Spatial evaluation method and system based on multi-dimensional perception information

Abstract

The invention relates to a space evaluation method and system based on multi-dimensional perception information, and the method comprises the steps: collecting multi-dimensional perception data which comprise building space data, behavior data, positioning data, physical environment indexes and survey data; performing space-time unification on the multi-dimensional sensing data, and performing data denoising and missing value processing to obtain a first data set; calculating a core evaluation index through the first data set according to a preset index library; the core evaluation indexes comprise a moving line perception index, a behavior perception index, a subjective psychological index and the physical environment index; and constructing a neural network evaluation model, and predicting the satisfaction degree of the to-be-tested space after training through the core evaluation indexes. By adopting the technical scheme of the invention, the evaluation timeliness and reliability can be improved by intelligently collecting multi-dimensional data, determining indexes based on behaviors and using a neural network model, and an accurate basis is provided for space optimization.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of architecture, and in particular to a space evaluation method and system based on multi-dimensional perception information. BACKGROUND

[0002] In the field of architecture, reasonable planning and optimization of space are crucial for improving use efficiency and meeting user needs. With the increasing emphasis on space experience, post-occupancy evaluation (POE) has become a key means to measure the rationality and functionality of space design. Traditional POE mainly focuses on space function, environmental comfort, facility satisfaction, etc. Through multi-source data collection, index system construction and result analysis, it provides basis for space optimization. Among them, data collection covers user subjective feedback and objective physical environment parameters, index system is constructed by referring to relevant standards, and analysis relies on statistics and measurement methods, aiming to explore the correlation between design parameters and user experience and propose improvement suggestions.

[0003] In the prior art, the experience of personnel in the space is usually collected by manual methods such as questionnaire survey, interview, field observation and industry expert evaluation. This method is inefficient and has a long evaluation period, resulting in poor timeliness of the results. Subjective data is also affected by individual psychological feelings and expert experience, reducing the reliability of evaluation indicators. In addition, the existing indicators focus on architect design points as the core, and each dimension is evaluated independently, which leads to a disconnection with the actual behavior of users and ignores the coupling relationship, making it impossible to reveal comprehensive space defects. At the same time, traditional methods such as Delphi method and analytic hierarchy process are usually used in evaluation methods, which rely on subjective experience, resulting in fuzzy qualitative evaluation results and difficulty in accurately quantifying space characteristics.

[0004] It should be noted that the information disclosed in the above background section is only used to strengthen the understanding of the background of the present disclosure, and therefore can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0005] To solve the above problems, the present application provides a space evaluation method and system based on multi-dimensional perception information, which can improve the evaluation timeliness and reliability by intelligently collecting multi-dimensional data, based on behavior indicators, and using a neural network model, and provide accurate basis for space optimization.

[0006] To achieve the purpose of the present application, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a space evaluation method based on multi-dimensional perception information, comprising:

[0008] Collect multi-dimensional sensing data, which includes building space data, behavioral data, location data, physical environment indicators, and survey data;

[0009] The multidimensional sensing data is spatiotemporally unified, and data denoising and missing value processing are performed to obtain the first dataset;

[0010] Based on a preset indicator library, core evaluation indicators are calculated using the first dataset; the core evaluation indicators include movement perception indicators, behavioral perception indicators, subjective psychological indicators, and physical environment indicators.

[0011] A neural network evaluation model is constructed, and the satisfaction level of the test space is predicted after training with the core evaluation indicators.

[0012] Secondly, this application also provides a spatial evaluation system based on multidimensional sensing information, used to execute the above-described spatial evaluation method based on multidimensional sensing information, the system comprising:

[0013] The data acquisition module is used to collect multidimensional sensing data, which includes building space data, behavioral data, location data, physical environment indicators, and survey data.

[0014] The data processing module is used to perform spatiotemporal unification on the multidimensional sensing data, and to perform data denoising and missing value processing to obtain the first dataset;

[0015] The indicator evaluation module is used to calculate core evaluation indicators based on a preset indicator library and the first dataset; the core evaluation indicators include movement perception indicators, behavior perception indicators, subjective psychological indicators and physical environment indicators.

[0016] The satisfaction prediction module is used to construct a neural network evaluation model and predict the satisfaction level of the test space after training with the core evaluation indicators.

[0017] The technical solution provided in this application may include the following beneficial effects:

[0018] The spatial evaluation method and system based on multidimensional perception information provided in this application can collect multidimensional data through intelligent devices, improving data collection efficiency and accuracy, thereby enhancing the timeliness and reliability of the final results. Furthermore, it obtains core evaluation indicators based on actual user behavior, breaking through the limitations of current indicators that focus on architect design points and avoiding one-sided evaluation. At the same time, it uses a neural network evaluation model to replace traditional methods that rely on subjective experience, avoiding the influence of subjective experience on the final results and providing accurate basis for spatial optimization.

[0019] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Obviously, the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0021] Figure 1 A flowchart illustrating a spatial evaluation method based on multidimensional sensing information provided in an embodiment of this application;

[0022] Figure 2 A flowchart illustrating step S200 of a spatial evaluation method based on multidimensional sensing information provided in an embodiment of this application;

[0023] Figure 3 A flowchart illustrating step S400 of a spatial evaluation method based on multidimensional sensing information provided in an embodiment of this application;

[0024] Figure 4 A schematic diagram illustrating the mapping relationship between evaluation indicators and multidimensional sensing data for a spatial evaluation method based on multidimensional sensing information, provided in an embodiment of this application;

[0025] Figure 5 A schematic diagram of a neural network spatial evaluation and prediction model for a spatial evaluation system based on multidimensional perception information, provided in an embodiment of this application;

[0026] Figure 6 This application provides a spatial evaluation method based on multi-dimensional perception information, which uses a heatmap to visualize behavioral perception indicators.

