A Cell Evaluation Method and System Based on Noise Maps

Abstract

The present invention discloses a cell evaluation method and system based on noise maps, which relate to the fields of intelligent perception and the Internet of Things. The method and system include constructing a digital twin model of a cell, dividing the cell into multiple sub-areas using a trained multi-agent reinforcement learning model, processing all sub-areas, outputting a dynamic noise distribution map of the cell, organizing the dynamic noise distribution map into a causal analysis data set, generating an initial causal relationship map through causal analysis software, collecting and identifying implicit noise sources in the initial causal relationship map, and adding them to the causal analysis data set to obtain a complete causal relationship map. The present invention analyzes the difference between the initial causal relationship map and the dynamic noise distribution map, collects implicit noise source data, and generates a complete causal relationship map after adding the causal analysis data set, thereby realizing the identification and integration of implicit noise sources.

Description

Technical Field

[0001] The present invention relates to the fields of intelligent perception and the Internet of Things, and in particular to a cell evaluation method and system based on noise maps. Background Art

[0002] Noise pollution, an increasingly prominent environmental issue in the urbanization process, has attracted widespread attention. In recent years, with the rapid development of sensor technology, geographic information systems, and artificial intelligence algorithms, noise monitoring and assessment technologies have made significant progress.

[0003] However, existing community noise assessments still have many problems. First, existing digital twin models often rely on default values for parameter calibration and fail to fully integrate actual environmental data for dynamic adjustment, resulting in deviations between simulation results and actual noise distribution. In addition, noise source analysis often focuses on explicit factors, and insufficient identification of implicit noise sources limits the comprehensiveness of causal relationships, making it difficult to accurately reflect the complex driving mechanisms and regional differences of noise. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a cell evaluation method based on noise maps to solve the problems of deviation between simulation results and actual noise distribution and insufficient identification of hidden noise sources.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a cell evaluation method based on a noise map, comprising:

[0008] Build a digital twin model of the community;

[0009] Use the trained multi-agent reinforcement learning model to divide the cell into multiple sub-areas, process all sub-areas, and output a dynamic noise distribution map of the cell;

[0010] The dynamic noise distribution map is organized into a causal analysis data set, and an initial causal relationship diagram is generated using causal analysis software;

[0011] Collect and identify the hidden noise sources in the initial causal relationship diagram and add them to the causal analysis data set to obtain a complete causal relationship diagram;

[0012] Identify key noise sources based on the causal strength in the improved causal relationship diagram, set intervention measures and simulate the effects to obtain the optimal noise reduction solution;

[0013] The dynamic noise distribution map is combined with the optimal noise reduction solution to obtain a comprehensive noise map. The noise conditions are then evaluated based on the comprehensive noise map and a cell score is generated.

[0014] As a preferred solution of the noise map-based cell evaluation method of the present invention, constructing a digital twin model of a cell specifically includes the following steps:

[0015] Collect noise data, traffic flow data and meteorological data of the community to form a multidimensional data set;

[0016] Establish and adjust the three-dimensional virtual model of the community through multidimensional datasets and historical multidimensional datasets.

[0017] As a preferred embodiment of the noise map-based cell evaluation method of the present invention, the cell is divided into multiple sub-areas using a trained multi-agent reinforcement learning model, and all sub-areas are processed to output a dynamic noise distribution map of the cell. The method specifically includes the following steps:

[0018] Use historical multidimensional datasets to train a multi-agent reinforcement learning model and divide the community into multiple sub-regions;

[0019] Each sub-region is managed by an agent, which outputs the noise prediction value of the sub-region it is responsible for;

[0020] The noise prediction values of the agents in each sub-region are combined to form a dynamic noise distribution map.

[0021] As a preferred solution of the noise map-based cell evaluation method described in the present invention, the generating of the initial causal relationship diagram by causal analysis software refers to generating the initial causal relationship diagram using pcalg based on the causal analysis data set.

[0022] As a preferred embodiment of the noise map-based cell evaluation method of the present invention, the following steps are specifically included: identifying and collecting the hidden noise sources in the initial causal relationship graph and adding them to the causal analysis data set to obtain a complete causal relationship graph:

[0023] When an implicit noise source is identified in the initial causal relationship diagram, the implicit noise source is collected;

[0024] The collected implicit noise sources are added to the causal analysis data set, and the perfect causal relationship diagram is obtained using pcalg based on the new causal analysis data set.

[0025] As a preferred embodiment of the noise map-based cell evaluation method of the present invention, the following steps are specifically included: identifying key noise sources based on the causal strength in the improved causal relationship graph, setting intervention measures and simulating the effects to obtain the optimal noise reduction solution:

[0026] Identify key noise sources by analyzing the causal strength in the refined causal relationship diagram and set intervention measures;

[0027] The intervention measures are simulated, a comprehensive evaluation score sheet is generated, and the optimal noise reduction solution is obtained.

[0028] As a preferred embodiment of the noise map-based cell evaluation method of the present invention, the step of evaluating the noise condition based on the comprehensive noise map and generating a cell score refers to analyzing the noise decibel value using a spatial analysis algorithm based on the comprehensive noise map to obtain the percentage of areas exceeding the standard.

[0029] The weighted scoring method was used to analyze the impact of the exceeding area on residents based on regional importance, and the weighted exceeding area ratio was obtained;

[0030] Based on the proportion of area exceeding the standard and the proportion of weighted area exceeding the standard, linear deduction is used to generate the community score.

