0.25), resulting in a seventh product term of 0.125. The central processing unit adds the fourth product term 0.225, the fifth product term 0.0375, the sixth product term 0.075, and the seventh product term 0.125 to obtain the basic seat evaluation value of 0.4625.

[0029] Meanwhile, the central processing unit sets the preset space range in step five as a rectangular area centered on the seat being evaluated, with half a row in front and behind and two seats on each side, based on the total number of seats being 80 (less than 100).

[0030] Step 3: The central processing unit (CPU) constructs a comprehensive seating arrangement evaluation value based on the seated personnel and occupied seats. This comprehensive evaluation value is composed of a social attraction evaluation value, a physical constraint evaluation value, and a basic seat evaluation value. The social attraction evaluation value is calculated as follows: For all seated personnel, the CPU randomly selects two different individuals, multiplies their social attraction coefficients, divides by the square of the distance between their seats, and sums the results of pairwise combinations for all personnel. The distance between seats is calculated as follows: The CPU obtains the planar coordinates of the seats of the two individuals, calculates the Euclidean distance, and divides it by the maximum diagonal length of the seating area in the exhibition venue to obtain a normalized seat distance value. The physical constraint evaluation value is calculated as follows: For each seated personnel, the CPU multiplies their physical constraint coefficient by the environmental attribute value of their current seat (composed of the seat's view obstruction coefficient, sound coverage quality, and distance from the podium), and sums the products of all seated personnel. The basic seat evaluation value is obtained by summing the basic seat evaluation values ​​of all occupied seats. Assuming that 40 participants have already checked in and taken their seats, the central processing unit calculates the overall seating evaluation value in the manner described above.

[0031] Step four: When a new attendee (let's say Professor Li from a university, the 41st person to arrive) completes check-in, the central processing unit (CPU) uses all available seats as a candidate set. In this embodiment, the currently available seats are the 40 seats in rows 5 to 10. The CPU calculates the marginal evaluation value increment after assigning Professor Li to each candidate seat. Taking the seat in row 5, column 3 as an example, assuming the overall evaluation value of the row before assignment is 156.8 and the overall evaluation value after assignment is 158.2, the CPU calculates the marginal evaluation value increment for this seat as the post-assignment evaluation value of 158.2 minus the pre-assignment evaluation value of 156.8, with a difference of 1.4. The CPU calculates the marginal evaluation value increment for each of the 40 candidate seats and selects the 5 seats with the smallest increments as initial selection seats (L is set to 5).

[0032] Step 5: The central processing unit (CPU) performs a local fine-tuning evaluation on each initially selected seat. Taking the initially selected seat with the smallest marginal increment (row 6, column 4) as an example: the CPU determines whether the local evaluation value decreases after allowing seated personnel in the same row and no more than 2 seats away from this seat (i.e., those already seated in rows 6, columns 2, 3, 5, and 6) to move no more than 2 seats left or right within their current row. The local evaluation value is calculated within a rectangular area centered on the evaluated seat, encompassing all seats and personnel seated there within the area formed by two seats in front and behind, and two seats to the left and right. The calculation shows that the local evaluation value before adjustment was 45.6, and the local evaluation value after adjustment was 43.2. The CPU calculates the reduction as 45.6 minus 43.2, a difference of 2.4. Dividing this by the original 45.6 yields a reduction of 5.26%, exceeding the preset threshold. The preset threshold is 0.03 (3%), therefore the CPU marks this seat as an adjustable seat. After evaluating all five initial seats, the central processing unit selects the adjustable seat with the largest decrease in local evaluation value as the final seat and adjusts the seats of the already seated personnel according to the corresponding adjustment plan. If there is no adjustable seat among the initial seats, the central processing unit selects the initial seat with the smallest marginal evaluation value increment as the final seat and does not perform seat adjustment.

[0033] Step Six: The CPU assigns Li to the final seat, updates the set of arrived personnel and the set of occupied seats, and uses a rectangular area centered on the final seat, consisting of two rows in front and behind and two seats to the left and right, as a local area. It calculates the variance of the overall seating arrangement evaluation value for all occupied seats within this local area. If the variance exceeds a preset convergence threshold (0.02), the CPU repeats Step Five until the variance is no greater than 0.02. In this allocation adjustment, after repeating Step Five twice, the local evaluation value variance is no greater than 0.02, and the CPU proceeds to Step Seven.

[0034] Step 7: The central processing unit (CPU) records the convergence time of this seat allocation adjustment as 3 iterations (the number of iterations from step 4 to step 6 when the local evaluation value variance is no greater than 0.02). Li participated in this seat allocation adjustment, and the CPU calculates Li's convergence time arithmetic mean as 3 iterations. The CPU updates Li's social attraction coefficient as follows: The CPU multiplies Li's current social attraction coefficient by a coefficient calculated as 1 minus a product value. This product value is a preset learning rate multiplied by Li's convergence time arithmetic mean divided by a preset maximum acceptable convergence time threshold. The preset learning rate is 0.05, Li's convergence time arithmetic mean is 3, and the preset maximum acceptable convergence time threshold is 10. Therefore, the product value is 0.05 multiplied by 3 divided by 10, which equals 0.015. The coefficient is 1 minus 0.015, which equals 0.985. Li's updated social attraction coefficient is his current social attraction coefficient multiplied by 0.985. The central processing unit stores the updated social attraction coefficient in the database, replacing the original social attraction coefficient.

[0035] Step 8: The central processing unit generates seat control instructions based on the final seat allocation results, and sends the seat control instructions to the electronic nameplates of each seat via wireless network, driving the electronic nameplates to display the corresponding attendee's name, organization, and position information.

[0036] Implementation results: In this embodiment, 80 attendees checked in sequentially from 8:30 AM to 9:10 AM. The central processing unit executed steps four through six for each new arrival. The average response time per attendee was 0.8 seconds, and the overall seating arrangement evaluation score was optimized from an initial 178.5 to a final 132.6, a reduction of 25.7%. All attendees accurately obtained their seating information via electronic nameplates, and no seating conflicts or information errors occurred. This embodiment verifies the effectiveness and real-time performance of the invention in small-scale conference and exhibition scenarios.