[0027] Figure 7 A schematic diagram illustrating the influence relationship between various indicators and satisfaction results of a spatial evaluation method based on multidimensional perception information provided in this application embodiment;

[0028] Figure 8 This is a schematic diagram of the structure of a spatial evaluation system based on multidimensional perception information, provided in an embodiment of this application. Detailed Implementation

[0029] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0030] This example implementation first provides a spatial evaluation method based on multi-dimensional perceptual information. (Reference) Figure 1 As shown, the spatial evaluation method based on multidimensional perceptual information may include the following steps:

[0031] Step S100: Collect multi-dimensional sensing data, which includes building space data, behavioral data, location data, physical environment indicators, and survey data.

[0032] Step S200: Perform spatiotemporal unification on the multidimensional sensing data, and perform data denoising and missing value processing to obtain the first dataset.

[0033] Step S300: Calculate core evaluation indicators using the first dataset based on a preset indicator library; the core evaluation indicators include movement perception indicators, behavior perception indicators, subjective psychological indicators, and the physical environment indicators.

[0034] Step S400: Construct a neural network evaluation model and predict the satisfaction level of the test space after training with the core evaluation indicators.

[0035] Below, we will refer to Figures 2 to 7 The steps of the spatial evaluation method based on multidimensional perception information described in this example embodiment will be explained in more detail.

[0036] In step S100, multidimensional sensing data is collected, including building space data, behavioral data, visitor data, and physical environment indicators.

[0037] In one possible implementation, the building space data includes: space area, ideal circulation length, and functional area area; the behavioral data includes: behavior type number, behavior personnel ID, behavior occurrence time, and behavior occurrence coordinates; the location data includes: location personnel ID, location time, and location coordinates; the physical environment indicators include: temperature, humidity, light intensity, noise intensity, PM2.5, CO2 concentration, TVOC concentration, wind direction, and wind speed; and the survey data includes: survey personnel ID and satisfaction score.

[0038] It is understandable that location data can be used to calculate the number of times a spatial unit is accessed, and the area of ​​a functional area includes the number of each functional partition; behavioral data can be used to calculate the number of times a behavior occurs, the duration of a single behavior, and the average duration of a single behavior; location data can also be used to calculate the number of people; the physical environment indicators also have corresponding collection times and sampling point coordinates.

[0039] The behavioral data is collected through behavioral mapping and video surveillance systems; the location data is collected through indoor positioning systems deployed within the building; the survey data is collected through satisfaction collection devices deployed within the building; the physical environment data is collected through environmental sensors deployed within the building; and the building space data is collected through laser point cloud modeling and spatial syntax parameters.

[0040] Understandably, combining behavior mapping with video surveillance systems can capture objective behavioral processes through video and use behavior mapping to record and classify behaviors in a structured way, ensuring the integrity and accuracy of behavioral data. Indoor positioning systems can accurately obtain real-time coordinates and movement trajectories of personnel, while satisfaction collection devices directly collect subjective evaluations. The combination of the two achieves the fusion of objective and subjective information in tourist data. Environmental sensors can continuously and automatically monitor physical environmental parameters, ensuring the continuity and timeliness of data, and intelligent robots can be used for data collection. Laser point cloud modeling can accurately obtain spatial geometric information, while spatial syntax parameters can analyze spatial structure from the perspective of topological relationships.

[0041] In step S200, the multidimensional sensing data is spatiotemporally unified, and data denoising and missing value processing are performed to obtain the first dataset.

[0042] In one possible implementation, step S200 may further include the following sub-steps:

[0043] In step S210, the timestamps and spatial coordinates of the multidimensional sensing data are unified to obtain the first spatiotemporal unified data.

[0044] It is important to note that the unification of timestamps and spatial coordinates is a prerequisite for achieving multi-source data correlation analysis. Different types of sensory data may be based on different time recording standards and spatial coordinate systems. Through unified processing, data such as movement patterns, behaviors, and environments can be correlated within the same spatiotemporal framework, providing a consistent data foundation for subsequent indicator calculations and model training.

[0045] In step S220, the error data in the first spatiotemporal unified data is processed by mean filtering or median filtering to obtain the second spatiotemporal unified data.

[0046] It should be noted that mean filtering is suitable for processing stationary data with relatively low random noise, smoothing errors by calculating the average value of data within a certain window; median filtering is more effective at handling impulse noise (such as sudden abnormal values ​​from sensors), eliminating extreme errors by taking the median value. The choice between the two methods should be determined based on the actual characteristics of the data noise, in order to retain the most effective information.

[0047] In step S230, the missing data of the second spatiotemporal unified data are filled in by interpolation or data fitting to obtain the third spatiotemporal unified data.

[0048] The interpolation method includes linear interpolation, polynomial interpolation, or nearest neighbor interpolation.

