Public facility layout method and system based on user preference and behavior simulation

Abstract

The invention provides a public facility layout method and system based on user preference and behavior simulation, and the method is technically characterized in that the method comprises the steps: obtaining the facility selection preference of a user through a narrative preference questionnaire, and describing the influence of each element on a selection behavior through a discrete selection model; a selection probability prediction model is established in combination with overall or sub-crowd parameters; simulating a facility selection pattern of a demand point through a Monte Carlo method, taking average utility maximization as a core optimization target, and adopting a genetic algorithm to realize iterative search of facility positions to obtain an optimized layout scheme meeting efficiency, fairness and multi-dimensional demands; matching relation analysis and index calculation of utility, distance, coverage rate and the like are carried out on the result, and a visual scheme is output, so that the defects that a traditional position-distribution model only considers the distance, ignores individual differences and depends on candidate points can be overcome, and a scientific and reasonable optimization tool is provided for public service facility planning.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of urban and regional planning optimization, in particular to a public facility layout method based on user preferences and behavior simulation and a system thereof. BACKGROUND

[0002] With the continuous advancement of urbanization, the rational layout of various public facilities has become an important task for urban and rural planning and public management disciplines. Scientific facility site selection not only affects the convenience of residents' travel and service fairness, but also directly relates to the efficiency of urban resource utilization and social benefits.

[0003] Currently, facility layout optimization generally uses the location-allocation (LA) model, typical representatives of which include the p-median model, the p-center model, and the maximum coverage model. Although these methods are mature, they have the following main shortcomings in practical application:

[0004] (1) The facility selection behavior is too simplified and cannot reflect reality

[0005] Traditional LA models usually simplify the complex facility selection decision-making process into the "nearest distance" rule, i.e., assuming that all demand points will only go to the nearest facility. This assumption ignores the multi-dimensional influencing factors in reality, resulting in a significant gap between the simulation results and the actual selection behavior. In fact, residents' facility selection decisions are not only affected by distance, but also by service level, price, environment, and other factors, which need to be considered comprehensively.

[0006] (2) Individual preference differences are ignored, and the results lack pertinence

[0007] Different groups of people have different selection preferences when using facilities, and the influence of various factors differs. Existing LA models do not introduce the selection preferences of facility users, only providing a unified allocation result, which cannot reflect the differentiated needs of different groups in terms of distance, service level, price, environment, etc., resulting in a lack of pertinence and adaptability of the scheme.

[0008] (3) The objective function is single, and it is difficult to measure comprehensive benefits

[0009] Traditional LA models mostly aim to minimize weighted distance or maximize coverage, without incorporating multi-dimensional factors such as price, service quality, and environment into utility calculation, nor reflecting the differences in the influence of these factors on different groups of people. The resulting scheme, although "optimal" in terms of distance, cannot fully reflect the actual satisfaction of residents and overall social benefits, and is likely to cause imbalance in resource allocation.

[0010] (4) The solution algorithm has strong dependence and limited applicability

[0011] Traditional LA model usually needs to set candidate facility locations in advance before solving, and then uses integer programming and other precise algorithms for optimization. This discrete processing method makes the result highly dependent on the setting of candidate points. If the candidate points are not reasonably selected, the optimal solution may be missed. On the other hand, integer programming has a significant increase in computational burden when facing large-scale or complex constraint problems, and has insufficient solving efficiency and scalability.

[0012] Therefore, the traditional LA model has limitations in candidate point dependence and solving method, and it is difficult to meet the flexible and diverse facility layout optimization requirements in reality. SUMMARY

[0013] Embodiments of the present application provide a public facility layout method and system based on user preferences and behavior simulation to at least achieve the technical effects of scientifically configuring and optimizing public facilities and space resources.

[0014] In order to solve the above technical problems, according to one aspect of the present application, a public facility layout method based on user preferences and behavior simulation is provided, comprising:

[0015] Collect and input narrative preference data, selection model data, demand point distribution data, existing facility distribution data, functionally related facility distribution data, and suitable construction area and prohibited construction area range data; generate a narrative preference questionnaire based on orthogonal design; estimate discrete choice model parameters using respondents' selection results, or load existing model parameters; perform behavior simulation based on the discrete choice model, calculate the selection probability of each demand point for the facility, and generate demand point-facility matching results through the Monte Carlo method; use average utility maximization as the core objective, and use genetic algorithm for iterative search to determine the spatial location of the new facility, compatible with average distance minimization, most disadvantaged distance minimization, and user-defined objective function; perform index calculation and multi-scheme comparison on the layout scheme, and generate visual display and exported results.

[0016] According to another aspect of the present application, a public facility layout system based on user preferences and behavior simulation is also claimed, which applies the aforementioned method, comprising:

[0017] A data module configured to collect and input narrative preference data, selection model data, demand point distribution data, existing facility distribution data, functionally related facility distribution data, and suitable construction area and prohibited construction area range data;

[0018] A narrative preference questionnaire generation module configured to generate a narrative preference questionnaire based on orthogonal design;

[0019] A discrete choice model setting module configured to estimate discrete choice model parameters using respondents' selection results, or load existing model parameters;

[0020] a facility selection behavior simulation module configured to perform behavior simulation according to the discrete choice model, calculate the selection probability of each demand point for a facility, and generate demand point-facility matching results through a Monte Carlo method;

[0021] a facility layout optimization module configured to take maximizing average utility as a core target, perform iterative search through a genetic algorithm, determine the spatial position of a newly added facility, and be compatible with minimizing average distance, minimizing the most unfavorable distance, and a user-defined objective function;

[0022] a scheme analysis and evaluation module configured to perform index calculation and multi-scheme comparison on a layout scheme, and generate visualized display and exported results.