[0037] Example 2 Purpose of implementation: This embodiment aims to illustrate the implementation process of the present invention in a large-scale industry summit scenario with a large number of attendees (500 people), focusing on verifying: (1) the dynamic setting logic of the system for the preset space range when the total number of seats is greater than or equal to 300; (2) the feasibility of batch calculation of the social attraction coefficient of attendees and the basic evaluation value of seats under large-scale data; and (3) the stability of incremental allocation and local fine-tuning when multiple people arrive continuously. Through this embodiment, it is demonstrated that the present invention can maintain good computational efficiency and seating arrangement quality in large-scale exhibition scenarios.

[0038] Implementation steps: Step 1: The central processing unit (CPU) acquires the object identification data of 500 attendees through a facial recognition camera. These 500 attendees include 120 company leaders, 180 department heads, 50 industry experts, and 150 general attendees. The CPU calculates the social attraction coefficient of each attendee in batches as follows: Taking an industry expert, Mr. Wang, as an example, the CPU maps Mr. Wang's job level (industry expert) to an industry influence index of 4; Mr. Wang's historical attendance records show interactions with 32 other attendees. The CPU adds the social association strength between Mr. Wang and all other attendees to obtain 32 (a value of 1 is assigned for each other attendee with whom there is an interaction record, and 0 otherwise); Mr. Wang has 8 "high" cooperation intention markers out of a total of 12. The CPU divides the 8 "high" markers by the total 12 markers to obtain a cooperation intention activity level of 0.667. The central processing unit (CPU) multiplies the industry influence index (4) by a first preset weight (0.3) to obtain the first product term (1.2); it multiplies the sum of social connection strengths (32) by a second preset weight (0.4) to obtain the second product term (12.8); and it multiplies the cooperation intention activity level (0.667) by a third preset weight (0.3) to obtain the third product term (0.2). The CPU adds the first product term (1.2), the second product term (12.8), and the third product term (0.2) to obtain Wang's social attraction coefficient as 14.2. Since Wang has no special needs, the physical constraint coefficient is set to the default value of 0.5.

[0039] Step two: The central processing unit (CPU) acquires spatial attribute data for 500 seats. Taking the seat in the first row, first column as an example, its field of vision obstruction coefficient is 0, sound coverage quality is 0.95, normalized distance from the podium is 0.02, and distance from the nearest exit is 1.5 meters. The CPU calculates the basic evaluation value of this seat: Calculate 1 minus the field of vision obstruction coefficient 0, multiply by the fourth preset weight 0.25 to obtain the fourth product term 0.25; calculate 1 minus the sound coverage quality 0.95, multiply by the fifth preset weight 0.25 to obtain the fifth product term 0.0125; multiply the normalized distance from the podium 0.02 by the sixth preset weight 0.25 to obtain the sixth product term 0.005; the distance from this seat to the exit is 1.5 meters, less than or equal to 2 meters, the accessibility coefficient is 1, calculate 1 minus 1, multiply by the seventh preset weight 0.25 to obtain the seventh product term 0. The central processing unit adds the fourth product term (0.25), the fifth product term (0.0125), the sixth product term (0.005), and the seventh product term (0) to obtain a basic seat evaluation value of 0.2675. Taking the seat in row 20, column 25 as an example, its field of vision obstruction coefficient is 0.3, its sound coverage quality is 0.4, its normalized distance from the podium is 0.95, and its distance from the nearest exit is 8 meters (greater than 5 meters, the accessibility coefficient is set to 0). The basic seat evaluation value for this seat is: 0.25 multiplied by 0.7 equals 0.175, plus 0.25 multiplied by 0.6 equals 0.15, plus 0.25 multiplied by 0.95 equals 0.2375, plus 0.25 multiplied by 1 equals 0.25, and the sum of the four terms is 0.8125.

[0040] Meanwhile, the central processing unit sets the preset space range in step five as a rectangular area centered on the seat being evaluated, consisting of two rows in front and behind and four seats on each side, based on a total of 500 seats (greater than or equal to 300).

[0041] Step 3: The central processing unit (CPU) constructs a comprehensive evaluation value for seating arrangement based on the number of seated personnel and the number of occupied seats. Since this embodiment represents the initial state before the start of the exhibition, and there are no seated personnel yet, the CPU does not execute the evaluation value calculation in Step 3 and directly proceeds to Step 4 to wait for the first attendee to sign in.

[0042] Steps four through eight: During peak registration hours (8:30 AM to 9:00 AM), 1 to 2 attendees complete registration every 10 seconds. The central processing unit (CPU) executes steps four through eight sequentially for each newly arriving attendee. Taking three consecutive attendees as an example: When attendee 301 (company representative, social attraction coefficient 8.7) arrives, the CPU executes steps four through seven, converging after three iterations; when attendee 302 (regular attendee, social attraction coefficient 1.2) arrives, due to their lower social attraction coefficient and smaller impact on local evaluation values, the CPU only needs one iteration to converge; when attendee 303 (industry expert, social attraction coefficient 15.6) arrives, due to their higher social attraction coefficient and larger impact on surrounding evaluation values, the CPU executes five iterations to converge. The CPU records the convergence time for each seat allocation adjustment and stores it in the database. In step eight, the central processing unit generates seat control commands based on the final seat allocation results and sends them to each electronic table card and smart seat indicator light via wireless network.

[0043] Implementation results: In this embodiment, 500 attendees checked in successively between 8:00 AM and 9:30 AM. The central processing unit executed steps four through eight for each new arrival. After all attendees were assigned seats, the overall seating rating was 523.6, with an average allocation response time of 1.6 seconds per attendee. Attendees with special needs (12 people with mobility impairments) were all assigned seats no more than 2 meters from the exit, achieving differentiated seating arrangements. This embodiment verifies the invention's dynamic adaptability to a preset spatial range and the high efficiency of batch calculation in large-scale exhibition scenarios.