[0049] Understandably, the application of interpolation and data fitting methods needs to be considered in conjunction with the degree and distribution characteristics of the missing data. For a small number of scattered missing values, linear interpolation or nearest neighbor interpolation can quickly fill in the missing values; for continuously missing or complexly distributed missing data, multinomial interpolation or data fitting (such as trend models based on adjacent time periods / regions) can more accurately restore the data trend and avoid analytical bias caused by missing values.

[0050] In step S240, the third spatiotemporal unified data is subjected to data standardization and normalization processing to obtain the first dataset.

[0051] Understandably, standardization, such as Z-score standardization, can eliminate differences in data units, enabling indicators with different units, such as temperature, humidity, and population density, to be analyzed in the same model. Normalization, such as scaling the [0,1] interval, can enhance data stability, avoid the model becoming overly sensitive to high-value indicators due to excessive differences in numerical ranges, and ensure the reliability of subsequent indicator calculations and model training.

[0052] In step S300, core evaluation indicators are calculated using the first dataset based on a preset indicator library; the core evaluation indicators include movement perception indicators, behavior perception indicators, subjective psychological indicators, and physical environment indicators.

[0053] In one possible implementation, the movement perception indicators include visit frequency, trajectory length, space usage time, movement speed, space dwell time, spatial clustering degree, spatial personnel density, spatial openness, trajectory topological connectivity, and trajectory length difference coefficient; the behavior perception indicators include the number of people engaging in a behavior, the duration of the behavior, the average occurrence rate of the behavior, the degree of behavior complexity, and the space function utilization rate; and the subjective psychological indicators include satisfaction scores.

[0054] It is understandable that physical environmental indicators can be collected directly using existing mature data collection equipment, or they can be calculated after collecting raw data. For example, raw data such as temperature-measuring gas pressure, temperature-measuring metal resistance, and photocurrent can be collected and processed using calculation methods within the scope of building physics to obtain temperature, humidity, etc.

[0055] It should be noted that, as Figure 4 As shown, the method for calculating visit frequency is as follows:

[0056] Visit frequency = the total number of times a spatial unit is visited;

[0057] a) Number of visits: For each spatial unit or spatial region, the number of times personnel visit that space.

[0058] b) Analyze spatial attractiveness through visual imagery: Analyze the usage popularity of different spatial units by comparing the frequency of visits to different coordinate points.

[0059] The method for calculating the trajectory length is as follows:

[0060] Trajectory length = the sum of the distances between adjacent coordinate points on the trajectory;

[0061] a) Obtain trajectory point data: Record the movement trajectory of personnel in space to obtain the coordinates (X, Y, F) of multiple consecutive trajectory points. i ,Y i ), where i represents the index of the trajectory point. Calculate the distance between adjacent trajectory points: for each pair of adjacent trajectory points (X... i ,Y i ) and (X i+1 ,Y i+1 ), calculate the Euclidean distance between them:

[0062] b) Accumulate the distances between all adjacent points:

[0063] Add the distances between all adjacent trajectory points to obtain the total trajectory length: Where n is the total number of trajectory points.

[0064] c) Calculate the maximum / minimum / average lengths of all personnel activity trajectories.

[0065] The method for calculating space usage time is as follows:

[0066] Space usage time = the total time period during which people use the space;

[0067] a) Count the time each person enters and leaves the space.

[0068] b) Filter the time periods used by individuals within the specified space, and sum all the time periods to obtain the total usage duration:

[0069] The method for calculating movement speed is as follows:

[0070]

[0071] a) Calculate the distance and time between adjacent trajectory points: For trajectory point (X) i ,Y i ), the adjacent trajectory points are (X i-1 ,Y i-1 ) and (X i+1 ,Y i+1 ), calculate the Euclidean distance and time difference:

[0072]

[0073] d i =d i-1 +d i+1

[0074] Δt i =T i+1 -T i-1

[0075] b) Calculate the movement speed: Calculate the movement speed between each pair of adjacent trajectory points. Calculate the average of all movement speeds. Maximum and minimum values: Comparison of V i To obtain the maximum or minimum rate.

[0076] The method for calculating the duration of stay in a space is as follows:

[0077] Duration of stay in a space = ∑(departure time - entry time);

[0078] a) Collect trajectory data: Record the trajectory points (X,Y) of each person in space and their corresponding timestamps.

[0079] b) Identify points of stay or areas: Analyze trajectory data to determine which points or areas people stay in for extended periods of time; this can be done by detecting points of stay that show minimal change or remain in a particular area over a period of time.

[0080] c) Calculate the dwell time at each stop point: For each identified stop point or area, calculate the dwell time of the person, which is the departure time minus the entry time.

[0081] d) Add up the duration of all stops: Add up the duration of each person's stay at each point or area to get the total duration of stay at that point or area.

[0082] The method for calculating the degree of spatial clustering is as follows:

[0083] K-Means algorithm;

[0084] a) Determine the space and number of people: Select the space to be evaluated and record the location coordinates of the people in that space.

[0085] b) Calculate the degree of spatial clustering: Preset the number of clusters and use the K-means algorithm to perform the calculation (based on similarity measure, group similar coordinates into the same subset, so that the distance between coordinates in the same subset is minimized, while the distance between coordinates in different subsets is maximized, and judge whether the clustering result meets the requirements).

[0086] The method for calculating the density of people in a space is as follows:

[0087]

[0088] a) Determine the space and number of people: Select the space to be evaluated and record the total number of people in the space; measure or obtain the total area of ​​the space.