[0031] In a second aspect, the present invention provides a cell evaluation system based on a noise map, comprising:

[0032] Build modules and construct a digital twin model of the community;

[0033] The prediction module uses a trained multi-agent reinforcement learning model to divide the cell into multiple sub-areas, processes all sub-areas, and outputs a dynamic noise distribution map of the cell;

[0034] The analysis module organizes the dynamic noise distribution map into a causal analysis data set and generates an initial causal relationship diagram through the causal analysis software;

[0035] The improvement module collects and identifies the hidden noise sources in the initial causal relationship graph and adds them to the causal analysis dataset to obtain the improved causal relationship graph;

[0036] The identification module identifies the key noise sources based on the causal strength in the improved causal relationship diagram, sets intervention measures and simulates the effects to obtain the optimal noise reduction solution;

[0037] The scoring module combines the dynamic noise distribution map with the optimal noise reduction solution to obtain a comprehensive noise map. It then evaluates the noise conditions based on the comprehensive noise map and generates a cell score.

[0038] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the cell evaluation method based on the noise map as described in the first aspect of the present invention is implemented.

[0039] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the cell evaluation method based on the noise map as described in the first aspect of the present invention is implemented.

[0040] The beneficial effects of the present invention are as follows: by calibrating the digital twin model with a multidimensional dataset and a historical multidimensional dataset, the precise construction of a three-dimensional virtual model of a community is achieved, and the deviation between simulation and reality is eliminated through calibration, providing a reliable virtual mirror for noise prediction and distribution analysis. The accuracy of the three-dimensional virtual model of the community is greatly improved, and the physical characteristics of the community's acoustic environment can be accurately simulated, laying a solid foundation for subsequent analysis; in addition, by analyzing the difference between the initial causal relationship diagram and the dynamic noise distribution diagram, implicit noise source data is collected, and a complete causal relationship diagram is generated after adding the causal analysis dataset, thereby realizing the identification and integration of implicit noise sources, filling the noise increment that cannot be explained by explicit factors, and providing a complete perspective for noise source analysis. It also improves the explanatory power of the causal relationship diagram, so that the description of the noise cause is changed from partial to comprehensive, avoiding analytical deviations caused by omitting key factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0042] Figure 1 This is a flowchart of the cell evaluation method based on noise map in Example 1.

[0043] Figure 2 This is a schematic diagram of the system module composition in Example 1.

[0044] Figure 3 Schematic diagram of building the digital twin model in Example 1.

[0045] Figure 4 Schematic diagram of multi-agent collaborative prediction in Example 1. DETAILED DESCRIPTION

[0046] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0047] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0048] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0049] Example 1, with reference to Figures 1 to 4 This embodiment provides a cell evaluation method based on a noise map, comprising the following steps:

[0050] S1. Build a digital twin model of the community.

[0051] The specific steps include:

[0052] S1.1. A noise sensor network is fully deployed in the target community. High-precision microphone array devices are selected and placed in key locations, such as the boundaries near main roads, near the windows of residential buildings, internal green spaces, and community activity squares. A monitoring point is set up every 50 meters to cover the main areas of the community. The noise sensor automatically collects noise data every 5 minutes, records the decibel value, and uploads it to the cloud server in real time via the 5G network. At the same time, traffic flow data, such as the number of vehicles passing through per hour, is monitored in real time through cameras, and environmental data such as wind speed, wind direction, and humidity are obtained from the local weather station. The collected noise data, traffic flow data, and meteorological data will be aggregated to the cloud server to form a multidimensional dataset containing timestamps and spatial locations.

[0053] Optimally, by deploying a noise sensor network throughout the target community and combining it with traffic flow and meteorological data, a multidimensional dataset is generated that updates in real time, providing a rich foundation for building the digital twin model. Furthermore, noise data is collected every five minutes and rapidly uploaded via the 5G network, ensuring real-time and continuous data availability. Compared to traditional static noise measurements, this high-frequency, dynamic data collection method captures instantaneous noise changes, enabling the digital twin model to truly reflect the dynamic characteristics of the community's noise environment.

[0054] S1.2. Using the generated, real-time, multidimensional dataset as input, begin building a digital twin model of the community under test. First, the collected noise data, traffic flow, wind speed, wind direction, humidity, and other environmental variables, along with the community's geographic information file (including building layout, road layout, and green space locations), are input into a digital twin platform, such as Unity or similar software. A three-dimensional virtual model of the community is created on the digital twin platform. Real-time noise data collected by sensors is mapped to corresponding spatial locations. For example, monitoring points near main roads show higher decibel values, while green spaces show lower values. This forms an initial noise distribution baseline. To ensure the accuracy of the three-dimensional virtual model, noise data, traffic flow data, and meteorological data collected over the past 24 hours are used for calibration. Parameters in the three-dimensional virtual model are adjusted. For example, one parameter in the three-dimensional virtual model is the sound attenuation rate, which represents the degree to which sound attenuates with distance as it propagates through the air. The default value might be set to 6 decibels per 100 meters. However, in data from the past 24 hours, the actual noise level at 100 meters along the main road dropped from 60 decibels to 50 decibels (a 10 decibel attenuation), while the 3D virtual model of the community only simulated a drop to 54 decibels (a 6 decibel attenuation). This difference is 4 decibels, exceeding the 2 decibel margin of error. Based on the actual data from the past 24 hours, the sound attenuation rate was adjusted to 10 decibels per 100 meters and the simulation was rerun. The results showed that the noise level at 100 meters dropped to 50 decibels, consistent with the measured value. This adjustment reflects that the air humidity or vegetation density within the community may be higher than the default value, resulting in faster sound attenuation. Furthermore, the building reflection coefficient in the 3D virtual model of the community indicates the intensity of sound reflected from buildings. The default value may be 0.8 (meaning that 80% of the sound is reflected). However, data from the past 24 hours showed that the noise level in an area near a high-rise residential building was 65 decibels, while the 3D virtual model of the community simulated it as 60 decibels, a difference of 5 decibels. Analysis found that building surfaces (such as glass curtain walls) in the real environment reflect more sound, while the 3D virtual model of the community underestimated this reflection effect. Therefore, the building reflection coefficient was increased from 0.8 to 0.95 (95% reflection). After re-simulation, the noise level in the area rose to 64 decibels, narrowing the difference from the actual value of 65 decibels to 1 decibel, meeting the error requirement. This modification and adjustment takes into account the actual reflective properties of the building materials in the community. After completing this calibration, the digital twin model was obtained, which was named the 3D Virtual Community Model for its appropriate purpose.