[0044] Example 3 Purpose of implementation: This embodiment aims to illustrate the adaptation process of the present invention for attendees with special needs such as mobility impairment and audiovisual assistance, and focuses on verifying: (1) the influence mechanism of the physical constraint coefficient on the basic evaluation value of the seat; (2) the optimization process of automatically assigning attendees with special needs to suitable seats; and (3) the protective effect of the elastic binding coefficient on the seats of attendees with special needs. Through this embodiment, it is demonstrated that the present invention can effectively meet the differentiated seating requirements of people with special needs in exhibition scenarios.

[0045] Implementation steps: Step 1: The central processing unit (CPU) acquires the object identification data of 200 attendees. Among these 200 attendees are 3 wheelchair users and 2 visually impaired attendees who need to be near the podium. Taking Zhao, an attendee with mobility impairment, as an example, his object identification data includes a mobility assistance requirement identifier (wheelchair user). Based on this identifier, the CPU determines Zhao's physical constraint coefficient to be 2.0 (higher than the default value of 0.5 for ordinary attendees). Taking Sun, a visually impaired attendee, as an example, his object identification data includes an audiovisual assistance requirement identifier (visually impaired, needs to sit in the front row). Based on this identifier, the CPU determines Sun's physical constraint coefficient to be 1.8.

[0046] Step two: The central processing unit (CPU) acquires spatial attribute data for 200 seats. For example, a seat 1 meter from the nearest exit has an accessibility coefficient of 1; a seat 4 meters from the nearest exit has an accessibility coefficient of 0.5; and a seat 8 meters from the nearest exit has an accessibility coefficient of 0. When the CPU calculates the basic evaluation value for each seat, the accessibility coefficient directly affects the seat's basic evaluation value—the lower the accessibility coefficient (i.e., the farther from the exit), the higher the basic evaluation value (indicating that the seat is not user-friendly for people with mobility impairments and will be prioritized for exclusion during optimization).

[0047] Step 3: After some ordinary attendees have completed their registration and taken their seats, the central processing unit constructs a comprehensive evaluation value for seating arrangement based on the seated attendees and occupied seats.

[0048] Step four: When Zhao completes check-in and enters, the central processing unit (CPU) uses all available seats as a candidate set and calculates the marginal evaluation value increment after assigning Zhao to each candidate seat. Since Zhao's physical constraint coefficient of 2.0 is relatively high, his physical constraint evaluation value has a larger weight in the total evaluation value. Therefore, the candidate seats with the smallest marginal evaluation value increment naturally tend to be those closer to the exit and with higher accessibility coefficients. Calculations show that the top 5 candidate seats with the smallest marginal evaluation value increments are all located within 2 meters of the exit (all with accessibility coefficients of 1). The CPU selects these 5 seats as the initial selection seats.

[0049] Step 5: The central processing unit (CPU) performs a local fine-tuning evaluation on each initially selected seat. Taking one of the initially selected seats (row 3, column 4, 1.5 meters from the exit) as an example: the CPU determines whether the decrease in the local evaluation value exceeds a preset threshold of 0.03 after allowing seated individuals in the same row and no more than two seats away from the seat to move left or right within their current row. The calculation shows that the local evaluation value before adjustment was 38.2, and the local evaluation value after adjustment was 36.8. The CPU calculates the decrease as 38.2 minus 36.8, a difference of 1.4. Dividing this by the original 38.2 yields a decrease of 3.66%, exceeding the preset threshold of 0.03. The CPU marks this seat as an adjustable seat. The CPU selects the adjustable seat with the largest decrease as the final seat and performs seat fine-tuning.

[0050] Step Six: The central processing unit assigns Zhao to the final seat (1.5 meters from the exit). A rectangular area centered on the final seat, consisting of one row in front and behind, and three seats to the left and right, is designated as a local area. The variance of the overall seating arrangement evaluation value for all occupied seats within this local area is calculated. If the variance exceeds the preset convergence threshold of 0.02, Step Five is repeated until the variance is no greater than 0.02.

[0051] Step 7: The central processing unit (CPU) records the convergence time and updates Zhao's social attraction coefficient. Simultaneously, the CPU assigns Zhao a relatively high elastic binding coefficient of 0.9. When other attendees arrive and trigger the local fine-tuning in Step 5, if the fine-tuning involves moving Zhao's seat, the system will prioritize avoiding moving his seat when evaluating the fine-tuning plan due to Zhao's high elastic binding coefficient (0.9).

[0052] Step 8: The central processing unit generates seat control instructions and sends them to the electronic table cards.

[0053] Implementation results: In this embodiment, the three attendees with mobility impairments were all assigned seats no more than 2 meters from the exit (accessibility coefficient of 1 for all), and the two visually impaired attendees were assigned to the center position in the second row (visual obstruction coefficient of 0, normalized distance from the podium of 0.05). As the remaining 32 attendees arrived and triggered local adjustments, the seats of attendees with special needs, such as Mr. Zhao, remained unmoved by any of the adjustments due to their high flexibility in being bound together. This embodiment verifies the accurate adaptation and effective protection of attendees with special needs by the present invention.

[0054] Example 4 Purpose of implementation: This embodiment aims to illustrate the seat allocation optimization process for attendees with close social relationships, focusing on verifying: (1) the effectiveness of social relationship identifier resolution and social association strength quantification; (2) the guiding mechanism of social attraction coefficient on social group aggregation in the seat allocation results; and (3) the optimization role of social attraction evaluation value in the total evaluation value. Through this embodiment, it is demonstrated that the present invention can fully utilize the social relationship data of attendees to achieve collaborative seat allocation and promote the effectiveness of conference and exhibition communication.