[0023] According to another aspect of the present application, a computer-readable storage medium is also provided, the storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the method embodiments when running.

[0024] According to another aspect of the present application, an electronic device is also provided, comprising a memory and a processor, the memory storing a computer program, and the processor being configured to execute the computer program to perform the steps in any of the method embodiments.

[0025] The present application comprehensively uses the stated preference method, the discrete choice model, the Monte Carlo simulation, and the genetic algorithm, combines preference analysis and behavior simulation with the classic location-allocation model, realizes comprehensive modeling and efficient solution of the public facility layout problem, and has the following technical effects:

[0026] First, in terms of simulation accuracy, the Monte Carlo simulation method proposed by the present application breaks through the limitation of the traditional nearest facility allocation method which only considers spatial distance, and the population-specific simulation can comprehensively reflect individual differences in preferences, significantly enhancing the authenticity and reliability of the simulation.

[0027] Second, in terms of optimization targets, the average utility maximization method proposed by the present application can simultaneously consider distance, service level, and environment and other multi-dimensional factors, and more comprehensively reflect the overall satisfaction of residents. Through multiple calculations, under the same input conditions, the average utility maximization method proposed by the present application improves the overall utility level by 15%-25% compared with the traditional p-median model, significantly improves the degree of compliance of the layout scheme with user preferences, and thus realizes comprehensive optimization effect that is more in line with actual needs.

[0028] Again, in terms of computational implementation, the present application adopts heuristic genetic algorithm for solution, which can directly search in continuous space and obtain feasible solutions without pre-setting candidate facility locations, thus breaking through the limitation of traditional integer programming model which must rely on discrete candidate points. This feature makes the method of the present application better adapt to large-scale and complex layout problems existing in actual planning.

[0029] Finally, in terms of system applicability, the present application intuitively presents the optimization results and simulation process through the visualization and interaction functions of the presentation layer, supports multi-scheme comparison, and facilitates user interpretation and decision-making. This feature helps to improve the transparency and acceptability of the scheme in the practical application of public facility layout planning, and has good social and economic application prospects.

[0030] In summary, the present application significantly improves the authenticity of the model and the rationality of the optimization results, while improving the computational efficiency and decision support value, and can effectively solve the problems of over-simplification, single objective and strong dependence of traditional facility layout technology. BRIEF DESCRIPTION OF DRAWINGS

[0031] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the present application and serve to explain the present application. In the drawings:

[0032] Figure 1 A main flow schematic diagram provided by an embodiment of the present application is shown;

[0033] Figure 2 A system architecture diagram provided by an embodiment of the present application is shown;

[0034] Figure 3 An example of a narrative preference questionnaire for layout optimization of a nursing home provided by an embodiment of the present application is shown;

[0035] Figure 4 A basic space environment diagram for layout optimization of a nursing home provided by an embodiment of the present application is shown;

[0036] Figure 5 A layout optimization result diagram for a nursing home provided by an embodiment of the present application, which aims to maximize average utility, is shown;

[0037] Figure 6 A layout optimization result diagram for a nursing home provided by an embodiment of the present application, which aims to minimize average distance (p-median model), is shown;

[0038] Figure 7 A system architecture diagram provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0039] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the preferred embodiments of the present application. Obviously, the described embodiments are only part of, rather than all of, the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of the present application.

[0040] A public facility layout method and system based on user preference and behavior simulation are provided in the embodiments of the present application.

[0041] As shown in Figure 1 and Figure 2 , a public facility layout method based on user preference and behavior simulation comprises:

[0042] S101, collecting and inputting narrative preference data, selection model data, demand point distribution data, existing facility distribution data, functionally related facility distribution data, and buildable area and prohibited area range data;

[0043] S102, generating a narrative preference questionnaire based on orthogonal design;

[0044] S103, estimating discrete choice model parameters using the selection results of respondents, or loading existing model parameters;

[0045] S104, performing behavior simulation according to the discrete choice model, calculating the selection probability of each demand point for a facility, and generating demand point-facility matching results through the Monte Carlo method;

[0046] S105, taking average utility maximization as the core target, using a genetic algorithm for iterative search to determine the spatial location of the new facility, and compatible with average distance minimization, most disadvantaged distance minimization, and user-defined objective function;

[0047] S106, performing index calculation and multi-scheme comparison on the layout scheme, and generating visual display and exported results.

[0048] It should be noted that the workflow of the present technology comprises the following steps.

[0049] First, collect preference data. Through preliminary research, determine which elements will significantly affect the selection preference of users in the public facilities to be laid out, and determine the typical level of each element. Then, use the narrative preference questionnaire generation module of the function layer to construct a virtual selection experiment based on orthogonal design method to generate a survey questionnaire, requiring respondents to make a choice between two virtual public facilities (or choose "none"), thereby obtaining narrative preference data.

[0050] Secondly, estimate the preference model. The discrete choice model setting module of the function layer is used to model the collected narrative preference data, and a multinomial Logit model is used as the specific form. The dependent variable of the model is the selection result of the virtual experiment, and the independent variables are the preference factors and their levels. The estimated model parameters reflect the influence direction and degree of each factor on the selection behavior, and can be used for explanation and prediction. Through this model, the probability of selecting any facility by any demand point under any facility layout scheme can be predicted. If the selection model parameters are directly obtained through other means, the selection model data can be directly input without the foregoing steps.