[0089] b) Calculate the personnel density of the space: Divide the number of people in the space by the area of ​​the space to obtain the personnel density of the space; the average / maximum / minimum values ​​can be further calculated based on the time.

[0090] The method for calculating the degree of spatial openness is as follows:

[0091]

[0092] a) Determine the spatial area and the number of people:

[0093] The space is divided into logical sub-regions based on architectural drawings, and the total number of people in the space is recorded; the total area of ​​the space is measured or obtained.

[0094] b) Calculate the openness of the space: Divide the space area by the number of people in the space to get the openness of the space; you can further calculate the average / maximum / minimum value based on the time.

[0095] The method for calculating trajectory topological connectivity is as follows:

[0096] Suppose there are N functional areas in the space, we can construct an N×N connection matrix C, where the elements of the matrix are C. ij This represents the number of connections from functional region i to functional region j.

[0097]

[0098] Among them, C ijP represents the total number of connections from region i to region j, M represents the total number of participants, and P represents the total number of connections from region i to region j. k (i→j) indicates whether there is a connection from region i to region j in the trajectory of the k-th participant (if it exists, it is marked as 1, otherwise as 0), indicator(P k (i→j) is an indicator function used to determine whether a connection exists.

[0099] Connection matrix C: Each element C of this matrix ij This represents the total number of connections made by all participants from region i to region j. Symmetry: If the connections from region i to region j are symmetrical with the connections from region j to region i (i.e., Ci...), then... ij =C ji If the connectivity between these areas is positive, it indicates that the functional connections between these areas are balanced in both directions; otherwise, the asymmetrical spatial usage pattern can be analyzed by comparing the connectivity between different areas.

[0100] a) Define functional areas: Clearly define the different functional areas in the space, such as A, B, C, etc.

[0101] b) Record personnel activity trajectories: Track and record the movement paths of personnel between different functional areas.

[0102] c) Statistical connection relationships: For each person's trajectory, count their movement between functional areas. For example, if someone moves from area A to area B, it is recorded as A→B=1.

[0103] d) Superimpose the topology connection count: Superimpose the topology connection counts of all personnel to obtain the total number of connections between each functional area.

[0104] e) Construct a topology connectivity matrix: Construct a matrix where rows and columns represent functional areas, and values ​​in the matrix represent the topology connectivity between corresponding functional areas.

[0105] The method for calculating the trajectory length difference coefficient is as follows:

[0106]

[0107] a) Obtain trajectory point data:

[0108] Record the movement trajectory of personnel in space to obtain the coordinates (X, Y, F) of multiple consecutive trajectory points. i ,Y i ), where i represents the index of the trajectory point. Calculate the distance between adjacent trajectory points: for each pair of adjacent trajectory points (X... i ,Y i ) and (X i+1 ,Y i+1 ), calculate the Euclidean distance between them:

[0109] b) Accumulate the distances between all adjacent points:

[0110] Add the distances between all adjacent trajectory points to obtain the total trajectory length:

[0111] Where n is the total number of trajectory points.

[0112] c) Determine the ideal streamline length: Calculate the ideal streamline length based on the spatial design scheme.

[0113] d) Calculate the trajectory length difference coefficient: Divide the actual trajectory length by the ideal streamline length to obtain the trajectory length difference coefficient.

[0114] e) Calculate the maximum / minimum / average of the coefficient of variation for all trajectory lengths.

[0115] The method for calculating the number of times an action occurs is as follows:

[0116]

[0117] Where n is the total number of people and the number of times the behavior occurs. i It represents the number of individuals who perform a certain action.

[0118] a) Define the preset behaviors: Determine the types of behaviors that need to be counted, such as "rest behavior" and "conversation behavior".

[0119] b) Statistical behavioral data: Calculate the total number of times each individual performs the preset behavior in the space.

[0120] The method for calculating the duration of an action is as follows:

[0121]

[0122] Where n is the total number of times the behavior occurs, and the duration of the behavior is... i Let be the duration of the i-th action.

[0123] a) Define the preset behaviors: Clarify the types of behaviors that need to be statistically analyzed.

[0124] b) Collect behavioral data: Record the start and end times of all preset behaviors that occur, and calculate the duration of each behavior.

[0125] c) Add up the duration of all behaviors: Add up the duration of all behaviors to get the total duration of the behavior.

[0126] The method for calculating the average incidence rate of the behavior is as follows:

[0127]

[0128] a) Define the preset behavior types: Clearly define the behavior categories to be counted (such as "rest", "collaboration", "passage", etc.) and ensure that all behavior types have a unique identifier in the data records (such as by behavior code or name).

[0129] b) Count the frequency of each behavior: Extract the personnel ID and behavior content fields from the behavior data. Classify the behavior according to the preset behavior type and count the total number of times each type of behavior occurs in the space (i.e., the sum of the number of times all personnel perform this behavior). Where n is the total number of people.

[0130] c) Calculate the dwell time of everyone: Add up the dwell times of everyone to get the total dwell time in the space;

[0131] d) Calculate the average incidence rate for each type of behavior: For each behavior type, use the proportion of its duration to the total time spent in the space as the average incidence rate of that behavior.

[0132] The method for calculating behavioral complexity is as follows:

[0133] The degree of behavioral complexity equals the sum of the information entropy of all types of behaviors;

[0134] a) Define the preset behavior types: specify the behavior categories to be counted (such as "rest", "collaboration", "travel"), and count the total number of categories (m).