[0055] Implementation steps: Step one: The central processing unit (CPU) acquires the object identification data of 300 attendees. These 300 attendees come from 10 different companies in the same industry, with strong social connections among attendees from the same company. The CPU constructs a social relationship graph of the attendees by analyzing the interaction object identifiers in their historical attendance records. Taking five attendees (A, B, C, D, and E) from a technology company as an example, after analyzing their historical attendance records, the CPU finds that: A interacts with B and C; B interacts with A and D; C interacts with A and E; D interacts with B; and E interacts with C. The central processing unit calculates the social attraction coefficient of each participant as follows: A's social connection strength (the sum of social connection strengths with all other participants) is 2 (1 with B and C each), B's social connection strength is 2 (1 with A and D each), C's social connection strength is 2 (1 with A and E each), D's social connection strength is 1 (only with B), and E's social connection strength is 1 (only with C). A's social attraction coefficient is calculated as follows: A's industry influence index multiplied by a first preset weight, plus A's social connection strength 2 multiplied by a second preset weight, plus A's cooperation intention activity multiplied by a third preset weight.

[0056] Step two: The central processing unit acquires the spatial attribute data of 300 seats and calculates the basic evaluation value of each seat.

[0057] Step 3: The central processing unit constructs a comprehensive seating arrangement evaluation value based on the seated individuals and occupied seats. In the social attraction evaluation value, pairs of individuals with high social attraction coefficients have a larger product. Dividing this by the square of the seat distance, closer seats contribute more to the evaluation value increment. Since lower evaluation values ​​are better, placing individuals with high social attraction coefficients in close proximity results in a larger contribution to the social attraction evaluation value (i.e., a greater increase in the overall evaluation value). In minimizing the overall evaluation value, the optimization algorithm naturally arranges individuals with high social attraction coefficients and strong inter-group relationships in close proximity, thus minimizing the increment of the social attraction evaluation value.

[0058] Step four: After A completes check-in and takes his seat, B, C, D, and E arrive in succession. The central processing unit executes step four for each new arrival in turn: calculating the marginal evaluation value increment after assigning the new attendees to their respective candidate seats. Taking B as an example, since A is already seated, B's social attraction coefficient is multiplied by A's social attraction coefficient and then divided by the square of the seat distance. The closer the distance, the larger this value (the larger the evaluation value increment, the worse the solution). Therefore, the candidate seat with the smallest marginal evaluation value increment naturally tends to be the seat closest to A.

[0059] Steps five through seven involve the central processing unit performing local fine-tuning assessments, iterative convergence, and feedback corrections for each newly arrived participant. After multiple incremental allocations and local fine-tunings, the five participants, A, B, C, D, and E, were ultimately assigned to five adjacent seats in the same row (row 8, columns 8 through 12).

[0060] Step 8: The central processing unit generates and sends seat control instructions based on the final seat allocation results.

[0061] Implementation results: In this embodiment, five attendees from the same company were ultimately assigned to five adjacent seats in the same row, achieving group seating for social participants. The central processing unit (CPU) sent seating control commands to the mobile terminal application in step eight, and all five attendees accurately obtained their seat locations and navigation routes through the application. Actual interaction monitoring data during the event showed that these five attendees engaged in 23 interactions, significantly higher than the interaction frequency of similar individuals in a randomly assigned seating scenario (historical data showed 8 to 12 interactions). This embodiment verifies the effective utilization of social relationship identifiers and the precise aggregation capability of social groups in this invention.

[0062] Example 5 Purpose of implementation: This embodiment aims to illustrate the complete process of automatic optimization of preset weights through cross-exhibition data accumulation, focusing on verifying: (1) the construction of the historical database and the data storage structure; (2) the weight optimization and update method with the goal of minimizing the average convergence time; and (3) the effect of the optimized weights on the efficiency of subsequent exhibition seating arrangements. Through this embodiment, it is demonstrated that the present invention has the adaptability of continuous evolution and can continuously improve seating efficiency as the frequency of use increases.

[0063] Implementation steps: An event organizer held 12 events of varying sizes and types within a year, each utilizing the intelligent seating optimization method provided by this invention. After each event, the central processing unit stored all seating process data in a historical database. Taking the first event (an industry summit with 500 attendees) as an example, the data stored in the historical database included: object identification data for the 500 attendees (identity information, industry tags, interaction object identifiers from historical attendance records, cooperation intention markers, special needs information, and estimated arrival time), initial and final evaluation values ​​for each seating allocation adjustment, convergence time for each seating allocation adjustment, and attendee and seat identifiers involved in each adjustment. The basic data for the 12 events is as follows: The first event was an industry summit with 500 attendees and 500 seats, and an average convergence time of 3.2 rounds. The second event was a technical seminar with 120 participants and 120 seats. The average convergence time was 2.1 rounds. The third event was a product launch, with 300 attendees and 300 seats, and an average convergence time of 2.8 rounds. The fourth conference was an academic exchange meeting with 80 participants and 80 seats. The average convergence time was 1.8 rounds. The fifth event was an industry summit, with 450 attendees and 450 seats, and an average convergence time of 3.0 rounds. The 6th event was an investment promotion and negotiation meeting, with 200 participants and 200 seats. The average convergence time was 2.3 rounds. The 7th event was a technical seminar with 150 participants and 150 seats. The average convergence time was 2.0 rounds. The 8th event was an industry summit with 600 attendees and 600 seats, and an average convergence time of 3.5 rounds. The 9th event was a product launch, with 250 attendees and 250 seats. The average convergence time was 2.5 rounds. The 10th conference was an academic exchange meeting with 60 participants and 60 seats. The average convergence time was 1.5 rounds. The 11th event was an investment promotion and negotiation meeting, with 180 participants and 180 seats. The average convergence time was 2.2 rounds. The 12th event was an industry summit with 550 attendees and 550 seats, and an average convergence time of 3.3 rounds.

[0064] Before the execution of the 13th exhibition (industry summit, with an estimated 520 attendees), the central processing unit used data from 12 exhibitions in the historical database as training samples and optimized and updated the first, second, third, fourth, fifth, sixth, and seventh preset weights with the goal of minimizing the average convergence time.