[0051] Then, set the environmental inputs and boundary conditions. The demand point distribution data, existing facility distribution data, and functionally associated facility distribution data of the data layer are loaded in the system as the environmental inputs for simulation and optimization. The factors involved in the preference model can be divided into two categories: one is the inherent attribute of the facility, which should be set in the existing facility; the other is the spatial relationship attribute with other facilities, which should be reflected through the input of functionally associated facilities. At the same time, the boundary conditions for simulation and optimization are set in the representation layer, including the number of new facilities, and optionally the range of buildable area and the range of prohibited construction area; if the range of buildable area is not set, the global spatial range is the default construction area, and if the range of prohibited construction area is not set, there is no prohibited construction area.

[0052] Next, simulate and optimize the facility layout. The facility selection behavior simulation module of the function layer is used to calculate the selection probability of each demand point for different facilities according to the discrete choice model, and the facility selection results of each demand point are obtained through Monte Carlo simulation. On this basis, the average distance, average utility, and other indicators are calculated and used as the target basis for facility layout optimization to form the fitness evaluation function for optimization search.

[0053] The present application takes the maximization of average utility as the core optimization goal, which can comprehensively reflect the satisfaction of users in distance, service level, environment and other preference elements, and is the most comprehensive measurement of real demand. At the same time, the system is also compatible with traditional optimization goals, such as p-median (minimum average distance) and maximum coverage rate, so as to compare with existing methods. Further, the system supports customizing the objective function, and users can flexibly combine multiple indicators such as utility and distance, or even introduce other evaluation elements, so as to build an optimization goal that is more in line with specific planning needs. Since the optimization problem is difficult to be efficiently solved by exact algorithms in the case of large solution space and complex constraints, the present application uses a heuristic genetic algorithm to realize optimization search, thereby improving the calculation efficiency while ensuring the result quality. The population is composed of a series of candidate layout schemes, and the schemes that perform better in the iteration process have a higher probability of entering the next round, so as to gradually evolve until convergence, and obtain a facility layout scheme that meets the optimization goal.

[0054] In an embodiment, the step of S102 specifically comprises:

[0055] The generation of the preference questionnaire adopts an orthogonal design method to combine the preference elements and their typical levels, and uses an orthogonal table to screen out representative questionnaire schemes.

[0056] Among them, the preference elements include at least two categories: one is the inherent attribute of the facility itself, including location condition, service level, cost, and environmental quality; the other is the spatial relationship attribute, reflecting the spatial accessibility and proximity of the facility to other functional associated facilities.

[0057] In an embodiment, the step of S103 is further configured to:

[0058] S1, the discrete choice model is a multinomial Logit model, and its utility function includes a deterministic part composed of observable elements and a random disturbance item subject to a Gumbel distribution;

[0059] S2, the model obtains parameters through maximum likelihood estimation, and the parameters are estimated based on the overall sample or estimated according to different groups;

[0060] S3, the parameter estimation result is used to explain the direction and influence degree of each element on the selection behavior, or the selection probability of individuals at demand points to each candidate facility is outputted to predict the potential user selection result under any facility layout scheme;

[0061] Among them, the different groups include: population characteristics, socio-economic characteristics or behavior characteristics.

[0062] In an embodiment, the step of S104 is further configured to:

[0063] S1, for each individual of each demand point, calculate its selection probability among facilities according to the discrete choice model, and divide the interval [0, 1] into several non-overlapping sub-intervals according to the probability values, and the length of the sub-interval is equal to the selection probability of the corresponding facility;

[0064] S2, generate a random number in the interval [0, 1] for each individual of each demand point, and compare the random number with the cumulative probability interval to determine the selection result of the individual;

[0065] S3, form a demand point-facility matching pattern by repeating the random sampling process, and obtain a stable statistical distribution after multiple simulation runs;

[0066] S4, calculate indexes based on the simulation results to evaluate the efficiency and fairness of the facility layout scheme, wherein the indexes include: average utility, average travel distance, worst distance, service coverage, utility standard deviation and distance standard deviation;

[0067] S5, the Monte Carlo simulation is uniformly performed under the overall model, or independently performed under the sub-population model.

[0068] In an embodiment, the step S105 comprises:

[0069] S1, the planar coordinates (x, y) of the new facility are encoded as the gene units of the chromosome, each candidate solution is composed of a plurality of coordinate pairs to represent a set of facility locations, and a population is formed by random generation in the initialization stage;

[0070] S2, determine whether each facility location is located within the suitable area and avoids falling into the forbidden area; after the population is generated, the service distribution and user selection result corresponding to each candidate solution are calculated to obtain the average utility value or other optimization objective function value as the fitness;

[0071] S3, in the iteration process, the genetic algorithm preferentially retains candidate solutions with higher fitness to enter the next generation through selection operation, exchanges part of the facility locations of different candidate solutions to generate new solution combinations through crossover operation, and applies random disturbance to some facility coordinates to explore new solution space regions through mutation operation;

[0072] S4, and after each crossover or mutation, it is determined whether the obtained facility location enters the forbidden area, if an invalid solution appears, it is discarded and regenerated, and finally a facility layout scheme satisfying the optimization target is obtained.

[0073] In an embodiment, the step S106 comprises:

[0074] The optimal spatial location coordinates of the facility and its layout scheme, the matching relationship between the demand points and the facility, and the allocation pattern of the service object; further calculate and output a plurality of indexes for evaluating the pros and cons of the layout, the indexes at least including average utility, average travel distance, most unfavorable travel distance, service coverage rate, and standard deviation of utility or distance, and support users to visually compare and analyze the optimization results and the control scheme on the interactive interface.