[0055] The authors further explained that calibration using data collected over the past 24 hours and adjusting the parameters of the 3D virtual model of the residential area further reduced the error between the simulated values and the actual measured values. This significantly improved the reliability of the 3D virtual model of the residential area, providing a solid foundation for subsequent noise prediction and evaluation.

[0056] S2. Use the trained multi-agent reinforcement learning model to divide the cell into multiple sub-areas, process all sub-areas, and output a dynamic noise distribution map of the cell.

[0057] The specific steps include:

[0058] S2.1 first train the multi-agent reinforcement learning model.

[0059] First, extract noise data, traffic flow data, and meteorological data from the past seven days from the cloud server, with timestamps and spatial locations, to form a continuous time series that records the changes in community noise and environmental variables over the past seven days. For example, the data may show that the noise level at 8 a.m. on weekdays is higher than at the same time on weekends, or that the noise level in certain areas increases significantly when the wind speed increases.

[0060] A multi-agent reinforcement learning model is used to divide a residential area into multiple subareas. For example, each building or road is considered an independent subarea, and each subarea is assigned to an agent. Each agent is then tasked with predicting the noise trend for that subarea over a period of time (e.g., two hours) based on a multi-dimensional input dataset (noise level, number of vehicles, wind speed, etc.). To help the agent learn how to predict noise trends over a period of time (e.g., two hours), the training process simulates a real-world environment, allowing the agent to learn patterns through trial and error from historical data. For example, an agent assigned to a main road subarea needs to learn to recognize the pattern of increased noise levels caused by an increase in vehicle traffic. Training is conducted in a virtual environment constructed based on the past seven days of historical data. A starting point is set, for example, at 00:00 on the first day, and historical data is played back chronologically. The agent's goal is to predict the noise level over the next two hours as accurately as possible. For example, given a noise level of 60 decibels at 8:00 AM on day 1, 200 vehicles, and a wind speed of 3 meters per second, the agent is required to predict the noise level at 10:00 AM and compare it with the actual data (e.g., 62 decibels). The training goal is to minimize the difference between the predicted and actual values, for example, to within 1.5 decibels. To incentivize the agent's learning, a reward mechanism is implemented: if the prediction is close to the actual value, the agent receives a high score; if the deviation is large, the agent receives a low score.

[0061] At the beginning of training, each agent has no knowledge of noise fluctuation patterns and can only randomly guess future noise values based on the input data (noise, number of vehicles, and wind speed). For example, the main road agent might initially guess that the noise level at 10:00 will be 65 decibels, while the actual value is 62 decibels, a deviation of 3 decibels. Feedback (such as a low score) is provided based on the size of the deviation, and the results of the attempts are recorded. As time progresses, with historical data from 00:00 on day 1 to 23:55 on day 7, the agent will repeatedly try its predictions, adjusting its strategy based on the actual data after each attempt. For example, it might discover that for every 100 additional vehicles, the noise level increases by an average of 2 decibels, or that for every 2 m / s increase in wind speed, the noise range increases by 10 meters. Through these seven days of trials, the agent gradually learns patterns in noise fluctuations, such as a 5-dB increase in noise levels when the number of vehicles surges at 8:00 a.m., or that the noise level stabilizes at night when wind speeds are low.

[0062] Preferably, the intelligent agent gradually optimizes the predicted value to be close to the actual value through repeated trial and error and feedback mechanism. The dynamic and high-precision output significantly improves the reliability and foresight of the noise distribution map, providing more accurate data support for community environmental management.

[0063] We also need to consider that the noise distribution in a community is holistic. For example, the noise from the main road may spread to the residential buildings, so the agents need to collaborate during training. Therefore, a sharing mechanism is set up so that each agent not only learns its own sub-area, but also refers to the prediction results of the neighboring areas. For example, if the main road agent predicts that the noise will increase by 5 decibels, it will notify the residential building agent and prompt it to consider the impact of the propagation. By combining the predictions of all agents, the strategy is adjusted to ensure the noise distribution of the entire community. Figure 1 For example, if the prediction for a main road is too high and the prediction for a residential building is too low, the two results are reconciled to optimize overall accuracy. Collaborative training enables the multi-agent reinforcement learning model to capture the spatial propagation patterns of noise, such as how noise affects one sub-region to another when wind direction changes.

[0064] Preferably, a collaborative mechanism between agents is used. By setting up a sharing mechanism, each agent not only focuses on its own sub-region but also refers to the prediction results of neighboring regions. Collaborative training ensures the consistency of the dynamic noise distribution map.