[0135] b) Calculate the sum of the information entropy of each behavior: the sum of the information entropy of all types of behaviors. in, n is the preset number of behavior types.

[0136] c) Normalize the information entropy: The maximum value of H is logn.

[0137] The calculation method for space function utilization rate is as follows:

[0138]

[0139] Among them, the number of people who participated in the behavior is the total number of people who participated in a certain preset behavior in the functional area; the area of ​​the functional area is the total area of ​​space allocated to the behavior.

[0140] a) Determine functional areas and behavior types: Divide the space into different functional areas and define the preset behavior types in each functional area.

[0141] b) Number of times a behavior occurs: Record the total number of times a preset behavior occurs in each functional area.

[0142] c) Measure the area of ​​functional areas: Calculate or measure the total area of ​​each functional area.

[0143] d) Calculate space utilization rate: Divide the number of people participating in the activity by the area of ​​the functional zone to obtain the space utilization rate. Then, further calculate the maximum / minimum / average / variance based on demand.

[0144] In step S400, a neural network evaluation model is constructed, and the satisfaction level of the test space is predicted after training with the core evaluation index.

[0145] In one possible implementation, step S400 may further include the following sub-steps:

[0146] In step S410, a neural network evaluation model is constructed.

[0147] It should be noted that this study uses a multilayer perceptron (MLP) to construct the satisfaction model, currently with two hidden layers, each containing 10 neurons. The network structure of the MLP can be adjusted according to the evaluation scenario, and the number of hidden layers and neurons needs to be optimized experimentally. For example, for the evaluation of complex large-scale architectural spaces, the number of hidden layers or neurons can be increased to improve the model's fitting ability; for small spaces, simplifying the structure can avoid overfitting and ensure the model's generalization performance.

[0148] In step S420, the core evaluation indicators are divided into training sets and validation sets, which are then input into the neural network evaluation model to train the neural network evaluation model. Subjective psychological indicators serve as the output, while motion perception indicators, behavioral perception indicators, and physical environment indicators serve as the inputs.

[0149] It should be noted that, in one embodiment, taking a museum post-visit evaluation as an example: during training, the maximum number of training epochs was set to 1000, and the target mean squared error was set to 10⁻⁶ (the model structure and parameters can be adjusted according to actual needs). During model training, a 5-fold cross-validation strategy was used to comprehensively evaluate the model's performance. Specifically, the dataset was divided into 5 subsets, with 4 subsets used as the training set for each training iteration, and the remaining subset used as the test set to validate the model's performance. This method not only reduces the risk of overfitting but also provides higher reliability for model performance evaluation. After training, the root mean squared error for each fold was recorded in detail, and the model's root mean squared error was calculated to be 0.97. This result indicates that for a single visitor, there is a certain deviation between the predicted visit experience using evaluation metrics and the actual situation. This deviation is within an acceptable range and can be further reduced by increasing the sample size and optimizing model parameters. 5-fold cross-validation effectively balances the distribution differences between the training and test sets, ensuring the model's stability across different data subsets and providing a reliable performance reference for satisfaction prediction in subsequent practical applications.

[0150] Optionally, after step S420, the method further includes: cross-validating the mean square error of each fold, and adjusting the parameters of the neural network evaluation model using a data fusion algorithm.

[0151] Understandably, by applying data fusion algorithms (such as Kalman filtering), this prediction error can be continuously reduced, thereby improving the accuracy of the prediction.

[0152] In step S430, the preprocessed multidimensional sensing data of the space to be tested is used to obtain the motion perception index, behavior perception index, and physical environment index of the space to be tested.

[0153] It should be noted that the preprocessing of the multidimensional sensing data to be tested must use the same methods as the training data, including spatiotemporal unification, denoising, missing value imputation, and standardization / normalization. Ensuring that the format and distribution characteristics of the data to be tested are consistent with those of the training data is crucial to guaranteeing the accuracy of the model's predictions and avoiding prediction bias caused by differences in preprocessing.

[0154] In step S440, the motion perception index of the space to be tested, the behavior perception index of the space to be tested, and the physical environment index of the space to be tested are input into the trained neural network evaluation model to obtain the satisfaction score of the space to be tested.

[0155] It should be noted that, optionally, the satisfaction result output by the neural network evaluation model is a quantitative score, which can be divided into levels such as "excellent," "good," "average," and "poor" based on preset thresholds. This result not only reflects the overall experience of the space under test, but can also be used to back-analyze the influence weights of each indicator through the model, providing targeted directions for space optimization.

[0156] In one possible implementation, the spatial evaluation method based on multidimensional sensing information further includes:

[0157] Step S500: Using a partial dependency graph visualization tool, explain the impact of a single feature on the model prediction results in the neural network evaluation model, and use a spatial heatmap to visualize the index calculation results of the test space, and propose optimization suggestions.

[0158] Understandably, Partial Dependency Graphs (PDPs) can visually demonstrate the non-linear relationship between a single metric (such as trajectory length or temperature) and satisfaction, helping to identify the optimal range of the metric (such as the optimal temperature range or reasonable path length); while spatial heatmaps can associate the distribution of the metric with the physical space, accurately locate the area that needs to be optimized (such as high noise area or low dwell time area), making the optimization suggestions more actionable.