[0065] The specific optimization process is as follows: The central processing unit (CPU) divides the data from 12 exhibitions into a training set (the first 8 exhibitions) and a validation set (the last 4 exhibitions) in a 7:3 ratio. The CPU uses the convergence time of each seat allocation adjustment in the training set as the target value, and different combinations of the first to seventh preset weights as input variables to construct a Gaussian process surrogate model. The CPU searches the weight space using a data acquisition function for the weight combination that minimizes the average convergence time predicted by the surrogate model. After 50 iterations, the optimal weight combinations are: first preset weight 0.28, second preset weight 0.42, third preset weight 0.30, fourth preset weight 0.22, fifth preset weight 0.28, sixth preset weight 0.25, and seventh preset weight 0.25. The central processing unit validates the optimized weight combination on the validation set: compared with the original weight combination (first preset weight 0.3, second preset weight 0.4, third preset weight 0.3, fourth preset weight 0.25, fifth preset weight 0.25, sixth preset weight 0.25, seventh preset weight 0.25), the optimized weight combination reduces the average convergence time from 2.8 rounds to 2.3 rounds, a reduction of 17.9%.

[0066] The central processing unit uses the optimized weight combination as the calculation parameters in steps one and two of the 13th conference. In the actual operation of the 13th conference, the average convergence time for 520 participants was 2.4 rounds, which is basically consistent with the 2.3 rounds predicted by the validation set.

[0067] After the 13th event concludes, the central processing unit will add the data from this event to the historical database, providing more training samples for weight optimization of subsequent events.

[0068] Implementation results: In this embodiment, after accumulating data and optimizing the weights of 12 exhibitions, the average convergence time of the 13th exhibition decreased from 2.8 rounds before optimization to 2.4 rounds, a reduction of 14.3%. The optimization and updating of the first to seventh preset weights by the central processing unit enabled the system's seating arrangement efficiency to continuously improve with increasing usage frequency. After the 13th exhibition, the historical database contained data from all 13 exhibitions, providing richer training samples for subsequent weight optimization. This embodiment verifies the effectiveness of the cross-exhibition knowledge transfer and weight self-evolution mechanism of the present invention.

[0069] Comparative Example 1 Purpose of implementation: This comparative example aims to compare the existing genetic algorithm-based seating arrangement method for exhibitions (hereinafter referred to as "existing method") with the intelligent seating arrangement optimization method of this invention. The focus is on verifying the technical advantages of this invention in the following five aspects: (1) computational efficiency – the difference between the total time spent on seating arrangement calculation and the average response time per person; (2) seating arrangement quality – the difference between the overall seating arrangement evaluation value and the evaluation values ​​of each sub-item; (3) scenario adaptability – whether it supports the flow of attendees and local fine-tuning; (4) adaptability – whether it supports closed-loop feedback and cross-exhibition knowledge transfer; (5) special needs handling – whether it has a dynamic seating protection mechanism for attendees with special needs. This comparative example demonstrates that this invention has significant technical advancements and practical value compared to the existing method.

[0070] Implementation steps: This comparative example runs both the existing method and the method of this invention in the same exhibition setting, sharing the same hardware environment and input data. The hardware environment uses an Intel Xeon E5-2680v4 CPU and 64GB of memory. The exhibition venue is a single location with 300 seats arranged in a rectangular layout of 20 rows by 15 columns. The total number of attendees is 300, including 30 company leaders, 60 department heads, 20 industry experts, and 190 general attendees, including 2 attendees with mobility impairments requiring wheelchairs and 1 visually impaired attendee needing to be near the podium.

[0071] For input data, both methods use the exact same participant identification data (including identity information, industry tags, interaction object identifiers from historical attendance records, cooperation intention markers, special needs information, and estimated arrival time) and seating space attribute data (including two-dimensional plane coordinates, field of vision obstruction coefficient, sound coverage quality, distance to the podium, and distance to the nearest exit). Both methods use the same preset weights: the first preset weight is 0.3, the second is 0.4, the third is 0.3, the fourth is 0.25, the fifth is 0.25, the sixth is 0.25, and the seventh is 0.25.

[0072] In this comparative example, the specific implementation method of the present invention is the same as that in Example 2. The central processing unit sets the preset space range as a rectangular area centered on the seat being evaluated, consisting of two rows in front and behind and four seats on the left and right, based on the total number of seats of 300 (greater than or equal to 300). The preset threshold value is 0.03, the preset convergence threshold value is 0.02, the preset learning rate value is 0.05, the preset maximum acceptable convergence time threshold value is 10, and L is 5.

[0073] The existing method uses a global sorting method based on genetic algorithms, and its complete execution steps are as follows: Step 1: The central processing unit acquires the object identification data of all 300 participants and the spatial attribute data of 300 seats, including the social attraction coefficient and physical constraint coefficient of each participant, and the basic seat evaluation value of each seat.

[0074] Step 2: The central processing unit incorporates all 300 attendees into the seating algorithm at once, generating an initial population in a random manner. Each individual in the population represents a complete seating allocation scheme (i.e., a one-to-one mapping relationship between 300 attendees and 300 seats). The initial population size is set to 200 individuals.

[0075] Step three: The central processing unit (CPU) calculates the fitness value of each individual in the population, with the optimization objectives of maximizing social proximity and minimizing physical incongruity. The fitness value is calculated in the opposite way to the overall seating arrangement evaluation value in step three of this invention—in this invention, a lower overall seating arrangement evaluation value indicates a better solution, while in existing methods, a higher fitness value indicates a better solution. The CPU calculates the fitness value of each individual by weighting and summing the social attraction evaluation value, physical constraint evaluation value, and basic seating evaluation value according to preset weights, and then takes the reciprocal of the sum.

[0076] Step 4: The central processing unit generates a new generation of individuals through selection (using roulette wheel selection, where individuals with higher fitness values ​​have a greater probability of being selected), crossover (using two-point crossover with a crossover probability of 0.8), and mutation (using exchange mutation with a mutation probability of 0.05). The size of the new generation of individuals is maintained at 200.

[0077] Step 5: The central processing unit repeats steps 3 and 4 until the preset maximum number of iterations of 200 generations is reached, or the fitness value of the best individual in the population no longer increases after 50 consecutive generations, at which point the iteration stops.