[0065] After the multi-agent reinforcement learning model is trained, it is put into use. Every 15 minutes, the agent in each sub-area adjusts its prediction strategy based on the latest multidimensional datasets, such as real-time noise levels and traffic flow, to predict noise trends in the area over the next two hours. For example, noise levels near main roads may increase due to an increase in vehicles, while levels in inner courtyards may remain low. During the prediction process, factors such as wind direction and building obstruction are taken into account to further refine the results. For example, when wind speed increases, noise may spread to more distant areas, or high-rise buildings may reduce noise in certain areas. The prediction results are completed collaboratively by all agents, forming a complete dynamic noise distribution map for the next two hours. The dynamic noise distribution map not only shows the noise level in each sub-area but also reflects the trend of noise spreading from high-noise sources (such as main roads) to low-noise areas (such as inner courtyards). Finally, the dynamic noise distribution map is presented as a heat map in the three-dimensional virtual model of the community, with red areas indicating excessive noise and green areas indicating quiet areas.

[0066] It is further explained that since the trained multi-agent reinforcement learning model can generate the latest dynamic noise distribution map in real time based on the latest multidimensional data set, and can also reflect the propagation trend of noise, it greatly improves the timeliness and practicality of the response.

[0067] S3. Organize the dynamic noise distribution map into a causal analysis data set, and generate an initial causal relationship diagram through causal analysis software.

[0068] The specific steps include:

[0069] Based on the dynamic noise distribution map, predicted noise data is extracted (e.g., 65 decibels on a main road at 10:00 AM). This data is then integrated with relevant environmental variables, such as traffic flow, wind speed, and the timing of resident activities, such as square dancing or construction. Environmental variables are derived from sensor networks, cameras (providing real-time traffic flow and wind speed), and community management records (providing activity times). The predicted noise data and relevant environmental variables are organized into a causal analysis dataset, ensuring that each data item is timestamped and spatially located. For example, the predicted noise value at a point on a main road at a certain time corresponds to the number of vehicles in the same time period.

[0070] S3.1 First, prepare a data set in a tabular format, for example, containing columns such as predicted noise level, traffic flow, wind speed, and resident activity times. Each row represents a data point at a specific time and location, such as 10:00 a.m., east side of the main road, predicted noise level of 65 decibels, 300 vehicles, wind speed of 5 m / s, and construction activity. Next, load the causal analysis dataset into pcalg in R (pcalg is an R package for causal structure learning that automatically infers causal relationship graphs from observational data. It uses the PC algorithm to identify causal directions between variables through conditional independence tests). The software automatically identifies each column in the table as a variable (also called a node), for example, traffic flow and noise level each become an independent variable.

[0071] After loading a causal analysis dataset, PCALG assumes that all variables may be related. For example, traffic flow may affect noise levels, wind speed may affect noise levels, or even the time residents spend outdoors. It then uses statistical tests to analyze the statistical dependence between the variables. In layman's terms, this examines whether traffic flow and noise levels consistently change together: for example, when the number of vehicles increases from 200 to 300, does the noise level increase from 60 decibels to 65 decibels? If statistical dependence between the variables recurs in the data, PCALG concludes that the two are likely related. After identifying a correlation between the variables, PCALG further determines the direction of the correlation. For example, if the data shows that increases in traffic flow always precede increases in noise levels, while increases in noise levels do not necessarily reduce traffic flow, the PCALG algorithm infers that traffic flow is the cause and noise levels are the effect, thus directing the arrow from traffic flow to noise levels. This inference is based on the temporal nature of the data and the statistical regularity of the data. For example, if wind speed changes and noise levels change, but the reverse is not true, the arrow will be directed from wind speed to noise levels. The PCAL algorithm strives to avoid loops (e.g., A affects B, which in turn affects A) to ensure a clear causal network. Furthermore, during the generation process, PCALG verifies the reliability of each causal relationship through statistical tests, for example, checking whether changes in traffic flow significantly affect noise levels. Once these tests are successful, an initial causal diagram is generated. However, this initial diagram currently only reflects the primary correlations between variables in the causal analysis dataset and does not establish a basic connection between noise and environmental variables, making it incomplete.

[0072] Furthermore, by integrating predicted noise data with multi-source environmental variables into a causal analysis dataset and using PCLG software to generate an initial causal relationship diagram, we can comprehensively capture the various factors influencing noise. PCLG analyzes the dependencies between variables through statistical tests and infers causal directions based on temporal and statistical laws. This ensures that the generated causal relationship diagram is not limited to a single variable but reflects the complex relationships between multiple variables. Furthermore, PCLG's conditional independence test verifies the reliability of each causal relationship, reducing the possibility of misjudgment.

[0073] S4. Collect and identify the implicit noise sources in the initial causal relationship graph and add them to the causal analysis data set to obtain a complete causal relationship graph.

[0074] The specific steps include:

[0075] As mentioned earlier, the initial causal relationship diagram generated only establishes the basic connection between noise and environmental variables. Now we begin to identify hidden noise sources and improve the causal analysis dataset.