[0159] It should be noted that, as Figures 6-7 As shown, based on the spatial visualization images of the index calculation results, specific spatial diagnostic results and optimization suggestions are further proposed, taking the post-evaluation of a museum as an example:

[0160] Partial Dependence Plot (PDP) is used to interpret the impact of individual features on the model's prediction results in a neural network evaluation model, visually revealing the nonlinear relationship between features and model predictions. This method observes the marginal impact of changes in a single variable on the model output while keeping other variables fixed. The impact of different indicators on visitor satisfaction is shown in the example. Figure 7 As shown.

[0161] As shown in Figure (a), younger visitors generally have higher satisfaction with the museum, while older visitors may have relatively lower ratings due to physical limitations, interests, or other factors. To improve the experience satisfaction of older visitors, museums could consider adding facilities, services, or exhibits tailored to them. Figure (b) reveals a positive correlation between visitor satisfaction and trajectory length. Trajectory length can reflect visitor satisfaction to some extent, as a longer trajectory often means visitors can see more exhibits and activities, resulting in a richer experience. However, when the trajectory length increases to a certain critical point (approximately 1500 meters in the figure), the rate of increase in visitor satisfaction begins to slow down. This may be due to visitors feeling fatigued by excessively long trajectories, or due to a decrease in visitor interest caused by excessive information. Therefore, museums can refer to this critical point when planning visitor routes to optimize the visitor experience. Figure (c) shows that when visitors are in areas with lower satisfaction, they tend to move faster, a phenomenon more pronounced among male visitors. To guide visitors to maintain a relatively appropriate pace of movement and improve their overall satisfaction, museums can achieve this by setting up interactive exhibits and providing personalized services. Figures (d), (e), and (f) show the relationship between the duration of various behaviors and satisfaction among visitors of different age groups. The results show that museum exhibits are more attractive to younger visitors, while the satisfaction of older visitors does not change significantly with the duration of their visit, a finding consistent with the conclusion in Figure (a). Meanwhile, the ideal duration of a museum visit is approximately 40 to 60 minutes. Furthermore, appropriate interaction among visitors and participation in other leisure activities also have a positive effect on improving satisfaction. Therefore, it is recommended to rationally plan and set up dedicated leisure areas within the museum to meet visitors' leisure needs during breaks. Figure (g) shows the relationship between visitor trajectory length and satisfaction at different visiting speeds. First, it can be observed that when visitor satisfaction is high, their visiting speed is generally relatively low, a finding consistent with the conclusion in Figure (c). Furthermore, when the trajectory length exceeds 1500 meters, the change in satisfaction with trajectory length is no longer significant, a pattern consistent with the conclusion in Figure (b). Figure (h) reveals that rest periods significantly impact visitor satisfaction as the tour length increases. Specifically, both excessively short and excessively long rest periods lead to decreased satisfaction. A reasonable rest period is approximately 12 minutes, which effectively improves visitor satisfaction. Figure (i) illustrates how visitor satisfaction varies with tour speed under different tour durations. It can be seen that visitors exhibit higher satisfaction when touring at a slower pace, an observation consistent with the conclusions in Figure (g).Further investigation revealed that visitor satisfaction was highest when the visiting speed was between 15 and 20 meters per minute.

[0162] like Figure 6 As shown, a spatial heat map is used to visualize the calculation results of the indicators:

[0163] Analysis of Physical Environment Assessment Results: The temperature in the museum's exhibition area fluctuated between 22℃ and 24℃, lower than the standard indoor thermal environment value; the humidity in the exhibition hall was maintained at 60%, which is suitable. Temperatures varied between different spaces; the temperature in Hall 1 on the first floor and Exhibition Hall 4 on the second floor was higher, fluctuating around 26℃. The museum's exhibition halls utilize a combination of artificial and natural lighting, with natural lighting used in the halls and entrance areas. There was a significant difference in light intensity between areas with artificial and natural lighting. The overall light intensity inside the exhibition halls was relatively low, fluctuating between 6 Lux and 403 Lux, and visitors reported low illumination when viewing the exhibits. The main entrance hall had higher light intensity, fluctuating around 2000 Lux. The average noise level at all measuring points remained within the range of 40-50 dB, indicating minimal impact on the visitor experience.