[0075] According to another aspect of the present application, a public facility layout system based on user preference and behavior simulation is also claimed, applied to the aforementioned method, comprising:

[0076] A data module configured to collect and input narrative preference data, selection model data, demand point distribution data, existing facility distribution data, functionally associated facility distribution data, and buildable area and prohibited area range data;

[0077] A narrative preference questionnaire generation module configured to generate a narrative preference questionnaire based on orthogonal design;

[0078] A discrete choice model setting module configured to estimate discrete choice model parameters using the selection results of respondents, or load existing model parameters;

[0079] A facility selection behavior simulation module configured to perform behavior simulation according to the discrete choice model, calculate the selection probability of each demand point for the facility, and generate demand point-facility matching results through the Monte Carlo method;

[0080] A facility layout optimization module configured to maximize average utility as the core target, use genetic algorithm for iterative search, determine the spatial location of the new facility, and be compatible with the minimization of average distance, the minimization of most unfavorable distance, and user-defined objective functions;

[0081] A scheme analysis and evaluation module configured to perform index calculation and multi-scheme comparison on the layout scheme, and generate visual display and exported results.

[0082] It should be noted that the system is mainly through three-tier system architecture, including presentation layer, function layer and data layer. Among them, the presentation layer mainly provides the interactive interface of the user and the system, is used for receiving various input conditions set by the user, and presents the calculation and optimization results in the form of visual graphics or report forms; the function layer bears the core algorithm and business logic of the application, is responsible for completing the facility selection behavior simulation, facility layout heuristic optimization and scheme analysis and evaluation and other key functions; the data layer is used for storing and managing various data required for operation, including narrative preference data, demand point distribution data, existing facility distribution data, function-related facility distribution data and the like. Through the design of this three-tier architecture, the complete processing flow from data input, model calculation to result output can be realized, the efficiency and stability of system operation are guaranteed, and the facility layout optimization method proposed in the application is effectively supported.

[0083] Specifically, the presentation layer is used for boundary condition setting, user input, spatial layout visualization, optimization process monitoring and result export;

[0084] The function layer includes a narrative preference questionnaire generation module, a discrete choice model setting module, a facility selection behavior simulation module, a facility layout optimization module, and a scheme analysis and evaluation module.

[0085] The data layer includes narrative preference data, selection model data, demand point distribution data, existing facility distribution data, function-related facility distribution data, suitable construction area and forbidden construction area data, and result data.

[0086] The connection relationship between the modules is that the model setting and the output of the boundary conditions jointly drive the behavior simulation and the layout optimization, the optimization results are input into the scheme analysis module and are output through the presentation layer. The data content in the data layer includes narrative preference data, selection model data, demand point distribution data, existing facility distribution data, function-related facility distribution data, and suitable construction area range data.

[0087] Among them, the narrative preference data is the use preference data of the potential users of the facility collected through offline or online questionnaire survey by using the narrative preference questionnaire, and is used for estimating and calibrating the selection model parameters.

[0088] The selection model data is the value of each parameter in the discrete choice model, including the overall parameters and the sub-population parameters, and is used for describing the preferences of people when selecting facilities.

[0089] The demand point distribution data is the spatial distribution of facility users and the demand amount of each distribution unit, and is used for describing the spatial pattern of potential service objects.

[0090] The existing facility distribution data is the spatial distribution and attributes of the existing facilities of the same type as the to-be-laid-out facility, which is used to consider the complementary or competitive relationship between the new facility and the existing facility. The functionally associated facility distribution data is the spatial distribution of other types of facilities that have functional association with the to-be-laid-out facility and can affect the selection preference of users. The buildable area range data is used to define the spatial range of the buildable area and the prohibited construction area of the to-be-laid-out facility.

[0091] The function layer includes the following modules: narrative preference questionnaire generation, discrete choice model setting, facility selection behavior simulation, facility layout heuristic optimization, layout scheme analysis and evaluation.

[0092] The narrative preference questionnaire generation module includes setting preference elements and levels, generating an orthogonal design scheme, checking unreasonable problems, and generating a survey questionnaire.

[0093] The discrete choice model setting module includes estimating the overall model and estimating the sub-population model.

[0094] The facility selection behavior simulation module supports the traditional nearest facility simulation, as well as the overall Monte Carlo simulation and sub-population Monte Carlo simulation proposed in the present application. The selection results output by the module are used as the input for the subsequent evaluation index calculation and optimization fitness evaluation.

[0095] The facility layout heuristic optimization module supports the traditional p-median optimization and maximum coverage optimization, as well as the maximum utility optimization and self-defined objective function optimization proposed in the present application. The module generates alternative schemes of facility distribution within the buildable area range. For each alternative scheme, the simulation module is called to obtain the selection results, and the average utility and average distance indexes are calculated. According to the optimization target, the selected indexes are used as the fitness function values, and heuristic optimization search is performed until the optimal layout scheme is obtained.

[0096] The layout scheme analysis and evaluation module includes facility selection pattern analysis, evaluation index calculation, and multi-scheme comparison analysis, and can analyze and evaluate the optimal layout scheme, the user input scheme, and other schemes.

[0097] The functions of the presentation layer include boundary condition setting, user scheme setting, spatial layout visualization, optimization process monitoring, simulation result visualization, and result export.

[0098] According to another aspect of the present application, a computer-readable storage medium is also provided, which stores a computer program. When the computer program is executed, the steps in any of the method embodiments described above are performed.

[0099] According to another aspect of the present application, an electronic device is also provided, comprising a memory and a processor, the memory having a computer program stored therein, and the processor being configured to execute the computer program to perform the steps in any of the method embodiments described above.

[0100] Optionally, in the embodiment, the electronic device described above can be located in at least one of the network devices in the computer network.