[0076] First, we analyze whether the causal relationships already established in the initial causal relationship diagram fully explain the predicted results from the dynamic noise distribution map. For example, the initial causal relationship diagram may show that traffic flow and wind speed contribute to noise levels, but the predicted noise level rises by 5 decibels. However, traffic flow and wind speed can only explain 3 decibels of this increase, leaving the remaining 2 decibels unexplained. This suggests the presence of hidden factors not directly measured, such as infrasound (low-frequency sound waves below 20 Hz) caused by building structural resonance. To verify this hypothesis, we use low-frequency microphones as a tool to conduct tests in potentially affected areas, such as near high-rise buildings. The testing method involves placing low-frequency microphones at specific locations and recording continuously for a period of time (e.g., several hours). The presence of infrasound is measured, along with its intensity (decibel level) and distribution (which areas are most pronounced). For example, we found that infrasound near high-rise buildings is 30 decibels, concentrated during the morning rush hour. The newly collected infrasound data will be organized into a table format, for example, at 10:00, on the west side of the high-rise building, the infrasound level is 30 decibels, and then added to the causal analysis dataset as a new column for infrasound level. This way, the new causal analysis dataset includes explicit variables (traffic flow, wind speed, etc.) and newly added implicit variables (such as infrasound).

[0077] The causal analysis dataset containing the latent variables was then re-entered into pcalg in R, and the causal inference was rerun using the PC algorithm. Specifically, the process involves first analyzing the statistical dependencies among all variables (including the newly added infrasound level), for example, examining whether noise levels increase with increasing infrasound intensity. Causal directions are then inferred, such as infrasound resonance → noise level, and the reliability of the inferred causal direction is verified through conditional independence tests (for example, confirming whether the impact of infrasound is independent of traffic flow). Over multiple iterations, pcalg updates the causal relationship graph, for example, by adding a new node, infrasound resonance, and drawing an arrow pointing to noise level, indicating that infrasound contributes to the noise increase. Statistical tests (such as significance tests) are then used to continuously verify the accuracy of each relationship in the updated causal relationship graph. For example, combining a prediction from a dynamic noise distribution map (a 5dB noise increase on a main road in the next two hours) with the prediction, the authors can analyze the proportion of the increase: 80% from traffic flow, 10% from wind speed, and 10% from infrasound resonance. After adjustments, the final causal relationship diagram clearly shows the main sources of noise. The final causal relationship diagram (also known as the improved causal relationship diagram) includes both explicit factors (traffic flow, wind speed) and implicit influences (infrasound).

[0078] Optimally, by analyzing the initial causal relationship diagram and comparing it with the predicted results of the dynamic noise distribution map, we identified noise increases that could not be fully explained by explicit factors (such as traffic flow and wind speed) (for example, 2 dB unexplained out of 5 dB). The causal analysis dataset was then refined by supplementing it with data on implicit noise sources (such as infrasound). This updated causal analysis dataset was fed into pcalg and iteratively generated into a refined causal relationship diagram, significantly improving the diagram's explanatory power. Compared to the initial causal relationship diagram, which only reflected basic connections (e.g., traffic flow → noise level), the refined causal relationship diagram not only includes explicit factors but also explores implicit influences (such as the contribution of infrasound to noise), providing a more comprehensive description of noise sources. This comprehensive approach ensures that the causal relationship diagram accurately reflects the true driving mechanisms of community noise and avoids analytical bias caused by omitting key factors.

[0079] S5. Identify key noise sources based on the causal strength in the improved causal relationship diagram, set intervention measures and simulate the effects to obtain the optimal noise reduction solution.

[0080] The specific steps include:

[0081] Using visualization tools to analyze the refined cause-and-effect diagram, we identified the key noise sources that have the greatest impact on the neighborhood. For example, we found that traffic flow had the thickest arrow pointing to noise level, and the data showed that for every 100 vehicles added, noise levels increased by an average of 2 decibels. Wind speed and residential activity, on the other hand, had thinner arrows and smaller impacts (e.g., for every 2 meters per second increase in wind speed, noise levels increased by only 0.5 decibels). This confirmed that traffic flow was the primary cause, with wind speed and residential activity being secondary factors.

[0082] For clarification, the refined causal relationship diagram consists of nodes and arrows. For example, traffic flow pointing to noise level indicates that an increase in vehicles leads to an increase in noise. The thickness of the arrow or the value labeled (such as the correlation coefficient) reflects the strength of the connection. Furthermore, an impact threshold is set based on noise changes (e.g., a contribution to noise change > 1 dB). Traffic flow (2 dB) exceeds the impact threshold, while wind speed (0.5 dB) is below the impact threshold but still has an impact. Residential activities, like wind speed, do not exceed the impact threshold. Factors exceeding the impact threshold are also identified as primary causes (i.e., key nodes).

[0083] After confirming that traffic flow is the primary cause, with wind speed and residential activities as secondary factors, various intervention measures should be implemented. These should be tailored to key noise sources and integrated with domain knowledge. For example, considering traffic flow as the primary cause and wind speed and residential activities as secondary factors, a few examples are possible. First, installing 2-meter-high sound barriers along main roads is expected to block 50% of traffic noise. Next, speed limits should be implemented, reducing traffic speeds on main roads from 50 km / h to 40 km / h during peak hours to reduce vehicle noise. Finally, increasing green belts upstream of the wind direction (for example, by planting 3-meter-tall trees) can mitigate the wind's amplification effect on noise. These interventions should ensure coverage of both primary and secondary factors.

[0084] Ideally, visualization tools are used to analyze the refined cause-and-effect diagram to identify the key noise sources that have the greatest impact on the community. Subsequent intervention measures (such as installing noise barriers, limiting vehicle speeds, and adjusting green belts) are directly targeted at these critical nodes, incorporating domain knowledge to ensure targeted measures. Furthermore, the clear guidance provided by the refined cause-and-effect diagram allows resources to be focused on addressing key drivers (such as traffic flow), significantly improving the scientific nature and efficiency of noise reduction measures.