[0164] Analysis of visitor flow evaluation results: The exhibition area is smaller than the standard requirements, which cannot fully showcase the museum's exhibits or provide diverse display formats, potentially affecting the functionality of the exhibition space and the visitor experience. Visitor dwell time in the exhibition halls is mainly influenced by two factors: hall size and exhibition content. Although the first and second exhibition halls are similar in size, visitors spent 5.5 minutes longer in the first hall than in the second, which may be related to the design (display method, spatial layout) or the attractiveness of the exhibits in the first hall. Despite being the largest hall, the third hall saw less visitor dwell time than the smaller first hall, indicating that the exhibits in the third hall were less attractive and attracted less visitor attention than in other halls. The temporary exhibition hall was visited only 7 times. This may be due to two reasons: the temporary exhibition hall is in a hidden location, and visitors left the museum without noticing it after visiting the fourth hall; or visitors experienced visitor fatigue after visiting the four halls, and although they noticed the temporary exhibition hall, they still left the museum. The third and fourth exhibition halls on the second floor of the museum were visited 14 times each, fewer than the first and second exhibition halls on the first floor. This may be related to the location of the exhibition halls and their connection to visitor flow. The total visitor flow length in first-floor lobby 1 is approximately twice that of first-floor lobby 2. This may be due to two reasons: to make full use of the circulation space, the staff placed some exhibits in the lobby; first-floor lobby 1 was visited 3 more times than first-floor lobby 2, indicating that some visitors did not visit first-floor lobby 2. The space area of ​​first-floor lobby 1 is approximately 1.2 times that of first-floor lobby 2, but the total time visitors spend in first-floor lobby 1 is approximately 4 times that of first-floor lobby 2. This may be due to three reasons: first-floor lobby 1 has a richer collection of exhibits than first-floor lobby 2; first-floor lobby 1 has a more comfortable spatial scale; and first-floor lobby 1 has rest areas. The lobby on the second floor of the museum was visited only 14 times, fewer than the lobby on the first floor. This could be due to two reasons: unclear path guidance signage and a chaotic flow of visitors, preventing them from continuing their visit to the second floor. The total time spent in Hall 2 on the first floor was less than the combined total time spent in Halls 1 on the first floor and the second floor halls. This could be due to two reasons: the smaller size of Hall 2 on the first floor reduces visitor willingness to linger; or the unsuitable scale of Hall 2 on the first floor discourages visitors from staying there. The souvenir shop and cultural and creative experience area were not open, hindering the museum's cultural dissemination function. The information desk area remained largely unused due to low visitor numbers, indicating that the service failed to effectively meet current visitor needs. The rest area effectively alleviated visitor fatigue and provided a social space. Visitor guidance information was unclear or indistinct. Some visitors entered through the back door of the second exhibition hall, viewing it in reverse order before moving on to the first, which could disrupt the flow of the exhibition content.Visitors leaving the museum directly after completing the first-floor exhibition without entering the second-floor exhibition hall reflects an unreasonable spatial organization of traffic flow between the two floors. The museum's functional distribution is also illogical; specifically, although the film screening room is located on the main axis, it remains largely unused. Visitors are easily drawn to this space due to its location, leading to a chaotic flow of visitors. Furthermore, the museum places too much emphasis on the influence of traditional residential architecture on spatial organization, neglecting the impact of spatial organization on visitor behavior.

[0165] Analysis of behavioral evaluation results: Significant individual differences exist among visitors, necessitating differentiated exhibition experiences for different visitor types. Some visitors' total activity time was significantly shorter than others, potentially indicating lower interest. The progressively decreasing viewing time in each exhibition hall suggests that the exhibits or exhibition design are less appealing in subsequent halls, failing to maintain visitor interest. The content in the second exhibition hall may be more discussion-oriented and likely to encourage interaction, while the content in the temporary exhibition halls failed to stimulate visitors' desire to engage. Some exhibition halls lacked rest areas, failing to meet visitors' need for rest. As viewing time increases, visitors require timely breaks, but the lack of rest areas in the exhibition halls may negatively impact the visitor experience.

[0166] Analysis of psychological perception evaluation results: Some visitors felt the guidance was insufficient, which may have affected their viewing experience, especially for first-time visitors. The mixed evaluations of the activity experience indicate that the museum's activities need improvement in terms of content innovation, participation, and interactivity. Attractive exhibits are a key factor in visitors' decisions to recommend the museum. A well-designed flow of visitors and a comfortable environment further enhance the overall experience. The museum performed well in terms of visitor experience, flow design, and environmental comfort.

[0167] Assessment Recommendations: Install an intelligent temperature control system, optimize lighting design, and regularly monitor and evaluate museum environmental parameters. Enhance the experience of circulation spaces, optimize flow guidance, improve the interactivity of exhibits, increase the attractiveness of temporary exhibition halls, add service functions, enhance spatial adjustability, and introduce a modular display system. Strengthen flow guidance signage, reposition the film and television hall, revitalize the film and television hall space, and enhance the attractiveness of the second-floor exhibition halls. Establish diversified visiting modes, optimize the exhibition hall atmosphere, rationally allocate rest facilities, and provide rest area signage. Improve the signage system, plan exhibition activities, and continuously optimize service facilities.

[0168] Improvement effect: Transforms the traditional fuzzy qualitative assessment results of the whole space into precise quantitative diagnosis of specific problems in the space.

[0169] Furthermore, this example embodiment also provides a spatial evaluation system based on multi-dimensional sensing information for performing the aforementioned spatial evaluation method based on multi-dimensional sensing information. (See reference...) Figure 8 As shown, the system may include a data acquisition module, a data processing module, an indicator evaluation module, and a satisfaction prediction module.

[0170] The data acquisition module is used to collect multidimensional sensing data, which includes building space data, behavioral data, location data, physical environment indicators, and survey data.

[0171] The data processing module is used to perform spatiotemporal unification on the multidimensional sensing data, and to perform data denoising and missing value processing to obtain the first dataset.

[0172] The indicator evaluation module is used to calculate core evaluation indicators based on a preset indicator library and the first dataset; the core evaluation indicators include movement perception indicators, behavior perception indicators, subjective psychological indicators and physical environment indicators.

[0173] The satisfaction prediction module is used to construct a neural network evaluation model and predict the satisfaction level of the test space after training with the core evaluation indicators.

[0174] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the appended claims.