[0101] Reference Figure 7 Optionally, in the embodiment, the processor described above can be configured to perform the following steps by the computer program:

[0102] S101, collecting and inputting narrative preference data, selection model data, demand point distribution data, existing facility distribution data, function-related facility distribution data, and data of the range of buildable area and prohibited area;

[0103] S102, generating a narrative preference questionnaire based on orthogonal design;

[0104] S103, estimating discrete choice model parameters using the selection results of respondents, or loading existing model parameters;

[0105] S104, performing behavior simulation according to the discrete choice model, calculating the selection probability of each demand point for the facility, and generating demand point-facility matching results by the Monte Carlo method;

[0106] S105, taking the maximization of average utility as the core target, using genetic algorithm for iterative search to determine the spatial location of the new facility, and compatible with the minimization of average distance, the minimization of the most disadvantaged distance, and user-defined objective function;

[0107] S106, performing index calculation and multi-scheme comparison on the layout scheme, and generating visual display and exported results.

[0108] As Figure 7 shown, the electronic device comprises a memory 402 and a processor 404, the memory 402 having a computer program stored therein, and the processor 404 being configured to execute the steps in any of the method embodiments described above by the computer program. Optionally, in the embodiment, the electronic device described above can be located in at least one of the network devices in the computer network.

[0109] Optionally, those skilled in the art can understand Figure 7 that the structure shown is only schematic, and the electronic device can also be a Mobile Internet Device (MID), a PAD, or other terminal device. Figure 7It does not limit the structure of the electronic device. For example, the electronic device can further include more or less components (such as a network interface, etc.) than those shown in Figure 7 or have a different configuration than that shown in Figure 7 .

[0110] The memory 402 can be used to store software programs and modules, such as program instructions / modules corresponding to the event occurrence probability determination method and the training method and device of the neural network model for its application in the embodiments of the present application. The processor 404 executes various functional applications and data processing by running the software programs and modules stored in the memory 402, that is, implements the event occurrence probability determination described above. The memory 402 can include a high-speed random access memory, and can further include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some examples, the memory 402 can further include a memory remotely arranged with respect to the processor 404, which can be connected to the terminal through a network. Examples of the network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. Specifically, the memory 402 can be used to store the program steps of the event occurrence probability determination, but is not limited to this.

[0111] As an example, as shown in Figure 7 , the memory 402 can include but is not limited to a data module, a narrative preference questionnaire generation module, etc. In this example, further description is not given.

[0112] Optionally, the transmission device 406 is used to receive or send data via a network. Specific examples of the network can include wired networks and wireless networks. In one example,

[0113] The transmission device 406 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices and routers through a network cable to communicate with the Internet or a local area network. In one example, the transmission device 406 is a radio frequency (Radio Frequency, RF) module, which is used to communicate with the Internet in a wireless manner.

[0114] In addition, the electronic device further includes a display 408 for displaying the operation or fault display of the rectifier side or the inverter side, and a connection bus 410 for connecting various module components in the electronic device.

[0115] Referring to Figures 3-6 , the following will be described through a specific example.

[0116] The invention uses a stated preference method to collect user preference data for public facilities. It is a hypothetical scenario-based questionnaire survey method, the basic idea of which is to infer the key factors affecting the choice behavior and their preference intensity by presenting a number of virtual facility selection schemes to the respondents and asking them to choose between different schemes with different advantages and disadvantages. This method can avoid the limitations of simply relying on existing behavior data, and is particularly suitable for new facilities or situations where there is a lack of historical observation data.

[0117] In order to improve the scientificity and efficiency of questionnaire design, the invention uses orthogonal design method to combine each preference element and its level to generate a representative set of schemes with statistical independence. In theory, if all factors and their levels are fully combined, the number of schemes will grow exponentially, which is not only difficult to implement in actual surveys, but also will increase the respondents' burden. Orthogonal design selects a limited but representative part of the combination from the vast full-factor combination by constructing an orthogonal array, the basic requirement of which is that the number of different levels of each factor appearing in the sample remains balanced, and the collocation between different factors is as independent as possible. Through this process, the main effect of each factor can be effectively estimated with less amount of questions. In other words, orthogonal design covers the potential space of "more and full" with "less and precise" combination, which reduces the complexity of the survey while ensuring the statistical significance and interpretability of the results. Its advantages are neat dispersion, uniform comparability, which can not only avoid the excessive burden of respondents, but also ensure the independent estimation of the influence of each element and the stability of the results in subsequent modeling.

[0118] Taking the facility layout of a nursing home as an example, the influencing factors determined through pre-research include location condition, distance from home, distance from general hospital, whether there is a subway station around, air quality, nursing service, and monthly expenditure, etc. For each factor, a number of typical levels are set, for example, "location condition" is set to city center, suburban area, and exurban area, and "distance from general hospital" is set to 5 minutes, 15 minutes, 30 minutes, and 45 minutes. By using orthogonal design method to combine these factors and their levels, a representative virtual nursing home scheme can be obtained, which is presented in the form of pairwise comparison in the questionnaire, and the respondents are required to choose a scheme they prefer to live in, or choose "none". Some questions in the questionnaire are shown in the table. Figure 3 The selection results of the respondents can be obtained by collecting the selection results of the respondents, which provides input for subsequent discrete choice model estimation.

[0119] The discrete choice model is used for modeling and analyzing the use preference of public facilities. The discrete choice model is based on the random utility theory, and the basic principle is that each potential user will choose the one that can bring the maximum utility when facing several different facility schemes. Specifically, the utility of facility j to individual i can be expressed as:

[0120] U ij = V ij + ε ij

[0121] In the formula, U ij represents the total utility, V ij is the deterministic utility part composed of observable elements, which is a linear combination of various preference factors, for example, the location condition of the above-mentioned old-age care institution, the distance from home, the distance from the general hospital, whether there is a subway station around, air quality, old-age care service, monthly cost and the like. ε ij is an unobservable random disturbance term, which is used to reflect the measurement error and other factors not included in the model. The core assumption of the model is that the individual will choose the facility scheme with the maximum utility, and therefore the selection probability of the facility is determined by the relative size of the utility function.