[0085] After obtaining the intervention measures, each intervention was entered into R. Using the refined causal diagram and Monte Carlo simulation, the causal effect simulation algorithm simulated the intervention's effects. The complete process begins by entering the intervention hypothesis, such as a 20% reduction in vehicle speed (e.g., a 20% reduction from 50 km / h to 40 km / h). The values at the traffic flow node were adjusted to reflect a 20% reduction in vehicle noise contribution. The causal effect simulation algorithm then calculated the noise change based on the arrow relationships in the refined causal diagram (e.g., traffic flow → noise level, with an intensity of 2 dB / 100 vehicles). For example, suppose the peak vehicle count is 500 and a 20% reduction to 400 vehicles results in a 2 dB noise reduction. Finally, secondary effects were analyzed, examining changes at other nodes. For example, a sound barrier might alter the wind speed path (blocking wind flow and reducing noise transmission range). Secondary effects (e.g., an additional 0.5 dB reduction in courtyard noise) were simulated using the interactions in the refined causal diagram. The causal effect simulation algorithm runs multiple times (for example, 1,000 Monte Carlo simulations), accounting for variable uncertainty (such as fluctuations in vehicle traffic) to determine the average effect of each measure. For example, a noise barrier reduces noise by 5dB overall, with a secondary impact of 0.3dB in the courtyard; a speed limit reduces noise by 3dB overall, with no secondary impact; and green belt adjustments reduce noise by 3dB overall, with the courtyard benefiting first. The effects of these noise reduction measures are then compiled and saved in a table (the preliminary evaluation results table).

[0086] The preliminary evaluation results table compares the noise reduction effect and implementation cost of each measure. For example, a noise barrier reduces noise by 5 decibels, with a construction cost of approximately 500,000 yuan (2 meters high, covering 200 meters on the east side of the main road) and an annual maintenance cost of 50,000 yuan. A speed limit reduces noise by 3 decibels, with an implementation cost of almost zero (requiring only adjustments to traffic rules and the installation of signs) and no maintenance costs. A green belt adjustment reduces noise by 3 decibels, with a planting cost of approximately 100,000 yuan (3-meter-high trees covering 100 meters upstream from the wind direction) and an annual maintenance cost of 20,000 yuan.

[0087] After comparing the noise reduction effect and implementation cost of each measure, a multi-criteria decision analysis method was used to comprehensively consider three dimensions: noise reduction effect, implementation cost, and feasibility. (Note: These dimensions can be increased or decreased based on actual needs and are not fixed.) Noise reduction effect was extracted from simulation results, for example, a 5dB noise barrier and a 3dB speed limit. Implementation costs were estimated based on budget considerations, including initial investment and long-term maintenance costs. Feasibility was assessed through research to assess potential issues. For example, speed limits may cause traffic congestion (increasing vehicle queues by 10% during peak hours), noise barriers must comply with community planning (occupying green space requires approval), and green belt adjustments may be subject to seasonal restrictions (planting must occur in spring). After determining the dimensions, assign a value and weight each dimension (e.g., 50% for effectiveness, 30% for cost, and 20% for feasibility). Calculate the score for each measure. For example, for a noise barrier, 5 points for effectiveness (out of 5), 2 points for cost (high on a 5-point scale), and 4 points for feasibility (requiring approval). The total score is 5 × 0.5 + 2 × 0.3 + 4 × 0.2 = 3.9. For a speed limit, 3 points for effectiveness, 5 points for cost (low on a 5-point scale), and 3 points for feasibility (congestion risk). The total score is 3 × 0.5 + 5 × 0.3 + 3 × 0.2 = 3.6. For a green belt planting, 3 points for effectiveness, 4 points for cost (moderate on a 5-point scale), and 4 points for feasibility (seasonal restrictions). The total score is 3 × 0.5 + 4 × 0.3 + 4 × 0.2 = 3.5. This score is then organized into a table (also known as the comprehensive evaluation score sheet).

[0088] The solution with the highest score in the comprehensive evaluation score table is selected as the optimal noise reduction solution, and a causal inference algorithm (such as do-calculus) is used to generate a counterfactual prediction, which shows that the noise level increases by 2 dB (65 dB to 67 dB) without intervention and drops to 61 dB after intervention, a decrease of 4 dB.

[0089] Furthermore, by selecting the optimal noise reduction solution (e.g., a soundproofing screen) as the optimal noise reduction solution and using a causal inference algorithm (do-calculus) to generate a counterfactual prediction (noise level increases by 2dB to 67dB without intervention, and decreases to 61dB after intervention, a 4dB reduction), a strong comparative basis for decision-making is provided. The counterfactual prediction demonstrates the consequences of not taking action (noise level increases) in stark contrast to the effect of intervention (a 4dB reduction), demonstrating the necessity of the optimal noise reduction solution. Predictions based on the refined causal relationship diagram not only enhance the persuasiveness of the optimal noise reduction solution but also provide managers with an intuitive analysis of the consequences of intervention or not.

[0090] S6. Combine the dynamic noise distribution map with the optimal noise reduction solution to obtain a comprehensive noise map. Then, based on the comprehensive noise map, evaluate the noise condition and generate a cell score.

[0091] The specific steps include:

[0092] Using GIS (Geographic Information System) tools, the dynamic noise distribution map is overlaid with the optimal noise reduction solution. Simply put, the optimal noise reduction solution is applied to the corresponding area on the dynamic noise distribution map, for example, reducing the noise level from 65 decibels to 61 decibels on a main road, generating a new noise value. The heat map color is adjusted based on the decibel range, for example, 65 decibels is red, 61 decibels is yellow, and 45 decibels is green. Improved areas are marked (for example, the change from red to yellow near a main road) to create a comprehensive noise map.