[0175] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. This application is not limited to the exact structures described above and illustrated in the accompanying drawings, and it should not be considered that the specific implementation of this application is limited to these descriptions. For those skilled in the art, various changes and modifications made without departing from the concept of this application should be considered to fall within the protection scope of this application.

Claims

1. A spatial evaluation method based on multidimensional perceptual information, characterized in that, include: Collect multi-dimensional sensing data, which includes building space data, behavioral data, location data, physical environment indicators, and survey data; The multidimensional sensing data is spatiotemporally unified, and data denoising and missing value processing are performed to obtain the first dataset; Based on a preset indicator library, core evaluation indicators are calculated using the first dataset; the core evaluation indicators include movement perception indicators, behavioral perception indicators, subjective psychological indicators, and physical environment indicators. A neural network evaluation model is constructed, and the satisfaction level of the test space is predicted after training with the core evaluation indicators.

2. The spatial evaluation method based on multidimensional sensing information according to claim 1, characterized in that, The building space data includes: space area, ideal circulation length, and functional area area; the behavioral data includes: behavior type number, behavior personnel ID, behavior occurrence time, and behavior occurrence coordinates; the location data includes: location personnel ID, location time, and location coordinates; the physical environment indicators include: temperature, humidity, light intensity, noise intensity, PM2.5, CO2 concentration, TVOC concentration, wind direction, and wind speed; the survey data includes: survey personnel ID and satisfaction score.

3. The spatial evaluation method based on multidimensional sensing information according to claim 1, characterized in that, The behavioral data is collected using behavioral mapping and video surveillance systems; the location data is collected using indoor positioning systems deployed within the building; the survey data is collected using satisfaction collection devices deployed within the building; and the physical environment data is collected using environmental sensors deployed within the building. The architectural space data is obtained through laser point cloud modeling and spatial syntax parameter acquisition.

4. The spatial evaluation method based on multidimensional sensing information according to claim 1, characterized in that, The step of performing spatiotemporal unification on the multidimensional sensing data, and processing the data for noise reduction and missing values ​​to obtain the first dataset includes: By unifying the timestamps and spatial coordinates of multidimensional sensing data, the first spatiotemporal unified data is obtained; The error data in the first spatiotemporal unified data is processed by mean filtering or median filtering to obtain the second spatiotemporal unified data; The missing data of the second spatiotemporal unified data are filled in by interpolation or data fitting to obtain the third spatiotemporal unified data. The third spatiotemporal unified data is subjected to data standardization and normalization processing to obtain the first dataset.

5. The spatial evaluation method based on multidimensional sensing information according to claim 4, characterized in that, The interpolation method includes linear interpolation, polynomial interpolation, or nearest neighbor interpolation.

6. The spatial evaluation method based on multidimensional sensing information according to claim 1, characterized in that, The movement perception indicators include visit frequency, trajectory length, space usage time, movement speed, space dwell time, spatial clustering degree, spatial personnel density, spatial openness, trajectory topological connectivity, and trajectory length difference coefficient; the behavior perception indicators include the number of people engaging in a behavior, the duration of the behavior, the average occurrence rate of the behavior, the degree of behavior complexity, and the space function utilization rate; the subjective psychological indicators include satisfaction scores.

7. The spatial evaluation method based on multidimensional sensing information according to claim 1, characterized in that, The step of constructing a neural network evaluation model and predicting the satisfaction level of the test space after training with the core evaluation indicators includes: Construct a neural network evaluation model; The core evaluation indicators are divided into training and validation sets, which are then input into the neural network evaluation model to train the model. Subjective psychological indicators serve as the output, while motion perception indicators, behavioral perception indicators, and physical environment indicators serve as the inputs. The preprocessed multidimensional perception data of the space to be tested yields the motion perception index, behavior perception index, and physical environment index of the space to be tested. The motion perception index, behavior perception index, and physical environment index of the space to be tested are input into the trained neural network evaluation model to obtain the satisfaction score of the space to be tested.

8. The spatial evaluation method based on multidimensional sensing information according to claim 7, characterized in that, After the step of dividing the core evaluation indicators into training and validation sets, inputting them into the neural network evaluation model respectively, and training the neural network evaluation model, the method further includes: The mean squared error of each fold in cross-validation is used to adjust the parameters of the neural network evaluation model through a data fusion algorithm.

9. The spatial evaluation method based on multidimensional sensing information according to claim 1, characterized in that, Also includes: By using a partial dependency graph visualization tool, the impact of a single feature on the prediction results of a neural network evaluation model is explained. Spatial heatmaps are used to visualize the index calculation results of the test space and optimization suggestions are proposed.

10. A spatial evaluation system based on multidimensional sensing information, characterized in that, The system is used to execute the spatial evaluation method based on multidimensional sensing information as described in any one of claims 1 to 9, the system comprising: The data acquisition module is used to collect multidimensional sensing data, which includes building space data, behavioral data, location data, physical environment indicators, and survey data. The data processing module is used to perform spatiotemporal unification on the multidimensional sensing data, and to perform data denoising and missing value processing to obtain the first dataset; The indicator evaluation module is used to calculate core evaluation indicators based on a preset indicator library and the first dataset; the core evaluation indicators include movement perception indicators, behavior perception indicators, subjective psychological indicators and physical environment indicators. The satisfaction prediction module is used to construct a neural network evaluation model and predict the satisfaction level of the test space after training with the core evaluation indicators.

Images