[0122] In the specific setting of the form of the utility function, the present application adopts a multinomial Logit model. The model assumes that the disturbance term is independently and identically distributed and subject to a Gumbel distribution, and thus the probability function of individual i selecting facility j can be derived as follows:

[0123]

[0124] Through the maximum likelihood estimation method, the collected descriptive preference questionnaire data can be used to estimate the parameters of the multinomial Logit model, and the regression coefficients of various factors are obtained, so as to quantitatively characterize the influence direction and influence degree of the facility selection behavior. For example, if the "distance from home" parameter is negative, it means that the closer the distance, the more likely the facility will be selected; if the "air quality" parameter is positive, it means that higher air quality can significantly improve the attractiveness of the facility.

[0125] In the present embodiment, the present application can not only estimate the overall model, but also set a grouping model for different groups of people to reflect the preference differences of different groups in the selection of public facilities. For example, in the site selection research of the old-age care institution, a model can be established for each age group (such as 60-69 years old, 70-79 years old, 80 years old and above) to compare the sensitivity differences of each group to the location condition, medical accessibility or cost level, so as to provide support for more targeted facility layout decisions.

[0126] By joint application of the above overall model and the sub-population model, the application can not only explain the existing facility selection behavior, but also predict the selection probability of different demand points and different populations to the candidate facility under any layout scheme. Based on this, the system can further take the selection probability as input to provide more refined and reliable model support for subsequent facility selection behavior simulation and facility layout optimization.

[0127] The selection behavior of public facilities is simulated based on the Monte Carlo method. The aforementioned multinomial Logit model can output the selection probability of any demand point to different facilities, but the probability itself cannot directly correspond to the actual selection result, so it needs to be converted into a specific facility matching relationship through a simulation process. The Monte Carlo method is a simulation method based on random number sampling, and its core idea is to generate a large number of random samples to convert the theoretical probability distribution into an approximate actual result, so as to obtain a selection pattern that is closer to the real behavior.

[0128] In the specific implementation process, the system first reads the probability matrix output by the multinomial Logit model, where each row corresponds to a demand point and each column corresponds to a candidate facility. Then, the system generates a random number between 0 and 1 for each individual in each demand point, and compares it with the cumulative probability distribution of the demand point to determine the final selected facility. For example, when the selection probabilities of an individual in a demand point to facilities A, B and C are 0.5, 0.3 and 0.2 respectively, the cumulative probability distribution is: [0, 0.5] corresponds to facility A, (0.5, 0.8] corresponds to facility B, and (0.8, 1.0] corresponds to facility C. If the generated random number is 0.62, the selection result of the individual is facility B. By randomly sampling all individuals in all demand points, a complete facility selection simulation result can be obtained.

[0129] In order to improve the accuracy and pertinence of the results, the application supports two ways of overall simulation and sub-population simulation. When only the overall preference model is used, all individuals use the parameters of the model for probability prediction and sampling simulation to obtain the overall simulation result; when different populations are distinguished and the model parameters are estimated respectively, different groups use their own model parameters and selection probabilities to simulate independently within the group to obtain the sub-population simulation result to reflect the preference differences of different populations in facility selection. In this way, the application can balance the grasp of the overall pattern and the presentation of the group differences, so as to maintain high rationality and adaptability under different modeling granularity.

[0130] Through multiple repeated Monte Carlo simulation, a statistical distribution of the facility selection result can be obtained, and based on this, a series of evaluation indexes reflecting the efficiency and fairness of the layout scheme are calculated, including average utility, average travel distance, most unfavorable travel distance, service coverage rate, utility standard deviation, travel distance standard deviation and the like. The simulation process can not only intuitively reflect the service distribution characteristics under the facility layout scheme, but also provide an objective basis for subsequent facility layout optimization.

[0131] A genetic algorithm is used for heuristic optimization of the public facility layout scheme. The genetic algorithm is an intelligent algorithm simulating natural evolution and group search, and the core idea is to generate and screen candidate solutions in a huge solution space through population iterative evolution, and gradually approach the optimal scheme.

[0132] In the specific application process, firstly, the representation mode of the optimization variable needs to be determined. In the application, the spatial position of the to-be-constructed facility is taken as the optimization object, and the plane coordinates (x, y) thereof are used as the gene units of the chromosome. When the number of the to-be-laid-out new facilities is p, each candidate solution (namely, a chromosome) is composed of 2p values, which correspond to the x coordinates and y coordinates of the p facilities respectively. A plurality of candidate solutions form an initial population, which is used as the search starting point of the genetic algorithm. In order to ensure the feasibility of the candidate scheme, when the initial population is randomly generated, whether the facility position falls into the forbidden construction area is automatically determined. If an illegal position appears, the solution is discarded and a new solution is generated.

[0133] In the fitness evaluation link, the facility selection behavior simulation module is called, the service distribution and user selection result of each scheme are calculated based on the discrete selection model and Monte Carlo simulation, and the target function value is further obtained. The target function can be the maximization of the average utility, or the minimization of the average distance, the minimization of the most unfavorable distance, or a multi-index combination defined by the user. The target function value is the fitness of the chromosome, which reflects the advantages and disadvantages of the chromosome in the optimization target.