[0093] After the application is complete, a spatial analysis algorithm (raster analysis) is used to calculate the percentage of residential areas with noise levels exceeding the standard, in accordance with the national standard GB3096-2008 (Acoustic Quality Standard). The specific steps are as follows: A noise threshold is defined, with the daytime standard for residential areas set at 55 decibels (according to the Acoustic Quality Standard). The comprehensive noise map is converted into a spatial data format and divided into regular grids, for example, each grid cell is 5 meters x 5 meters (25 square meters in area). For a residential area of 1,000 square meters, this is divided into 40 grid cells (1,000 ÷ 25 = 40). Each grid cell is assigned a value based on the comprehensive noise map, for example, 61 decibels for grids near main roads and 45 decibels for grids in courtyards. Rasterization converts the continuous noise distribution into discrete spatial units for statistical analysis. All grid cells are traversed to identify those with noise levels exceeding 55 decibels. For example, four grid cells near main roads have a value of 61 decibels, exceeding the standard, while 36 grid cells in courtyards have a value of 45 decibels, meeting the standard. Each grid cell has an area of 25 square meters, and the total number of grid cells exceeding the standard is multiplied by the area of each grid cell. For example, 4 grids exceeding the standard x 25 square meters = 100 square meters. Divide the exceeding area by the total area of the community to calculate the percentage. For example, 100 square meters ÷ 1000 square meters = 10%. This indicates that 10% of the community has noise exceeding the standard, primarily concentrated near main roads.

[0094] However, simply measuring the percentage of areas exceeding the standard cannot fully reflect the impact of noise on residents. Therefore, a weighted scoring method is needed to consider the importance of different areas. For example, excessive noise levels in high-activity areas such as kindergartens and residential buildings warrant greater attention. First, divide the residential area into key areas (such as kindergartens and residential buildings) and general areas (such as road edges). Suppose the kindergarten occupies two grids (50 square meters) and the other areas exceeding the standard along the main road occupy two grids (50 square meters). Then, weights are assigned based on the importance of residential activities: the kindergarten's excessive noise level is given a weight of 2 (highly sensitive), and the other areas are given a weight of 1 (moderately sensitive). Then, a weighted calculation is performed. For example, the excessive noise level of the kindergarten is 2 grids × 25 square meters × a weight of 2 = 100 weighted square meters; the excessive noise level of the other areas along the main road is 2 grids × 25 square meters × a weight of 1 = 50 weighted square meters. The total weighted excessive noise level is 100 + 50 = 150 weighted square meters. The weighted percentage is 150 ÷ (1000 × 1) = 15% (assuming the total area is normalized to a weight of 1). The weighted 15% is higher than the 10% area-based percentage alone, reflecting the severity of the kindergartens' exceeding standards. This results in a 15% impact, highlighting the priority need for governance in key areas.

[0095] Advantageously, a weighted scoring method takes into account the importance of different areas (e.g., kindergartens are weighted 2, other areas are weighted 1), further enhancing the comprehensiveness of the assessment and its sensitivity to the actual impact on residents. For example, a kindergarten exceeding the standard by 50 square meters is weighted to 100 square meters, bringing the total weighted excess area to 150 square meters, representing a 15% share, higher than the 10% share based on area alone. This weighted scoring method prioritizes violations in key areas (e.g., kindergartens and residential buildings) and reflects the true extent of noise impact on residents' lives.

[0096] Now, we'll quantitatively score the community. First, we'll define the scoring criteria. A maximum score of 10 represents an ideal state, meaning that noise levels in all areas of the community are below the GB3096-2008 standard (55 decibels), with no violations. Scores of 1 to 3 indicate a severe violation, 4 to 6 indicate a relatively severe violation, 7 to 9 indicate a minor violation, and scores greater than 9 and less than 10 indicate near-compliance. Point deductions will be applied linearly based on the percentage of area exceeding the standard, with 1 point deducted for every 10% increase in the area exceeding the standard. This is because the area exceeding the standard directly reflects the scope of the noise problem; a higher percentage indicates worse environmental quality. We'll also add an additional deduction to account for the importance of key areas. For example, as mentioned earlier, violations near kindergartens have a greater impact on residents, so we'll include an additional deduction to prioritize these key areas. For example, if the noise level exceeds the standard by 10%, 1 point will be deducted, and if the kindergarten exceeds the standard, 0.5 point will be deducted. The total score = 10 - 1 - 0.5 = 8.5. This 8.5 represents the community's quantitative score, which also corresponds to the minor violation, completing the community evaluation.

[0097] This embodiment further provides a cell evaluation system based on a noise map, including:

[0098] Build modules and construct a digital twin model of the community;

[0099] The prediction module uses a trained multi-agent reinforcement learning model to divide the cell into multiple sub-areas, processes all sub-areas, and outputs a dynamic noise distribution map of the cell;

[0100] The analysis module organizes the dynamic noise distribution map into a causal analysis data set and generates an initial causal relationship diagram through the causal analysis software;

[0101] The improvement module collects and identifies the hidden noise sources in the initial causal relationship graph and adds them to the causal analysis dataset to obtain the improved causal relationship graph;

[0102] The identification module identifies the key noise sources based on the causal strength in the improved causal relationship diagram, designs intervention measures and simulates the effects to obtain the optimal noise reduction solution;

[0103] The scoring module combines the dynamic noise distribution map with the optimal noise reduction solution to obtain a comprehensive noise map. It then evaluates the noise conditions based on the comprehensive noise map and generates a cell score.