[0078] Step six: The central processing unit selects the seat allocation scheme corresponding to the individual with the highest fitness value from the final population as the optimal seat allocation scheme output by the existing method.

[0079] The method of this invention employs steps one through eight identical to those in Embodiment 2 to optimize seating arrangement. Specifically, step one involves acquiring participant identification data and calculating the social attraction coefficient and physical constraint coefficient; step two involves acquiring seat space attribute data and calculating the basic seat evaluation value; step three involves constructing a comprehensive seating evaluation value based on already seated participants and occupied seats; step four involves calculating the marginal evaluation value increment using vacant seats as a candidate set and selecting initial seats when new participants arrive; step five involves performing local fine-tuning evaluations on the initial seats and determining the final seats; step six involves assigning new participants to the final seats and iteratively converging until the variance is no greater than a preset convergence threshold; step seven involves recording the convergence time and updating the social attraction coefficient; and step eight involves generating and sending seat control instructions.

[0080] Implementation results: In this comparative example, after the existing method and the method of this invention completed the seating arrangement for all 300 participants in the same exhibition setting, the results of various comparative indicators are as follows: Regarding the total time spent on seating arrangement calculation, the existing method (genetic algorithm) takes about 180 seconds, while the method of this invention takes about 8 seconds. This invention is 22.5 times faster than the existing method.

[0081] Regarding the average response time per person, the existing method is a one-time global seating arrangement, which requires obtaining information from all 300 participants before calculation can begin. This method is not applicable to the "response time per person" metric in streaming scenarios. The method of this invention is approximately 1.6 seconds per person, supporting streaming arrival scenarios where participants check in and enter one after another.

[0082] Regarding the overall evaluation value of seating arrangement, the existing method is 78.6, while the method of this invention is 52.3. This invention reduces the value by 33.5% compared to the existing method (a lower overall evaluation value of seating arrangement indicates a better seating arrangement scheme, including higher social proximity, higher physical adaptability, and more reasonable utilization of seat resources).

[0083] Regarding the social attraction score, the existing method is 45.2, while the method of this invention is 28.7, which is 36.5% lower than the existing method (the lower the social attraction score, the higher the proximity of the social relationship).

[0084] Regarding the physical constraint evaluation value, the existing method is 18.4, while the method of this invention is 12.6. This invention reduces the physical constraint evaluation value by 31.5% compared to the existing method (the lower the physical constraint evaluation value, the higher the physical fit of the participants with special needs).

[0085] Regarding the basic seat evaluation value, the existing method is 15.0, while the method of this invention is 11.0. The present invention reduces the value by 26.7% compared to the existing method (the lower the basic seat evaluation value, the more reasonable the utilization of seat resources).

[0086] Regarding whether streaming arrival is supported, existing methods do not support it (seat arrangement can only begin after all attendee information is obtained), while the method of this invention supports it (attendees can check in and enter the venue one after another, and the system responds in real time).

[0087] Regarding whether local fine-tuning is supported, existing methods do not support it (once the solution is output, it is fixed and cannot be adjusted), while the method of this invention supports it (it allows for very small-scale adjustments to local areas based on the already established solution).

[0088] Regarding whether closed-loop feedback is supported, existing methods do not support it (the process terminates immediately after seating arrangement, without any feedback mechanism), while the method of this invention supports it (the social gravity coefficient is adjusted in real time based on the convergence time).

[0089] Regarding whether cross-exhibition learning is supported, existing methods do not support it (seating is arranged independently for each exhibition, and historical data is not utilized), while the method of this invention supports it (weight optimization and update driven by historical data).

[0090] In terms of computational efficiency, existing methods take 180 seconds to calculate seating arrangements. This is because existing methods require obtaining information on all attendees before starting the seating calculation, and the genetic algorithm involves a large amount of iterative computation (200 generations of evolution, with fitness values ​​calculated for 200 individuals per generation, requiring a total of 40,000 fitness evaluations), resulting in enormous computational overhead. This invention employs a streaming incremental allocation method, with an average response time of 1.6 seconds per new attendee and a total time of approximately 8 seconds (300 attendees multiplied by 1.6 seconds, each attendee divided by 60 parallel calculations), significantly improving computational efficiency compared to existing methods.

[0091] In terms of seating arrangement quality, the existing method achieves a comprehensive evaluation score of 78.6, while this invention achieves 52.3, a reduction of 33.5%. Specifically, the social attraction evaluation score of this invention is reduced by 36.5%, indicating better social proximity; the physical constraint evaluation score is reduced by 31.5%, indicating higher physical fit for attendees with special needs; and the basic seat evaluation score is reduced by 26.7%, indicating more rational utilization of seat resources. Considering all sub-indicators, this invention comprehensively outperforms the existing method in seating arrangement quality. The reason for this is that the existing method uses a global one-time seating strategy, where all attendees' seats are determined simultaneously, making dynamic adjustments based on the actual check-in order impossible; while this invention uses a streaming incremental allocation strategy, where the seat allocation for each newly arrived attendee is locally optimized based on the actual distribution of currently seated attendees. The seating arrangement dynamically evolves with the actual arrival of attendees, ultimately converging to a better global state.

[0092] Regarding scenario adaptability, existing methods do not support streaming arrival scenarios—if attendees arrive after seating arrangements are completed, existing methods cannot dynamically insert them into the existing seating plan, requiring manual arrangement or re-execution of the entire seating calculation (taking more than 180 seconds), severely impacting the on-site experience of the event. This invention, through incremental allocation and local fine-tuning mechanisms in steps four to six, can respond in real-time to the arrival of new attendees (averaging 1.6 seconds per attendee), and each adjustment only involves a small number of seats in a local area (no more than a rectangular area consisting of two rows in front and behind, and four seats on each side), without affecting the overall stability of the seating arrangement.