[0134] In the reproduction operation, the genetic algorithm mainly includes three steps of selection, crossover and mutation. The selection operation selects individuals from the current population according to the fitness, and the probability of being selected by the individual with high fitness is greater, so as to ensure that the excellent scheme can be continued to the next generation. The crossover operation simulates the gene recombination process in natural heredity, and exchanges part of the gene fragments of two chromosomes, so as to generate new candidate solutions. In the application scenario of the application, the crossover can be represented as the combination adjustment of different facility coordinates, namely, the exchange of the positions of part of the facilities in two layout schemes. The mutation operation introduces diversity by randomly changing some gene values, for example, adding or reducing a small perturbation to the (x, y) coordinates of a facility, so as to explore a new solution space area. If the solution generated by the crossover or mutation falls into the forbidden construction area, the individual is determined to be invalid and discarded, and the system will generate a new valid individual.

[0135] After selection, crossover and mutation, the new generation population will replace the last generation population. Through this "survival of the fittest" iterative evolution mechanism, the genetic algorithm makes the better layout scheme gradually accumulate and the inferior scheme gradually eliminate. With the increase of the number of iterations, the overall fitness of the population gradually improves, and finally converges to the facility layout scheme that meets the optimization target.

[0136] Compared with traditional exact optimization methods (such as integer programming), the genetic algorithm method of the present application can still maintain high computational efficiency in the case of large solution space and complex constraints, and has strong robustness and flexibility. It not only can effectively solve complex problems such as facility layout, but also can adapt to the expansion of different objective functions and constraint conditions.

[0137] Further assume an application example in a practical scenario. It is assumed that two new nursing home facilities need to be added in Pudong New Area, Shanghai, and the optimal site is to be determined in the spatial environment shown in the figure. Different sizes of circles in the figure represent the distribution and quantity of the elderly population as demand points for nursing homes; hollow pentagons represent existing nursing homes, which are complementary and competitive with the proposed new nursing homes; cross-shaped icons and subway station icons represent general hospitals and subway stations, which are the functional facilities that need to be considered when the elderly choose a nursing home. Figure 4

[0138] Firstly, the elderly's selection preferences for nursing homes are collected through narrative preference surveys. The considered preference factors include location conditions, distance from home, distance to general hospitals, whether there is a subway station around, air quality, nursing service level, and monthly cost. Subsequently, a discrete choice model is used to model and estimate the parameters of these factors, and the results show that the elderly prefer to choose nursing homes located in the city center, close to home, close to general hospitals, with subway stations around, better air quality, higher service level, and lower cost. These conclusions are consistent with common sense, and more importantly, the parameters of the discrete choice model can quantify the relative importance of each factor. For example, the results show that when the distance from home of a nursing home decreases by 1 minute, the elderly are willing to pay 15.07 yuan more per month for it; and when the distance from a nursing home to a general hospital decreases by 1 minute, the elderly are willing to increase the distance from home by 2.25 minutes as a trade-off. Based on these parameters, the selection probability of the elderly when facing different conditions of the alternative nursing homes can be calculated, and further Monte Carlo simulation is called to predict the nursing home selection results under a given layout scheme.

[0139] Then, the genetic algorithm is used to heuristically optimize the location of the two new nursing homes with the optimization target of maximizing the average utility of the elderly. In the iterative process, the system performs Monte Carlo simulation on each candidate layout scheme and calculates the average utility as the fitness index. The final optimization result is as shown in​Figure 5 As shown in the figure, the solid pentagram in the figure is the optimal location of the new nursing home. It can be seen that the two nursing homes are close to the general hospital and the subway station, and a large number of elderly population demand points are gathered around, which is highly consistent with the factors such as distance from home, distance from general hospital, whether close to subway station and other factors that have been proved to have strong influence in the preference model, indicating that the optimization result is reasonable and consistent with the actual preferences of the elderly.

[0140] As a comparison, Figure 6 The optimization results of the traditional p-median model (minimizing the average distance) are given. Since there are already many nursing homes in the northern part of the region, the newly added nursing homes obtained by this method are all in the southern part, which preferentially serves the surrounding elderly population. According to the index calculation, the average utility of the utility maximization scheme is -0.85, the average distance from the demand point to the selected facility is 7.00 kilometers, the most unfavorable distance is 33.17 kilometers, and the market share of the newly built nursing home is 36%; while the average utility of the p-median model scheme is -1.09, the average distance is 4.21 kilometers, the most unfavorable distance is 10.86 kilometers, and the market share is 25%. The comparison results show that although the utility maximization scheme increases the average distance and the most unfavorable distance by 66% and 205% respectively, the average utility is improved by 22%, and the market share is significantly improved. It should be noted that the utility index takes into account multiple factors such as distance from home, accessibility of general hospital, accessibility of subway, location conditions, etc., which can better reflect the overall satisfaction of the elderly to the nursing home than the simple distance from home. Therefore, although the utility maximization scheme sacrifices the advantage of "close to home" to some extent, it seems to bring inconvenience, but by being close to important factors such as general hospitals and subway stations, it improves the overall utility level and the attractiveness of the facility, making it more scientific and reasonable, and more in line with the real needs of the elderly population.

[0141] Finally, it should be noted that the above examples are only used to illustrate the technical solutions of the present application, and are not limited thereto; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing examples, or make equivalent substitutions for some of the technical features; and these modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A public facility layout method based on user preferences and behavior simulation, characterized by, The method comprises the following steps: Collect and input narrative preference data, selection model data, demand point distribution data, existing facility distribution data, functionally associated facility distribution data, and buildable area and non-buildable area range data; Generate a narrative preference questionnaire based on orthogonal design; Estimate discrete choice model parameters using the selection results of respondents, or load existing model parameters; Perform behavior simulation based on the discrete choice model, calculate the selection probability of each demand point for a facility, and generate demand point-facility matching results through the Monte Carlo method; Determine the spatial location of the new facility by iteratively searching using a genetic algorithm, with the core objective of maximizing average utility, and compatible with the objectives of minimizing average distance and minimizing the worst distance, as well as user-defined objective functions; Calculate indicators and compare multiple schemes, and generate visual displays and export results.