[0104] This embodiment also provides a computer device suitable for the case of a cell evaluation method based on a noise map, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the cell evaluation method based on the noise map proposed in the above embodiment.

[0105] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0106] This embodiment further provides a storage medium having a computer program stored thereon. When the program is executed by a processor, the program implements the noise map-based cell evaluation method proposed in the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0107] In summary, the present invention achieves the precise construction of a three-dimensional virtual model of a community by calibrating the digital twin model with a multidimensional dataset and a historical multidimensional dataset. The deviation between simulation and reality is eliminated through calibration, providing a reliable virtual mirror for noise prediction and distribution analysis. The accuracy of the three-dimensional virtual model of the community is greatly improved, and the physical characteristics of the community's acoustic environment can be accurately simulated, laying a solid foundation for subsequent analysis. In addition, by analyzing the difference between the initial causal relationship diagram and the dynamic noise distribution diagram, implicit noise source data is collected, and a complete causal relationship diagram is generated after adding the causal analysis data set, thereby realizing the identification and integration of implicit noise sources, filling the noise increment that cannot be explained by explicit factors, and providing a complete perspective for noise source analysis. It also improves the explanatory power of the causal relationship diagram, so that the description of the cause of noise is changed from partial to comprehensive, avoiding analytical deviations caused by omitting key factors.

[0108] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A cell evaluation method based on noise maps, characterized by: include, Build a digital twin model of the community; Use the trained multi-agent reinforcement learning model to divide the cell into multiple sub-areas, process all sub-areas, and output the dynamic noise distribution map of the cell. The specific steps include the following: Use historical multidimensional datasets to train a multi-agent reinforcement learning model and divide the community into multiple sub-regions; Each sub-region is managed by an agent, which outputs the noise prediction value of the sub-region it is responsible for; Combine the noise prediction values of the agents in each sub-region to form a dynamic noise distribution map; The dynamic noise distribution map is organized into a causal analysis data set, and an initial causal relationship diagram is generated using causal analysis software; Collect and identify the hidden noise sources in the initial causal relationship diagram and add them to the causal analysis data set to obtain a complete causal relationship diagram. The specific steps include the following: When an implicit noise source is identified in the initial causal relationship diagram, the implicit noise source is collected; The collected implicit noise sources are added to the causal analysis data set, and causal inference is performed using pcalg based on the new causal analysis data set to obtain a complete causal relationship diagram; Identify key noise sources based on the causal strength in the improved causal relationship diagram, set intervention measures and simulate the effects to obtain the optimal noise reduction solution; The dynamic noise distribution map is combined with the optimal noise reduction solution to obtain a comprehensive noise map. The noise conditions are then evaluated based on the comprehensive noise map and a cell score is generated.

2. The cell evaluation method based on noise map according to claim 1, characterized in that: Constructing a digital twin model of a community includes the following steps: Collect noise data, traffic flow data and meteorological data of the community to form a multidimensional data set; Establish and adjust the three-dimensional virtual model of the community through multidimensional datasets and historical multidimensional datasets.

3. The cell evaluation method based on noise map according to claim 2, characterized in that: Generating an initial causal relationship diagram by using causal analysis software refers to generating an initial causal relationship diagram using pcalg based on a causal analysis data set.

4. The cell evaluation method based on noise map according to claim 3, characterized in that: According to the causal strength in the improved causal relationship diagram, the key noise sources are identified, intervention measures are set and the effects are simulated to obtain the optimal noise reduction solution, which specifically includes the following steps: Identify key noise sources by analyzing the causal strength in the refined causal relationship diagram and set intervention measures; The intervention measures are simulated, a comprehensive evaluation score sheet is generated, and the optimal noise reduction solution is obtained.

5. The cell evaluation method based on noise map according to claim 4, characterized in that: The step of evaluating the noise situation based on the comprehensive noise map and generating a cell score refers to analyzing the noise decibel value using a spatial analysis algorithm based on the comprehensive noise map to obtain the percentage of areas exceeding the standard; The weighted scoring method was used to analyze the impact of the exceeding area on residents based on regional importance, and the weighted exceeding area ratio was obtained; Based on the proportion of area exceeding the standard and the proportion of weighted area exceeding the standard, linear deduction is used to generate the community score.

6. A noise map-based cell evaluation system, based on the noise map-based cell evaluation method according to any one of claims 1 to 5, characterized in that: include, Build modules and construct a digital twin model of the community; The prediction module uses a trained multi-agent reinforcement learning model to divide the cell into multiple sub-areas, processes all sub-areas, and outputs a dynamic noise distribution map of the cell; The analysis module organizes the dynamic noise distribution map into a causal analysis data set and generates an initial causal relationship diagram through the causal analysis software; The improvement module collects and identifies the hidden noise sources in the initial causal relationship graph and adds them to the causal analysis dataset to obtain the improved causal relationship graph; The identification module identifies the key noise sources based on the causal strength in the improved causal relationship diagram, sets intervention measures and simulates the effects to obtain the optimal noise reduction solution; The scoring module combines the dynamic noise distribution map with the optimal noise reduction solution to obtain a comprehensive noise map. It then evaluates the noise conditions based on the comprehensive noise map and generates a cell score.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the cell evaluation method based on the noise map are implemented in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the cell evaluation method based on the noise map according to any one of claims 1 to 5 are implemented.

Images