[0093] Regarding adaptability, existing methods set the optimization objective and weight parameters all at once before seating arrangement, making it impossible to adjust them according to actual conditions during the arrangement process, let alone learn from historical events. This invention achieves real-time correction of the social attraction coefficient through a convergence time feedback correction mechanism in step seven—when a participant frequently triggers high-cost adjustments, their social attraction coefficient is automatically lowered; simultaneously, through a cross-event knowledge transfer mechanism, it achieves cross-event optimization updates of preset weights—as historical data accumulates, seating arrangement efficiency continuously improves. These two mechanisms enable the system to continuously evolve.

[0094] Regarding the handling of special needs, existing methods, while allowing the inclusion of special needs constraints in the fitness function (such as awarding fitness points to individuals with mobility impairments by assigning them seats closer to the exit), lack a dynamic protection mechanism for participants with special needs. In subsequent iterative optimizations, the desirable seats already allocated to these individuals may be reassigned to other individuals due to population evolution. This invention protects the seats of participants with special needs through an elastic binding coefficient, ensuring they are not improperly moved during subsequent local fine-tuning. Furthermore, the elastic binding coefficient dynamically increases based on the frequency of adjustment, further enhancing the protection effect.

[0095] Compared with Examples 1-5 and Comparative Example 1, Examples 1 to 5 systematically verified the intelligent seating optimization method for exhibitions provided by the present invention from different dimensions. Example 1, using an 80-person small seminar as a scenario, fully demonstrated the entire process from steps one to eight, verifying the logic of dynamically setting the preset space range to half a row at the front and back and two seats on each side when the total number of seats is less than 100, as well as the complete link of convergence time recording and social attraction coefficient feedback correction in a single local field reconstruction, with an average allocation response time of 0.8 seconds and a 25.7% reduction in the overall seating evaluation value. Example 2, using a 500-person large-scale industry summit as a scenario, verified the logic of dynamically setting the preset space range to two rows at the front and back and four seats on each side when the total number of seats is greater than or equal to 300, as well as the system's stable processing capability under conditions of continuous and dense arrival of multiple people, with an average allocation response time of 1.6 seconds for 500 participants. Example 3 focuses on attendees with special needs, verifying the influence mechanism of the physical constraint coefficient on the basic seat evaluation value and the protective role of the elastic binding coefficient. Three attendees with mobility impairments were all assigned seats no more than 2 meters from the exit and were not moved during subsequent minor adjustments. Example 4 focuses on highly socially connected groups, verifying the quantitative calculation of social connection strength and the guiding mechanism of social attraction evaluation value on group aggregation. Five employees from the same company were ultimately assigned to adjacent seats in the same row, resulting in 23 interactions during the exhibition, significantly higher than the 8 to 12 interactions in a random seating scenario. Example 5 focuses on cross-exhibition knowledge transfer, verifying the weight optimization and update mechanism driven by the historical database. After accumulating data from 12 exhibitions, the average convergence time for the 13th exhibition decreased from 2.8 rounds to 2.4 rounds, a reduction of 14.3%. These five examples correspond to five dimensions: small-scale scenarios, large-scale scenarios, special needs adaptation, social group aggregation, and system self-evolution. They are independent yet complementary, collectively constituting a verification system for the completeness and effectiveness of the technical solution of this invention.

[0096] The comparative study uses a global seating arrangement method based on a genetic algorithm as a reference, comparing it with the present invention in the same exhibition scenario (300 people, 300 seats). In the comparative study, the existing method requires obtaining information on all attendees before calculation can begin; the genetic algorithm requires 200 generations of evolution, with 200 individuals per generation, totaling 40,000 fitness assessments, and the total seating arrangement time is 180 seconds. The present invention uses a streaming incremental allocation method, with an average response time of 1.6 seconds per attendee and a total time of approximately 8 seconds, improving computational efficiency by 22.5 times. Regarding seating arrangement quality, the existing method achieves a comprehensive evaluation score of 78.6, while the present invention achieves 52.3, a reduction of 33.5%. Specifically, the social attraction evaluation score is reduced by 36.5%, the physical constraint evaluation score by 31.5%, and the seat basis evaluation score by 26.7%, demonstrating superior performance across all three sub-indicators. Regarding scenario adaptability, existing methods do not support streaming arrival and local fine-tuning; when attendees arrive late, they can only be manually arranged or the global calculation must be recalculated. This invention achieves real-time incremental allocation and local fine-tuning through steps four to six, with each adjustment involving only a local area. Regarding adaptability, existing methods cannot adjust the optimization target and weight parameters once set. This invention achieves real-time feedback correction of the social attraction coefficient through step seven, and continuously optimizes the weights through cross-exhibition knowledge transfer, possessing self-evolutionary capabilities. Regarding handling special needs, existing methods lack dynamic protection mechanisms. This invention effectively protects the seating of attendees with special needs through elastic binding coefficients. Five embodiments and comparative examples, from both positive verification and negative reference perspectives, jointly demonstrate that this invention has significant technical advancements in five aspects: computational efficiency, seating quality, scenario adaptability, adaptability, and handling special needs. Embodiments one to five address "what this invention can achieve," while the comparative examples address "how this invention is better than existing technologies." The combination of these two forms a complete system for demonstrating the technical effectiveness.

[0097] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of protection of the claims.

[0098] In conclusion, the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing intelligent seating arrangements at exhibitions, characterized in that, Includes the following steps: Step 1: Obtain the object identification data of the participants and calculate the social attraction coefficient and physical constraint coefficient for each participant; Step 2: Obtain the spatial attribute data of the seats and calculate the basic evaluation value of each seat; Step 3: Construct a comprehensive evaluation value for seating arrangement based on the seated personnel and occupied seats. The comprehensive evaluation value for seating arrangement is composed of the social attraction evaluation value, the physical constraint evaluation value, and the basic seat evaluation value. Step 4: When new participants arrive, use the available seats as the candidate set, calculate the marginal evaluation value increment, and select the L candidate seats with the smallest increment as the initial selection seats. Step 5: Perform local fine-tuning evaluation on each initially selected seat, allowing those already seated in the surrounding area to move within the preset space, and determine the final seat based on the reduction in local evaluation value; Step six: Assign new participants to their final seats. If the variance of local evaluation values ​​is greater than the preset convergence threshold, repeat step five until convergence is achieved. Step 7: Record the convergence time of each seat allocation adjustment, and update the social attraction coefficient of the participants involved in the adjustment based on the convergence time. Step 8: Generate a seat control command and send it to the seat binding terminal to drive the seat binding terminal to display the corresponding attendee information.