2. The public facility layout method based on user preferences and behavior simulation according to claim 1, characterized in that, The step of generating a narrative preference questionnaire based on orthogonal design specifically includes: The preference questionnaire is generated using the orthogonal design method, which combines preference elements and their typical levels, and uses an orthogonal table to select representative questionnaire schemes. Preference elements include at least two categories: one is the inherent attributes of the facility itself, including location conditions, service level, cost, and environmental quality; the other is spatial relationship attributes, reflecting the spatial accessibility and proximity of the facility to other functionally associated facilities.

3. The method for public facility layout based on user preferences and behavior simulation according to claim 1, characterized in that, The step of estimating discrete choice model parameters using the selection results of respondents, or loading existing model parameters, is further configured as: The discrete choice model is a multinomial Logit model, and its utility function includes a deterministic part composed of observable elements and a random disturbance term following a Gumbel distribution; The model obtains parameters through maximum likelihood estimation, and the parameters are estimated based on the overall sample or stratified by different groups; The parameter estimation results are used to explain the direction and degree of influence of each element on selection behavior, or to output the selection probability of each individual located at a demand point for each candidate facility, to predict potential user selection results under any facility layout scheme; The different groups include population characteristics, socio-economic characteristics, or behavior characteristics.

4. The method for public facility layout based on user preferences and behavior simulation according to claim 1, characterized in that, The step of performing behavior simulation based on the discrete choice model, calculating the selection probability of each demand point for a facility, and generating demand point-facility matching results through the Monte Carlo method is further configured as: For each individual of each demand point, calculate the selection probability between facilities based on the discrete choice model, and divide the interval [0, 1] into several non-overlapping sub-intervals according to the probability values, with the length of the sub-interval equal to the selection probability of the corresponding facility; Generate a random number in the interval [0, 1] for each individual of each demand point, and compare the random number with the cumulative probability interval to determine the selection result of the individual; Form a demand point-facility matching pattern by repeating the random sampling process, and obtain a stable statistical distribution after multiple simulation runs; Calculate indicators based on the simulation results to evaluate the efficiency and fairness of the facility layout scheme, where the indicators include average utility, average travel distance, worst distance, service coverage rate, utility standard deviation, and distance standard deviation. The Monte Carlo simulation is performed under a general model or independently under a sub-population model.

5. The method for public facility layout based on user preferences and behavior simulation according to claim 1, characterized in that, The step of taking the maximization of average utility as a core objective, using a genetic algorithm for iterative search to determine the spatial location of the new facility, and being compatible with the minimization of average distance, the minimization of the worst distance, and a user-defined objective function, comprises: The planar coordinates (x, y) of the new facility are taken as gene units of a chromosome for coding, each candidate solution is composed of multiple coordinate pairs to represent a set of facility locations, and a population is formed by random generation in the initialization stage; Meanwhile, it is determined whether each facility location is within the range of a buildable area and avoids falling into a forbidden build area; after the population is generated, the service distribution corresponding to each candidate solution and the selection results of users are calculated to obtain the average utility value or other optimization objective function value as the fitness; In the iterative process, the genetic algorithm preferentially retains candidate solutions with higher fitness to enter the next generation through selection operation, exchanges the facility locations of different candidate solutions to generate new solution combinations through crossover operation, and applies random disturbance to some facility coordinates to explore new solution space areas through mutation operation; And after each crossover or mutation, it is re-determined whether the obtained facility location enters the forbidden build area, and if an invalid solution appears, it is discarded and re-generated, and finally a facility layout scheme that meets the optimization objective is obtained.

6. The method for public facility layout based on user preferences and behavior simulation according to claim 1, wherein, The step of performing index calculation and multi-scheme comparison on the layout scheme, and generating visual display and exported results, comprises the optimal spatial location coordinates of the facility and its layout scheme, the matching relationship between demand points and facilities, and the allocation pattern of service objects, and further calculates and outputs multiple indexes for evaluating the pros and cons of the layout, the indexes at least including average utility, average travel distance, worst travel distance, service coverage rate, and standard deviation of utility or distance, and supports users to intuitively compare and analyze the optimization results and the control scheme on the interactive interface.

7. A public facility layout system based on user preferences and behavior simulation, applied to the method of any one of the preceding claims, characterized in that: a data module configured to collect and input descriptive preference data, selection model data, demand point distribution data, existing facility distribution data, functionally related facility distribution data, and buildable area and forbidden build area range data; a descriptive preference questionnaire generation module configured to generate a descriptive preference questionnaire based on orthogonal design; a discrete choice model setting module configured to estimate discrete choice model parameters using selection results of respondents, or load existing model parameters; a facility selection behavior simulation module configured to perform behavior simulation based on the discrete choice model, calculate the selection probability of each demand point for a facility, and generate demand point-facility matching results through a Monte Carlo method; a facility layout optimization module configured to take the maximization of average utility as a core objective, use a genetic algorithm for iterative search to determine the spatial location of the new facility, and be compatible with the minimization of average distance, the minimization of the worst distance, and a user-defined objective function; a scheme analysis and evaluation module configured to perform index calculation and multi-scheme comparison on the layout scheme, and generate visual display and exported results.

8. A computer readable storage medium, characterized in that, The storage medium stores a computer program, and the computer program is configured to execute the method in any one of claims 1 to 6 when running. 9.An electronic device comprising a memory and a processor, the electronic device characterized by, The memory stores a computer program, and the processor is configured to execute the computer program to execute the method in any one of claims 1 to 6.

Images