2. The intelligent seating arrangement optimization method for exhibitions according to claim 1, characterized in that, The object identification data includes identity information, industry domain labels, interactive object identifiers from historical participation records, cooperation intention markers, special needs information, and estimated arrival time; The identity verification information includes name, ID number, and / or facial feature code; The special needs information includes mobility assistance needs identifiers and / or audiovisual assistance needs identifiers.

3. The intelligent seating arrangement optimization method for exhibitions according to claim 1, characterized in that, The spatial attribute data includes two-dimensional plane coordinates, field of view obstruction coefficient, sound coverage quality, distance from the podium, and distance from the nearest exit; The field of view occlusion coefficient is a value that is greater than or equal to 0 and less than or equal to 1; The sound coverage quality is a value greater than or equal to 0 and less than or equal to 1.

4. The intelligent seating arrangement optimization method for exhibitions according to claim 1, characterized in that, The social attraction coefficient is calculated as follows: The job level of the participants is mapped to the industry influence index, with the mapping rule being that ordinary participants are mapped to 1, department heads to 2, company leaders to 3, and industry experts to 4. Multiply the industry influence index by the first preset weight to obtain the first product term; The sum of the social connection strengths of the participant with all other participants is multiplied by the second preset weight to obtain the second product term, where the value of the social connection strength between the participant and any other participant is: 1 if there is an interaction record with the other participant in the historical meeting records, and 0 otherwise. Multiply the participant's cooperation intention activity level by the third preset weight to obtain the third product term, where the cooperation intention activity level is the number of cooperation intention tags marked as "high" divided by the total number of tags; Add the first product term, the second product term, and the third product term to obtain the social attraction coefficient of the participant; The first preset weight, the second preset weight, and the third preset weight are all greater than 0, and the sum of the three is equal to 1.

5. The intelligent seating arrangement optimization method for exhibitions according to claim 3, characterized in that, The calculation method for the basic seat evaluation value is as follows: Multiply the difference between 1 and the field of view occlusion coefficient by the fourth preset weight to obtain the fourth product term; Subtract the difference in sound coverage quality from 1 and multiply it by the fifth preset weight to obtain the fifth product term; Multiply the normalized distance to the podium by the sixth preset weight to obtain the sixth product term; Multiply the difference between 1 and the accessibility coefficient by the seventh preset weight to obtain the seventh product term; Add the fourth, fifth, sixth, and seventh product terms together to obtain the basic evaluation value of the seat. The accessibility coefficient is set as follows: if the distance to the nearest exit is less than or equal to 2 meters, it is 1; if it is greater than 2 meters but less than or equal to 5 meters, it is 0.5; if it is greater than 5 meters, it is 0. The fourth preset weight, the fifth preset weight, the sixth preset weight, and the seventh preset weight are all greater than 0, and their sum equals 1.

6. The intelligent seating arrangement optimization method for exhibitions according to claim 1, characterized in that, The social attraction rating is calculated as follows: For all seated individuals, randomly select two different individuals, multiply their social attraction coefficients, divide by the square of the distance between their seats, and sum the results of all pairwise combinations. The distance between seats is a normalized value obtained by dividing the Euclidean distance in planar coordinates by the maximum diagonal length of the seating area in the exhibition venue.

7. The intelligent seating arrangement optimization method for exhibitions according to claim 1, characterized in that, The physical constraint evaluation value is calculated as follows: For each person who is already seated, multiply the physical constraint coefficient by the environmental attribute value of their current seat, and sum the products of all seated persons. The environmental attribute values ​​consist of the seat's field of vision obstruction coefficient, sound coverage quality, and distance from the podium; The basic seat evaluation value is obtained by summing the basic seat evaluation values ​​of all occupied seats.

8. The intelligent seating arrangement optimization method for exhibitions according to claim 1, characterized in that, The value of L is greater than or equal to 3 and less than or equal to 10; The preset threshold value ranges from greater than or equal to 0.01 to less than or equal to 0.

1. The preset space range is a rectangular area centered on the seat being evaluated, consisting of one row in front and behind, and three seats on each side. The variance of the local evaluation value is calculated as follows: taking the final seat as the center, the rectangular area formed by one row in front and behind and three seats on each side as the local area, and calculating the variance of the comprehensive evaluation value of all occupied seats in the local area.

9. The intelligent seating arrangement optimization method for exhibitions according to claim 1, characterized in that, The specific method for updating the social attraction coefficient is as follows: Multiply the current social attraction coefficient of the participant by a coefficient, which is 1 minus a product value, where the product value is a preset learning rate multiplied by the arithmetic mean of the participant's convergence time and then divided by a preset maximum acceptable convergence time threshold. The arithmetic mean of the convergence time is the arithmetic mean of the convergence times of all seat allocation adjustments in which the participant participated. The preset learning rate is greater than 0 and less than 1, and the preset maximum acceptable convergence time threshold is a positive integer. The preset learning rate has a value range of greater than or equal to 0.01 and less than or equal to 0.1, and the preset maximum acceptable convergence time threshold has a value range of greater than or equal to 5 and less than or equal to 20.

10. The intelligent seating arrangement optimization method for exhibitions according to claim 4 or 5, characterized in that, It also includes the following steps: All seating arrangement process data for each exhibition is stored in a historical database. The seating arrangement process data includes object identification data, initial and final evaluation values ​​for each seat allocation adjustment, convergence time for each seat allocation adjustment, and the identification of the participants and their seats involved. Before the execution of subsequent exhibitions, the data in the historical database is used as training samples, and the first to seventh preset weights are optimized and updated with the goal of minimizing the average convergence time. The optimized weights are then used as calculation parameters for subsequent exhibitions.

Images